Advances and Innovations in Aphasia Treatment Trials

Introduction

Aphasia is a common complication of stroke with a prevalence of nearly one million in the United States alone.¹ Post stroke aphasia (PSA) can be devastating, impacting an individual’s ability to express and/or comprehend language, often disrupting communication, socialization, and work. Intervention is often necessary to improve language, independence, and quality of life. This article summarizes advances in clinical trials of PSA treatment in the past five years. We completed a search of Pubmed to identify clinical trials completed within the past 5 years using the following terms: stroke, aphasia, treatment, speech therapy, speech-language pathology, transcranial direct current stimulation, and transcranial magnetic stimulation. We included the 40 most relevant studies found in our search. Research trends include Noninvasive Brain Stimulation (NIBS), novel speech-language therapy (SLT), pharmacological treatments, and alternative treatments.

Noninvasive Brain Stimulation

NIBS is a promising technique to augment traditional SLT for PSA. Transcranial Direct Current Stimulation (tDCS) and repetitive Transcranial Magnetic Stimulation (rTMS) are two such techniques used in both clinical research and clinical practice. Devido dos Santos et al.² compared a single session each with TMS, tDCS, and sham in randomized order with a naming task and found no statistical significance between before and after stimulation across conditions. Results might reflect the need for greater frequency or intensity of NIBS to improve language. The studies described below delivered NIBS paired with concurrent or subsequent SLT for 2 to 15 sessions. These trials are summarized in Tables 1 and 2.

Table 1.

Summary of TDCS Trials

Authors	Design	Participants	Type of stimulation	Location Targeted	# of Sessions	Results
Campana, Caltagirone & Marangolo3	Crossover trial	20 chronic nonfluent PSA	A-TDCS	Left IFG	20 (10 anodal and 10 sham)	Damage to specific sites associated with lower responses to A-TDCS
Rodriques da Silva et al.4	Double blind randomized, placebo-controlled trial	14 subacute-chronic PSA	C-TDCS	Right Broca’s homolog	5	Improved response time on the Boston Naming Test, C-TDCS > sham
Darkow et al.5	Randomized crossover trial	16 chronic PSA	A-TDCS	Left primary motor cortex	2	Increased activity in language networks with A-TDCS > sham
Devido dos Santos et al.2	Double blind randomized, placebo-controlled trial	13 chronic PSA	A-TDCS and rTMS 1 Hz	TDCS: Broca’s area; rTMS: right hemisphere Broca’s homolog	3; 1 A-TDCS, 1 rTMS; 1 sham	No statistically significant difference between conditions
Fridriksson et al.6,7	Double-blind, randomized sham-controlled trial	74 Chronic PSA	A-TDCS	Area of greatest left hemisphere activation	15	Improved Naming A-TDCS> sham; especially in participants with val/val BDNF polymorphism
Holland et al.8	Pseudorandomized within-subjects crossover trial	10 healthy subjects with left hemisphere language dominance	A-TDCS	Left inferior frontal cortex	2; each session consisted of A-TDCS and sham	A-TDCS had significantly better naming response times and blood oxygen level dependent signal in Broca’s compared to sham
Marangolo et al.9	Double-blind randomized, crossover trial	12 Chronic PSA	C-TDCS	Right cerebellum	20 (10 with TDCS)	Improved verb generation with C-TDCS only
Marangaolo et al. 10	Double-blind, randomized, sham controlled, within-subjects, crossover deign	9 Chronic PSA	A-TDCS and C TDCS	Bilateral stimulation of left IFG and right IFG	30 (15 sham and 15 with anodal and cathodal tDCS)	Improved articulation of treated and untreated stimuli and stronger functional connectivity in left hemisphere in TDCS condition
Meinzer et al. 11	Double blind, randomized, sham-controlled trial	26 chronic PSA	A-TDCS	Left primary motor cortex	8 days (2×1.5 hours/day)	Improved trained items and Communicative Effectiveness Index scores A-TDCS>sham, lasting 6 months
Pestalozzi et al.12	Double blind, sham controlled, within-subjects design	14 Chronic PSA	A-TDCS	Left dorsolateral prefrontal cortex	3 (1 with testing, 1 with sham, 1 with A-TDCS)	Improved verbal fluency A-TDCS> sham
Sebastian et al.13	Double-blind, within-subject, crossover trial	1 Chronic PSA	A-TDCS	Right cerebellum	15	Improved spelling of trained and untrained words A-TDCS > sham
Spielmann, van de Sandt-Koenderman, Heijenbrok, Ribbers14	Randomized, crossover trial	13 Chronic PSA	A-TDCS	Left IFG vs left STG	3 (Sham, TDCS to left IFG, TDCS to left STG)	Unable to determine optimal condition; no improvement on untrained items

Abbreviations PSA, post-stroke Aphasia; A-TDCS, anodal transcranial direct current stimulation; IFG, inferior frontal gyrus; C-TDCS, cathodal transcranial direct current stimulation; rTMS, Repetitive Transcranial Magnetic Stimulation; Hz, Hertz; BDNF, brain-derived neurotrophic factor; STG, superior temporal gyrus

Table 2.

Summary of rTMS Trials

Authors	Design	Participants	Type of stimulation	Location Targeted	# of Sessions	Results
Haghighi, Mazdeh, Ranjbar, & Seifrable15	Double blind, randomized, sham-controlled trial	12 subacute PSA	rTMS 1 Hz	Right inferior posterior frontal gyrus	10	Improved content, fluency, and WAB aphasia quotient rTMS> sham
Hara et al.16	Double-blind, sham controlled, parallel design	8 chronic PSA	rTMS: 1 Hz LF or 10 Hz HF depending on language activation	Right IFG	10	Improvement on Standard Language Test of Aphasia in both groups
Hu et al.17	Double-blind, randomized, sham condition trial	40 subacute-chronic nonfluent PSA	rTMS 10 Hz HF vs 1 Hz LF	Right Broca’s area homolog	10	Improved spontaneous speech, auditory comprehension, and WAB aphasia quotient LF > HF and sham
Khedr et al.18	Randomized, crossover trial	30 subacute nonfluent PSA	rTMS; sequential stimulation of each hemisphere	Right Broca’s area homolog (1Hz) then left Broca’s area (20 Hz)	10	Improved word comprehension, naming, repetition, frequency, and aphasia severity in rTMS>sham
Rubi-Fessen et al.19	Crossover trial	30 subacute PSA	rTMS 1 Hz	Right IFG	10	Improved functional communication TMS> sham
Tsai et al.20	Randomized, sham-controlled trial	56 chronic nonfluent PSA	rTMS 1 Hz	Right pars triangularis	10	Improvements in Concise Chinese Aphasia Test, object naming, and naming reaction time RTMS > sham
Wang et al.21	Double-blind, randomized trial	45 nonfluent PSA	rTMS 1 Hz	Right Broca’s area homolog	10	More improved action and object naming in TMS with synchronous SLT vs TMS with subsequent SLT or sham

Abbreviations PSA, post-stroke Aphasia; rTMS, Repetitive Transcranial Magnetic Stimulation; Hz, Hertz; WAB, Western Aphasia Battery; LF, low frequency; HF, high frequency; IFG, inferior frontal gyrus; SLT, speech-language therapy

Transcranial Direct Current Stimulation

tDCS utilizes surface electrodes to deliver a constant, weak current to the brain that reduces (or increases) the threshold of activation of neurons, influencing cortical excitability. tDCS is a safe, low cost adjunct to traditional SLT to maximize language outcomes in individuals with PSA. Anodal and cathodal tDCS (or both) have been utilized in clinical trials. Anodal tDCS is applied to the perilesional area to excite neuronal activity, while cathodal tDCS is applied to the healthy hemisphere to inhibit cross-hemisphere inhibition, allowing greater activation of the injured hemisphere. Early, small studies of tDCS to augment aphasia therapy in PSA reported mostly positive effects (see reference²² for review).

In all clinical trials we reviewed from the last five years, tDCS was administered for 20 minutes with a current ranging from 1 to 2 mA^2–14. The location of electrode placement, SLT treatment methods, and number of sessions were variable. tDCS resulted in better language outcomes (relative to sham) in each of these trials. Pestalozzi et al.¹² assessed the effects of anodal tDCS to left dorsolateral prefrontal cortex compared to sham, finding immediate improvement in verbal fluency after a single session of tDCS (6.5 words to 5.5 words, p=.010) and improvement in latency of naming high frequency pictures (1264.9 ms to 1913 ms, p=.034), but no improvement in naming accuracy. Marangolo et al.⁹ found cathodal tDCS to the right cerebellum significantly improved verb generation when compared to sham (44% vs 15%, p<.001), with effects persisting at 1 week follow up. Another study placed cathodal tDCS on the right homolog to Broca’s area, finding the tDCS group had quicker response times with naming therapy when compared to the sham group (1.29 to 2.57; p=0.050).⁴ Anodal tDCS placed at the left inferior frontal gyrus (IFG) in conjunction with conversational language therapy resulted in significantly greater improvement compared to sham in picture description (19.5±24.60 vs 10.61±24.50; p=.033), noun naming (18.30±12.87 vs 9.15±11.34; p=0.024), and verb naming (18.40±17.80 vs 7.30±8.86; p=0.019).³ Several studies have found anodal tDCS delivered at M1 paired with naming treatment stimulates language centers of the brain and improves functional language outcomes as compared to sham.^5,11

The largest clinical trial of tDCS to augment naming treatment for PSA targeted the area of greatest activation in the left hemisphere during spoken naming (localized on pre-treatment functional MRI). In this double blind, sham-controlled study of tDCS, 74 individuals with chronic PSA were randomized to anodal tDCS versus sham, matching for aphasia severity and type.⁶ Both groups had identical computer-delivered naming therapy for 15 sessions. TDCS was associated with greater change in number of correctly naming pictured objects, the primary outcome measure: 13.9 words (95% CI, 9.0–18.7) for tDCS vs 8.2 words (95% CI, 3.8–12.6) for sham.⁶ There was a 70% greater improvement in correct naming for anodal tDCS relative to sham. Furthermore, outcome was influenced by an interaction between anodal tDCS and a single-nucleotide polymorphism of the brain-derived neurotrophic factor (BDNF) gene, rs6265^.7 Participants with the normal val/val genotype who received tDCS showed greater response to aphasia treatment than val/val participants who received sham, and greater response to aphasia treatment than the Met allele carriers, regardless of tDCS condition.⁷

As expensive technology such fMRI is not readily available across clinical settings, some investigators have explored behavioral methods for identifying ideal electrode placement. Spielmann et al.¹⁴ compared electrode placement on the left IFG or the left superior temporal gyrus (STG) using anodal tDCS versus sham. On average, placement at the left IFG resulted in better post-treatment performance, though there was large variability in individual responses. It was not possible to establish optimal placement in some participants.¹⁴ Patients with nonfluent aphasia due to frontal cortex damage benefitted from stimulating the frontal cortex, whereas patients with fluent aphasia did not benefit from single session stimulation at either site.¹⁴ There was no improvement in performance on untrained items with single session stimulation; therefore, trained items serve as a better outcome measure to establish optimal placement.

Imaging studies have explored the neural mechanisms of tDCS effects on language in PSA. One voxel-based lesion symptom mapping study showed that individuals with damage to the left basal ganglia, insula, and superior and inferior longitudinal fasciculi had lower response to tDCS.³ The authors concluded that integrity of left subcortical structures and white matter language pathways influences the benefits of tDCS.³ Darkow et al.⁵ investigated in the effects of tDCS on functional brain activity to determine effects of tDCS in individuals with PSA compared to healthy controls. Participants were hooked up to an intra-scanner tDCS device with anode placed at M1 and underwent an fMRI while naming pictures they could consistently name, as established in baseline testing.⁵ Relative to sham, tDCS resulted in enhanced activity in language regions and reduced activity in domain-general brain regions associated with working memory and response selection including anterior cingulate cortex, left insula, and right lingual gyrus.⁵ The investigators indicated tDCS may improve efficiency at the neural level, by increasing activity in the language areas such that language success can be achieved with lesser demands on cognitive processes of working memory and response selection.⁵ Interestingly, activity was enhanced in only the language networks and not the motor or visual networks, even though stimulation was placed on M1. These and other results indicate that placement of the electrodes may not matter as much as activating the language network with language treatment during the tDCS. ^{5, 8} One study of aphasia after bilateral middle cerebral artery stroke showed that right cerebellar tDCS resulted in both (1) greater improvements in spelling, compared to sham paired with the same therapy (39/40 vs. 21/40; p<0.0001 for trained words and 33/40 vs. 11/40; p<0.0001 for untrained words) and (2) enhanced connectivity between right cerebellum and frontal and temporal cortex, evaluated with resting state fMRI.¹³ Other studies have also reported increased resting state connectivity between specific brain regions resulting from tDCS paired with language therapy.¹⁰

Transcranial Magnetic Stimulation (TMS)

rTMS is another form of NIBS that has been used to supplement SLT in PSA. rTMS can be delivered at both high frequency (excitatory) and low frequency (inhibitory). Low frequency rTMS is often applied to the contralesional right hemisphere to inhibit right hemisphere activation during language related tasks and to encourage perilesional left hemisphere activation¹⁶ in both the subacute and chronic PSA.

Subacute.

Both low frequency and high frequency rTMS have been beneficial in improving language outcomes when paired with SLT in subacute PSA. Khedr et al.¹⁸ applied bihemispheric rTMS using high frequency to the injured left IFG and low frequency to the right sided, homologous IFG for a total of 10 sessions followed by 30 minutes of SLT, compared to sham. Patients who received rTMS showed significantly greater improvements compared to sham in accuracy of word comprehension (p=.04), naming (p=.01), repetition (p=.002), and in aphasia severity (1.8±1.2 vs 0.9±0.3; p=.018).¹⁸ This significant improvement was present immediately after treatment and at 2 month follow up. Haghighi et al.¹⁵ found that low frequency rTMS applied to the right IFG at 1Hz for 30 minutes paired with SLT for 45 minutes, 5 times per week over 2 weeks, compared to sham with the same SLT, resulted in improved scores on the Farsi WAB (aphasia quotient =50.27 ± 28.37 vs. 39.50 ±18.14). Large effect sizes were observed for speech content and fluency scores and aphasia quotient.¹⁵ Rubi-Fessen et al.¹⁹ applied rTMS at 1 Hz to the right IFG compared to sham, each with subsequent 45 minute SLT. They found that rTMS led to greater improvement compared to sham in all language measures, including functional communication (34.20±12.09 vs. 32.93±14.84 on the Amsterdam-Nijimegen Everyday Language Test A-scale; p=.050).¹⁹

Chronic.

Both low and high frequency rTMS are effective in improving language outcomes in chronic PSA as well. Hara et al.¹⁶ used functional near infrared spectroscopy to determine the hemisphere of language activation to determine rTMS delivery method. They implemented 1 Hz (inhibitory) rTMS to contralesional right IFG in those with left sided language activation and implemented 10 Hz (excitatory) rTMS to right IFG in those with right hemisphere activation during language tasks.¹⁶ Both groups received intensive SLT following rTMS. The groups showed equally significant improvements in language functions,¹⁶ indicating that right hemisphere activation during language may not always be maladaptive. Authors suggest that lesion site and size may determine whether perilesional areas can take on language functions or the right hemisphere is needed to compensate.¹⁶ Likewise, Tsai et al.²⁰ found that applying low frequency rTMS in the area contralateral to the lesion followed by 1 hour of SLT also resulted in greater improvements compared to sham in object naming (47.4±28.3 vs. 35.3±30.1; p<0.05), object naming reaction time (12.1±4.9 vs13.9±5.1; p <0.01), action naming (34.8±24.6 vs. 25.9±20.4; p<0.01), and action naming reaction time (15.4±5.2 vs. 15.4±5.7; p<0.01) immediately after treatment, and lasting up to 3 months.²⁰ Hu et al.¹⁷ also found that both high and low frequency rTMS to the contralesional hemisphere resulted in significant mean improvement compared to sham and controls, although low frequency rTMS yielded greater improvements, perhaps because fewer participants had right hemisphere dominance for language after stroke.

The above studies delivered SLT immediately following delivery of rTMS. Wang et al²¹ compared SLT delivery immediately following rTMS to synchronous delivery of SLT with rTMS. They found that 1 Hz of rTMS to right IFG with synchronous SLT was more effective than either sham (SLT alone) or rTMS with subsequent SLT. Improvements in verbal expression included description, object naming, and action naming, and lasted up to 3 months.²¹ Although synchronous SLT may be more beneficial, rTMS produces a loud clicking noise during administration, which can prove distracting to the patient during SLT.

Advances in Speech-Language Therapy

Recent research in SLT has focused largely on interventions to improve expressive language, specifically in chronic PSA.^23–31 Constraint Induced Aphasia Therapy (CIAT) and Intensive Language Action Treatment (ILAT) are group therapy methods. CIAT requires patients to communicate verbally with each other while playing a card game, while prohibiting use of any nonverbal communication methods.²⁷ In ILAT patients interact by requesting picture cards from each other and naming pictures using carrier phrases.²⁴

In one-on-one treatment, a number of therapies utilize multiple modalities to stimulate language recovery. Power-Afa²³ is an Italian software program consisting of phonological, semantic, orthographic, morphological, and syntactic tasks of varying levels of complexity that are adjusted over time. Phonomotor treatment is an intensive protocol that trains individual sounds and progresses to training of 1–3 syllable words and nonwords using articulatory-motor, acoustic, tactile kinesthetic, and orthographic modalities.²⁵ In Verb Network Strengthening Treatment (VNEST), the therapist provides a verb and asks the patient to generate related agents and receivers of the action; additionally, participants answer wh- questions about each generated schema.²⁶ Other treatments focus specifically on verbal expression. Melodic Intonation Therapy (MIT) focuses only on verbal expression with an established protocol involving the therapist singing short utterances and tapping to the rhythm alongside the patient.^28,29 As the patient progresses, utterances become increasingly complex and the therapist cueing decreases.^28,29 Harnish et al.³¹ utilized an intensive naming treatment presenting 50 pictures 8 times across a session for a total of 400 repetitions a session. A phonological treatment, Earobics, uses software to target sound to picture matching, letter-to-sound mapping, auditory segmentation, rhyme detection, word to picture matching, and auditory discrimination.³⁰ The study design, participant number, and results of each study are summarized in Table 3.

Table 3.

Summary of Speech-Language Therapy Trials

Authors	Design	Participants	Treatment	Dosage	Results
Barbancho et al.32	Double-blind, randomized placebo controlled trial	27 Chronic PSA	CIAT with Memantine Or CIAT with placebo	3 hours/day for 2 weeks Total: 30 hours	Western Aphasia Battery scores improved with memantine; Improved even more with memantine + CIAT
Breitenstein et al.34	Multicenter, open-label, blinded endpoint randomized controlled trial	156 Chronic PSA	Evidenced based SLT vs deferral of same SLT for 3 weeks	10 hours a week of individual and group therapy (30+ hours)	Significant improvement lasting at least 6 months
Breitenstein et al.33	Double-blind, randomized placebo controlled trial	10 Chronic PSA	Levadopa with SLT; placebo with SLT	4 hours/day for 10 days Total: 40 hours	Improved naming and verbal communication in both groups
De Luca et al.23	Randomized, controlled trial	32 Chronic PSA	Power-Afa computer based intervention vs Traditional therapy	45 minute/day 3 days/week for 8 weeks; (18 hours)	Improved repetition, selective attention, denomination, and reading Power-Afa> traditional
Dignam et al.35	Nonrandomized, parallel-group, pre-post test design	34 Chronic PSA	Intensive therapy vs. distributed therapy	48 hours in 3 weeks or 48 hours in 8 weeks	Distributed delivery had greater impact on Boston Naming Test scores
Edmonds, Mammino, and Ojeda26	Multiple baseline design	11 Chronic PSA	VNeST	35 hours	Improved trained and untrained sentence probes, object and action naming.
Harnish et al.31	Feasibility study	8 Chronic PSA	Computer based naming treatment	1 hour/day; 4 days/week for 2 weeks	Improvements after 1 treatment session maintained at 2 months
Kendall et al.25	Non-randomized trial	26 Chronic PSA	Phonomotor Therapy va. delayed Phonomotor therapy	2 one hour sessions/day 5 days/week for 6 weeks Total: 60 hours	Improved naming of untrained items lasting up to 3 months post
Stahl et al.24	Randomized, crossover, controlled trial	18 Chronic PSA	ILAT va. Naming Therapy	3.5 hour/day 6 days (21 hours)	Greater language improvements in ILAT > Naming Therapy
van der Meulen et al.28	Multicenter, randomized, controlled trial	27 subacute PSA	MIT vs control intervention (comprehension oral and written	5 hours/week for 6 weeks Total: 30 hours	Significant improvement with MIT
Wan et al.29	Pre-post design	11 Chronic PSA	MIT vs untreated	1.5 hours/day 5 days a week for 15 weeks Total: 110 hours	Improvements in speech lead to structural changes in the right hemisphere
Woldag et al.27	Single blind, randomized, controlled trial with 3 arms	60 Acute PSA	CIAT group therapy 3 hrs/day for 10 days (30 hrs) vs. conventional group 2 hours/day for 10 days vs. individual therapy 2x/day and group therapy (14 hours)	3 hours/day for 10 days (30 hours) or 2 hours/day for 10 days (20 hours) or 2x/day individual and group for a (14 hours).	All groups showed significant improvement.
Woodhead et al.30	Double-blind, placebo controlled, crossover, within-subjects design	20 Chronic PSA	Earobics Vs. Earobics with Donepezil	10 hours a week for 5 weeks (50 hours)	Donepezil had a negative effect
Woodhead et al.36	Baseline-controlled, repeated measures, crossover design	21 Chronic PSA	EG1: iReadMore with anodal TDCS EG2: iReadMore with Sham	2 4 week blocks; 34 hours of training; 11 stimulation sessions	Improved reading of trained words with iReadMore, small facilitation with anodal stimulation

Abbreviations: PSA, post stroke aphasia; CIAT, Constraint Induced Aphasia Therapy; SLT, speech-language therapy; VNeST, Verb Network Strengthening Treatment; ILAT, Intensive Language Action Treatment; MIT, Melodic Intonation Therapy

Studies that utilize a mixed treatment approach, delivering treatment in both individual and group settings, have also been effective in improving language outcomes (Table 3).^{27, 33,35} Some studies have also incorporated computer-delivered or tablet-based therapies, such as “iReadMore”, an app designed to improve word recognition in people with reading deficits in PSA. Use of iReadMore resulted in significant improvement in reading trained words (8.7%, 95%CI: 6–11.4) but not untrained words.³⁶ However, when combined with TDCS, there was a significant improvement in untrained as well as trained words.³⁶

Although some studies have identified new treatment methods, much of the recent research explores the ideal intensity of therapy. Breitenstein et al. ³³ found that intensive SLT administered for greater than or equal to 10 hours a week with a therapist and 5 hours or more per week of self-practice, for 3 weeks including both individual and group settings, resulted in significantly greater improvements in verbal communication when compared to control group who deferred therapy for 3 weeks. Dignam et al.³⁵ compared an intensive (16 hours a week for 3 weeks) vs. distributed therapy (6 hours a week for 8 weeks) delivery, finding that distributed therapy resulted in greater change in scores on the Boston Naming Test (BNT). Similarly, Woldag et al.²⁷ found no significant difference in language outcomes with CIAT delivered 3 hours a day for 10 days (30 hours of CIAT), conventional group therapy delivered 3 hours a day for 10 days (30 hours of group therapy), or a combination of individual and group therapy delivered twice a day for a total of 14 hours over 2 weeks. Therefore, the mixed therapy seemed to be the most time-effective or efficient for facilitating language recovery.

Pharmacological and Medical Interventions

Other clinical trials have evaluated the efficacy of medications in improving PSA. However, like previous medication trials for PSA (see³⁷ for review), the recent trials have been small (N=10–156) and some have had weak study designs (e.g. open-labelled) (Table 3). Woodhead et al.³⁰ found that donepezil was not beneficial in improving PSA and actually had a negative effect on speech comprehension outcomes. Although some studies have indicated that levadopa may augment moderate intensity language therapy, Breitenstein et al.,³³ found it did not improve outcomes of high intensity language therapy (63.8% for levadopa vs 66.5% for placebo). On the other hand, memantine had a small positive impact on language functioning, resulting in greater improvement in Western Aphasia Battery (WAB) Aphasia Quotient, compared to placebo (67.1 ± 5.5 vs 65.8 ± 3.0; p <0.002)³². The effect was smaller than effects reported previously for memantine plus CIAT.³²

Other

Raglio et al.³⁸ reported that 10 patients randomized to music therapy in addition to 30 sessions of SLT showed significant improvement compared to 10 patients who received 30 sessions of SLT alone (Aachener Aphasia spontaneous speech subtest: p = 0.020; Cohen’s d = 0.35).

In a study of 60 patients with acute to chronic PSA randomized to “Heart-Gallbladder” acupuncture versus traditional acupuncture, both groups improved on the Aphasia Battery for Chinese, but the experimental group showed significantly greater improvement in fluency, repetition, naming, and reading scores (p<0.05).³⁹ However, results were not corrected for multiple comparisons.

Another modality currently being explored is a more invasive form of brain stimulation, epidural cortical stimulation (CS). Epidural CS requires surgical implantation of a device, which can then be turned on and off. Cherney⁴⁰ evaluated epidural CS paired with language therapy compared to language therapy alone. Language therapy included 3 hours a day, 5 days a week for 6 weeks including apraxia drills, confrontational naming, computer practice, and conversational practice. Overall, the CS group showed greater improvements than controls on the WAB aphasia quotient: 7.98 ±4.94 (95% CI, −0.83–2) vs 4.59±5.17 (95% CI −1.27–1.73).⁴⁰

Discussion and Limitations

It should be noted that methodological weaknesses in many of the aphasia treatment studies compromise strong conclusions about efficacy. Most trials have been small (see Tables 1,2, and 3); only 3 of the recent trials have included more than 50 participants^6,7,20,34, and only one has included more than 100 participants. ³⁴ Many have not reported that evaluators of outcome have been masked to the treatment group. Although most of the tDCS studies have reported improvement in untrained as well as trained items¹³ or on untrained standardized aphasia batteries^11,15–18, some have shown no improvement on untrained items¹⁴ or have not reported on generalization.⁶ Likewise, many behavioral³⁵ and medication³² trials have reported gains on standardized tests, and others report gains on untrained stimuli;²⁶ but many others failed to report generalization.

Most studies have been carried out in participants with chronic post-stroke aphasia (>6 months, often many years, post stroke), in which it is assumed that language performance is relatively stable with no intervention. Studies in subacute²⁸ and acute²⁷ stroke have mostly used randomized designs with evaluators masked to treatment group to try to control for variability in spontaneous recovery that takes place during the early months after stroke. However, the effect size of treatment must be very large (or the groups very large) to show an effect of treatment over and above the spontaneous improvement, as illustrated by the study by Woldag and colleagues,²⁷ which showed no significant effect of treatment with CIAT group therapy 3 hours/day for 10 days (30 hrs) compared to conventional group therapy 2 hours/day for 10 days (20 hours) or individual therapy twice a day and group therapy (14 hours), with 20 participants in each group. One study of tDCS¹⁹ used a crossover design and randomized order of treatment with 30 individuals with subacute PSA to partially control for variability of spontaneous recovery (as each participant is compared to themselves, across conditions). This study showed significant improvement in functional communication with tDCS compared to sham, along with language therapy.

Finally, most studies have included participants with a variety of aphasia subtypes, or have included individuals with various nonfluent aphasia subtypes (Global, Transcortical Motor, and Broca’s aphasia). The distribution of aphasia subtypes might influence efficacy. That is, it is possible that individuals with Broca’s aphasia respond more to certain types of treatment, while those with Wernicke’s aphasia respond more to other types. However, none of the studies have been adequately powered to identify differential efficacy across subtypes. Therefore, it is not possible to predict for whom the therapy will be effective, even for studies that report statistically significant results for their population.

One caveat about tDCS is that the current is much more disperse than in TMS, making it difficult to identify the optimal stimulation site. On the other hand, there is evidence from fMRI that stimulation over any area of the network being activated by the concurrent language task will facilitate activation throughout the network.⁸

Conclusions and Future Directions

This review of clinical trials for PSA in the past five years reveals that a multitude of interventions can be beneficial in improving language and functional outcomes for patients with PSA, with the majority of research focusing on the chronic phase of aphasia. The most effective or efficient interventions combine SLT with NIBS or medications. It is hypothesized that both NIBS and certain medications that influence neurotransmitters increase long term potentiation or depression required for neuroplasticity. Thus, these interventions can augment SLT in recruiting non-damaged areas of the left or right hemisphere to assume the functions of the damaged parts.

In regards to NIBS, both TMS and TDCS are generally effective interventions when paired with SLT.^{2,4, 9,12,22} Additionally epidural CS is another form of brain stimulation that may augment SLT in PSA.⁴⁰ However, further studies are needed to identify the most effective electrode placement, the optimal “dose”, and the mechanisms by which NIBS and CS facilitate improvement.

A number of SLT interventions are beneficial when administered at a moderate to high intensity in both subacute and chronic aphasia.^23–29

With regard to pharmacological interventions, preliminary studies indicate that donepezil negative effects on speech comprehension, while memantine may have a positive impact on language, but additional studies are needed to confirm these results.^30,32 Levadopa has had inconsistent effects on language recovery.³³ Music therapy and acupuncture may be effective adjuncts to SLT, although further research is needed to confirm preliminary findings.^38,39

Future studies should focus on identifying the most cost-effective combination of interventions. Combining medications with NIBS and SLT, for example, might result in improvement with fewer number of sessions. Randomized trials are also needed to evaluate the effect of computer-delivered or app-delivered interventions, with or without clinician-delivered therapy in both the clinic setting and remotely (telerehabilitation). Furthermore, more trials are needed in the early stage of recovery. These studies are more difficult, because they must show an effect of the intervention over and above spontaneous recovery. Thus, only interventions with large effect sizes will yield significant results.