Donepezil Does Not Enhance Perceptual Learning in Adults with Amblyopia: A Pilot Study

Abstract

Amblyopia is a developmental disorder that results in a wide range of visual deficits. One proven approach to recovering vision in adults with amblyopia is perceptual learning (PL). Recent evidence suggests that neuromodulators can enhance adult plasticity. In this pilot study, we asked whether donepezil, a cholinesterase inhibitor, enhances visual PL in adults with amblyopia. Nine adults with amblyopia were first trained on a low-contrast single-letter identification task while taking a daily dose (5 mg) of donepezil throughout training. Following 10,000 trials of training, participants showed improved contrast sensitivity for identifying single letters. However, the magnitude of improvement was no greater than, and the rate of improvement was slower than, that obtained in a previous study in which six adults with amblyopia were trained using an identical task and protocol but without donepezil (Chung et al., 2012). In addition, we measured transfer of learning effects to other tasks and found that for donepezil, the post-pre performance ratios in both a size-limited (acuity) and a spacing-limited (crowding) task were not significantly different from those found in the previous study without donepezil administration. After an interval of several weeks, six participants returned for a second course of training on identifying flanked (crowded) letters, again with concurrent donepezil administration. Although this task has previously been shown to be highly amenable to PL in adults with amblyopia (Chung et al., 2012; Hussain et al., 2012), only one observer in our study showed significant learning over 10,000 trials of training. Auxiliary experiments showed that the lack of a learning effect on this task during donepezil administration was not due to either the order of training of the two tasks or the use of a sequential training paradigm. Our results reveal that cholinergic enhancement with donepezil during training does not improve or speed up PL of single-letter identification in adults with amblyopia, and importantly, it may even halt learning and transfer related to a crowding task. Clinical Trial Registration: This study was registered with ClinicalTrials.gov (NCT03109314).

Keywords: amblyopia; cholinesterase inhibitors; contrast sensitivity; donepezil (aricept); perceptual learning; uncrowding.

Figures

Figure 1

Schematic summary of the training protocol and tasks for the different observers and for the two phases of training. Day numbers within gray squares represent those days in which observers ingested 5 mg donepezil (except for observer S11 who completed the two phases of training without taking any donepezil).

Figure 2

Training results from three observers, for single letter identification task (left) and flanked letter identification task (right). Data for all other observers are presented in Figure S1. Observers S1–S9 first trained on the single letter task (Phase 1). Note that for observer S10 (bottom panels), the order of training was reversed: flanked letter identification task first (Phase 1), followed by single letter identification task (Phase 2). Due to an unfortunate incident, S10 was only trained for 9, instead of 10 sessions, in Phase 2. Error bars represent ±1 SEM.

Figure 3

Boxplots showing the distribution of (left) the exponential decay constant, τ, derived from the best-fit exponential function for each observers' training data for the single letter identification training; and (right) the slope of the linear fit based on the best-fit line for each observer's training data for the flanked letter training, for observers with donepezil administration (present study, plotted in red) and without donepezil (Chung et al., , plotted in gray). For observers with donepezil administration, circles represent observers S1–S9, triangles represent data for S10, and stars represent data for S11. The stars are plotted adjacent to the boxplot for the present study because S11 did not take donepezil during training, and thus her values were not included in the boxplot.

Figure 4

Training results averaged across all observers in the current study (red circles). Single letter identification task (left). Flanked letter identification task (right). For comparison, the gray circles show the averaged data from our previous study of PL without donepezil in amblyopia (Chung et al., 2012). Note that in Chung et al. (2012), instead of sequential training, the two training tasks were performed by two separate groups of observers (single letter training: n = 6; flanked letter training: n = 5). Shaded regions represent 95% confidence intervals based on the model parameter values. Error bars shown for the 1st and the 100th block represent ±1 SEM of the group-average values.

Figure 5

Training and transfer effects for single letter identification training (left) and flanked letter identification training (right) for observers S1–S10. For each of these training tasks, changes in performance for the training task are plotted in the left sub-panel and changes in performance for the three baseline tasks are plotted in the right sub-panel. Except for the values plotted for the training task for flanked letter identification, values plotted represent the Post/Pre threshold ratio (PPR) of the last and the first block of the training task (based on the fitted curves), or one of the three baseline tasks (size, spacing, and contrast threshold). For the training effect of flanked letter identification, Pre/Post ratios are used, because performance on this task was proportion correct rather than a threshold. A ratio of 1 (indicated by the dashed lines) represents no change, and ratios <1 indicate improved performance. Black solid circles and error bars are the mean and 1.96 × SEM from the present study. Gray squares and error bars are the mean and 1.96 × SEM, replotted from our previous study of training without donepezil (Chung et al., 2012). Open symbols represent individual observer data (red for amblyopes with strabismus and green for the amblyope without strabismus, circles for observers S1–S9, triangles for observer S10). For flanked letter identification, we were not able to measure the spacing limit for two observers during pretests, thus the number of open symbols is fewer than seven.

Figure 6

Post/Pre Ratios for each of the three baseline tasks are plotted as a function of initial performance for single letter identification training (top) and flanked letter identification training (bottom). Each data point (red circle) represents the performance for one observer. Dashed lines represent linear regression lines for the data, with 95% confidence bands shown as solid red lines. The slope of each line (and 95% confidence interval) is displayed in the lower left corner in each plot. For the spacing measurement, the number of data points plotted is fewer than the number of observers for both training tasks because we were not able to obtain a reliable pre-test performance measure for some observers (the pre-test performance for all spacings except for the unflanked condition was poor, and we were therefore unable to fit a psychometric function to obtain a spacing threshold).