Randomized trial of iReadMore word reading training and brain stimulation in central alexia

Abstract

Central alexia is an acquired reading disorder co-occurring with a generalized language deficit (aphasia). We tested the impact of a novel training app, 'iReadMore', and anodal transcranial direct current stimulation of the left inferior frontal gyrus, on word reading ability in central alexia. The trial was registered at www.clinicaltrials.gov (NCT02062619). Twenty-one chronic stroke patients with central alexia participated. A baseline-controlled, repeated-measures, crossover design was used. Participants completed two 4-week blocks of iReadMore training, one with anodal stimulation and one with sham stimulation (order counterbalanced between participants). Each block comprised 34 h of iReadMore training and 11 stimulation sessions. Outcome measures were assessed before, between and after the two blocks. The primary outcome measures were reading ability for trained and untrained words. Secondary outcome measures included semantic word matching, sentence reading, text reading and a self-report measure. iReadMore training resulted in an 8.7% improvement in reading accuracy for trained words (95% confidence interval 6.0 to 11.4; Cohen's d = 1.38) but did not generalize to untrained words. Reaction times also improved. Reading accuracy gains were still significant (but reduced) 3 months after training cessation. Anodal transcranial direct current stimulation (compared to sham), delivered concurrently with iReadMore, resulted in a 2.6% (95% confidence interval -0.1 to 5.3; d = 0.41) facilitation for reading accuracy, both for trained and untrained words. iReadMore also improved performance on the semantic word-matching test. There was a non-significant trend towards improved self-reported reading ability. However, no significant changes were seen at the sentence or text reading level. In summary, iReadMore training in post-stroke central alexia improved reading ability for trained words, with good maintenance of the therapy effect. Anodal stimulation resulted in a small facilitation (d = 0.41) of learning and also generalized to untrained items.10.1093/brain/awy138_video1awy138media15796149281001.

Figures

Figure 1

Study design. G1 = Group1: received tDCS in Block 1 and sham in Block 2. G2 = Group2: received sham in Block 1 and tDCS in Block 2.

Figure 2

Consolidated Standards of Reporting Trials (CONSORT) flow diagram..

Figure 3

Patient structural MRI images and lesion overlap map. Crosshairs indicate the approximate location of the stimulation site. Bottom right tiles show the lesion overlay map with voxels where at least two patients had damage. The highest lesion overlap (n = 20) was seen in two areas: (i) the superior longitudinal fasciculus underlying the supramarginal gyrus; and (ii) the junction of the superior longitudinal, inferior longitudinal and inferior fronto-occipital fasciculi underlying the posterior superior temporal sulcus.

Figure 4

Therapy effects on word reading ability. Change over time in (A) mean word reading accuracy (n = 21) and (B) reaction times (n = 20). There were four different word lists: words trained in Block 1 (blue), words trained in Block 2 (red), untrained words (black) and the unmatched list of high-frequency, low-imageability core words (purple). Error bars indicate within-subject standard error of the mean (SEM). Training Block 1 was administered between T3 and T4; Block 2 was administered between T4 and T5.

Figure 5

Change in word reading ability after therapy. Effects of iReadMore and tDCS on change in (A) word reading accuracy (n = 21) and (B) word reading reaction times (RT) (n = 20). Block 1 change was calculated as accuracy or reaction time at T4 − T3; Block 2 change was T5 − T4. G1 = cross-over group 1; G2 = cross-over group 2. Error bars represent the within-subject SEM.

Figure 6

Change in word reading accuracy and self-reported reading by participant. (A) Raw percentage change in word reading accuracy for trained (black) and untrained (grey) words, averaged over Block 1 (T4 − T3) and Block 2 (T5 − T4). For trained words, this represents the average of the change in the 90 words trained in Block 1 between T3 and T4 and the change in the 90 (different) words trained in Block 2 between T4 and T5. For untrained words, this represents the change in the 90 untrained words over the same two time-periods. Participants are ordered according to tDCS group, followed by ascending CAT naming accuracy. (B) The CDP measures self-report ability in silent word, sentence, text and mail reading. Score for each level is out of 4, giving a total score out of 16. Change in CDP score is the difference between T5 (after training) minus T3 (before training). Positive scores represent improvements in self-reported reading ability. CDP data were unavailable for Patient P4.