May cure Prozac amblyopia in adults

Fluoxetine does not improve the effects of perceptual learning on visual function in adults with amblyopia


  • Extrastriate cortex
  • Translational research


Amblyopia is a common visual disorder that can be treated in childhood. However, the effectiveness of therapies in adult patients with amblyopia is limited. Fluoxetine can restore neural plasticity in the early stages and has been used to restore functional vision in adult rats with amblyopia. We conducted a phase 2, double-blind, randomized study (fluoxetine vs. placebo) to determine whether fluoxetine could improve visual acuity in amblyopic adults. This interventional study included 42 participants diagnosed with moderate to severe amblyopia. The subjects were randomized to receive either 20 mg fluoxetine (n = 22) or placebo (n = 20). During the 10-week treatment period, all subjects performed computer-aided perception training and eye patches every day. At the primary endpoint, the mean difference between treatment groups in improving visual acuity was only 0.027 logMAR units (95% CI: -0.057 to 0.110; p = 0.524). However, visual acuity was improved with fluoxetine (-0.167 logMAR; 95% CI: -0.226 to -0.108; p <0.001) and with placebo (-0.1194 logMAR; 95% CI: –0, 254 to –0, 133) significantly improved from the ground up for 10 weeks p <0.001) groups. While this study did not provide evidence that fluoxetine increased neuroplasticity, our data support other recent clinical studies that suggest that vision improvement may be achieved in adults with amblyopia.


Amblyopia is a condition in which best-corrected visual acuity (BCVA) is impaired in one eye or, less often, both eyes, although there are generally no ocular abnormalities. Amblyopia occurs when one or both eyes have abnormal (physical or physiological) vision during the sensitive phase of childhood (from birth to age 6) 1, 2 . Unilateral amblyopia is most commonly caused by strabismus, anisometropia, or both. Amblyopia is often defined in the clinical setting as a logarithm of the minimum angle of resolution (logMAR) of the BCVA of less than 0.30 (Snellen equivalent: equal to or less than 20/40) and / or an interocular BCVA difference of 0.2 logMAR more ( in unilateral amblyopia). The prevalence of amblyopia in the general population varies between 1.3% and 3.6% and is one of the most common causes of monocular visual impairment in adults 3, 4, 5 . Patients with monocular amblyopia have a significantly increased risk of visual impairment when the vision of their “good” eye is lost as a result of trauma or illness 6 .

Amblyopia can be treated at a young age 7, 8, 9, but the Eyesight of school-age children decreases as neuroplasticity of the visual system decreases and, possibly, therapy compliance 1, 10, 11, 12 . In the past, amblyopia was thought to be difficult, if not impossible, to treat in adults 13 .

An improved understanding of the neural mechanisms underlying amblyopia and neuroplasticity of the brain in adults has led to the development of visual rehabilitation methods that follow the critical period 17, 18, 19, 20, 21 can be applied . Experimental models of amblyopia are based on the effects of monocular withdrawal on the structure and function of the visual cortex 22, 23 . Espinosa and Stryker demonstrated on an animal model 23that the effects of amblyopia can be reversed in the critical phase of early postnatal development, but not later in life. However, recent evidence suggests that there is an enrichment of the environment 24, 25 and pharmacological treatment 26 reactivate critical periodic plasticity in the adult rodent visual cortex. In particular, fluoxetine, a selective serotonin reuptake inhibitor (SSRI), promotes neuroplasticity and neurogenesis 27 and reactivates the critical periodic plasticity in the rat visual cortex 15 .

An emerging literature suggests that using conventional therapies (e.g., occlusion therapy) in adolescents (i.e., older children and adolescents) 28, 29, 30 and adults 17, 18, 31, 32 with amblyopia even after Engaging a Vision Improvement Can Be Achieved Amblyopia in childhood has been considered untreatable in the past. In addition, catecholamine-based medical treatments can reduce eyesight amblyopic Temporarily improve patients, including adults ( 1, 33, 34) . The awareness training 35, the use of dichoptic non-action and action video games 36 and the use of video games while mending 37 can improve vision in the amblyopic eye and binocular vision in adults. This placebo-controlled study investigated whether fluoxetine could improve neuroplasticity and vision in adults with amblyopia. The treatment comprised eye patches and computer-aided perception training on a web-based system for all subjects.


Subjects of study

A total of 42 subjects were included in the study, with 22 and 20 subjects being randomly assigned to the fluoxetine and control group, respectively. Table 1 provides a complete list of the eligibility and exclusion criteria used in patient selection. 41 of 42 subjects (97, 6%) needed new glasses before randomization. Four subjects were not compliant, three subjects withdrew their consent, and one subject was lost for follow-up. A total of 37 subjects therefore completed the 10-week treatment period including the primary endpoint investigations and 34 completed the 3-month follow-up treatment period (20 in the fluoxetine group and 14 in the control group; 1). The data of all 42 randomized subjects were subjected to an intention-to-treat analysis and included in the analyzes. Subjects who completed the study showed good drug compliance (> 85%) and completed> 85% of the computer-aided awareness training sessions.

Full size table

Study design and process. ( A. ) Disposition and subjects. Subjects with moderate (0.3–0.6 logMAR interocular visual acuity difference) to severe (> 0.6 logMAR difference) amblyopia were included in the study. ( B. ) Visiting and assessment plan as well as duration of medication and active training.

Full size image

The demographic and ocular baseline parameters are summarized in Table 2. In short, the best-corrected logMAR visual acuity at the exit was between 0.30 and 1.08 for the amblyopic eye and no worse than 0.10 for the dominant eye. The majority of the subjects had anisometropic amblyopia [19 of 20 control subjects (95.0%), 18 of 22 fluoxetine subjects (81.8%)]. Four subjects [1 (5, 0%) in the control group and 3 (13, 6%) in the fluoxetine group] showed combined strabismic-anisometropic amblyopia, while 1 (4.5%) in the fluoxetine group showed strabismic amblyopia Showed amblyopia. These 5 subjects underwent strabismus surgery in childhood. Twenty subjects [8 (40, 0%) in the control group and 12 (54, 5%) in the fluoxetine group] had moderate amblyopia (0, 30–0, 60 logMAR) and 22 [12 (60, 0%) ) in the control group and 10 (45.5%) in the fluoxetine group] had severe amblyopia (> 0.60 logMAR) (Fig. 1A).

Full size table

The mean age of the test subjects was 36, 4 ± 11, 5 years in the control group and 38, 5 ± 12, 5 years in the fluoxetine group. The mean logMAR visual acuity in the amblyopic eye was 0.620 ± 0.190 (Snellen equivalent: 20/83) in the control group and 0.649 ± 0.252 (20/89) in the fluoxetine group. The mean interocular visual difference between the amblyopic and the fellow eye was 0.728 in both groups. For hyperopic subjects (n = 33) the mean refractive error in the amblyopic eye was +3, 13 ± 1, 49 and +3, 07 ± 1, 79 D in the control or fluoxetine group and for myopic subjects (n = 9) the mean refractive error. The error in the amblyopic eye was -1, 23 ± 0.66 and -2, 81 ± 1.07 D. The Mean anisometropy was 2.19 ± 1.86 and 1.95 ± 1.44 D in the control group and the fluoxetine group, respectively. A binocular basic test showed that 9 of 20 (45.0%) subjects in the control group and 13 of 22 (59.1%) subjects in the fluoxetine group exhibited suppression or abnormal retinal correspondence at a distance of 4 m. At a distance of 33 cm, 7 of 20 (35.0%) subjects in the control group and 8 of 22 (36.4%) subjects in the fluoxetine group showed suppression or an ARC at the start of the study. All subjects had impaired visual acuity (with displacement effect; visual acuity close to logMAR <0.7) at the start of the study, but only 2 subjects (10.0%) of the control group and 6 subjects (27.3%) of the fluoxetine group had it an abnormal contrast sensitivity at the mean spatial frequencies measured with the Pelli-Robson diagram. The mean visual basic parameters are summarized in Table 2.

The 10-week treatment regimen included a combination of medication, refraction correction, eye patches, and awareness training. Game-based perception training software has been specially developed to improve the use of the amblyopic eye when patching. The tasks of the game are shown in Fig. 2 and in additional video 1. A placebo control group was enrolled in the study for the medication only, and all participating subjects were prescribed the same daily patch and exercise instructions.

Schematic representation of the task design and the composition of the training game. For more information, see the Materials and Methods: Training Paradigm section. In short, the training program consisted of seven different games (Games 1-7) that focused primarily on visual acuity and contrast sensitivity in several attention and working memory tasks. The subjects were given a set selection of games for each training day. The total training time per week was ~ 3.5 hours, excluding the time spent adjusting the game parameters. For all tasks, the subject answered with a single keystroke or withheld the answer. Games 1 and 2 were visual tracking tasks with one or more objects, in which complex-shaped objects moved along curved paths on the screen and the subjects' task was to react when they saw a change in the function of one of the objects noticed. Different game segments had different numbers of objects to be visited (attention load 1, 2, 3 and 4). Before each game there was a calibration period with one (game 1) or two (game 2) objects during which the size of the feature change (C) was adjusted to achieve a recognition rate (HR) of 64-73%. Games 3 and 4 were visual chase games like Games 1 and 2 and had an identical calibration procedure and object mobility, but included only attention loads of 1 and 2 and had six characteristically different distractor objects imposing visual scrum under two out of four conditions. Game 5 was a continuous single object tracking task in which the subjects reported the functional changes of a single object (as in Games 1-4). Game 5 had no calibration, but began with very pronounced functional changes, which decreased by a factor of 1.6 in each of the 12 game segments, so that the test subjects were able to achieve segments 7–8 with a detection rate of> 25% on average. Game 6 was a go / no-go 1-back working memory task in which the subject was presented with an object in a quadrant with a duration of ~ 1 s and a stimulus rate of ~ 2.5 s. The subject's task was to indicate whether the object in the current stimulus was different from that in the previous stimulus, regardless of quadrant and object rotation. Game 7 was a task for recognizing stimulus thresholds, in which semitransparent complex visual objects were presented at random for 0.1 s and the test subjects' task was to report perceived stimuli. The object transparency was calibrated in such a way that a recognition rate of 0.5 at 0.5 A was obtained for an alpha level A. During the games, the objects were at five equiprobable levels of A, so A was 0, 0, 25, 0, 5, 0, 75 and 1, 0.

Full size image

Visual acuity

The visual acuity of the amblyopic eye improved significantly in both treatment groups (Fig. 3A, B). At the primary efficacy endpoint (10 weeks), the change in logMAR visual acuity from baseline was -0.167 [95% confidence interval (CI): -0.226 to -0.108; p <0.001] in the fluoxetine group and -0.194 (95% CI: -0.254 to -0.133; p <0.001) in the control group (3A, B and Table 3). The mean difference between treatment groups in improving visual acuity was only 0.027 logMAR units (95% CI: -0.057 to 0.110; p = 0.524). Nine subjects (42.8%) in the fluoxetine group and eight subjects (40.0%) in the control group improved visual acuity ≥ 0.2 logMAR units at the primary efficacy endpoint (FIG. 3C). Two subjects (9.5%) in the fluoxetine group and two subjects (10.0%) in the control group had improved to normal visual acuity (<0.1 logMAR) at the end of the 10-week treatment period.

Improve visual acuity. ( A. Scatterplots showing the visual acuity of each individual patient (amblyopic eye) at the beginning, at week 10 (end of treatment / training) and at week 22 (end of follow-up care), measured according to the ETDRS diagram (logMAR). The control group is shown on the left and the fluoxetine group on the right. The limit of normal visual acuity (logMAR 0) is shown hatched. ( B. ) Average change in visual acuity compared to baseline, measured using the ETDRS diagram (logMAR) at the beginning and after 2, 6, 10, 14 and 22 weeks. The mean +/- 95% CI at each point in time when visual acuity was determined by the ETDRS graph is displayed. ( C. ) Number of patients per group who had a visual acuity improvement of ≥0, 2 or <0, 2 logMAR units at weeks 10 and 22 compared to baseline.

Full size image

Full size table

The improvements in visual acuity observed at the 10-week primary endpoint visit were maintained in both groups during the 22-week follow-up visit (12 weeks after treatment discontinuation at the primary endpoint visit). In addition, many subjects in both groups had a visual gain of at least 0.2 logMAR units [9 of 20 fluoxetine subjects (45.0%), 6 of 14 control subjects (42.9%); Figure 3C]. The mean visual acuity in the control group changed from 0.636 (logMAR; 95% CI: 0.539 to 0.733) at the start of the study to 0.365 (95% CI: 0.245 to 0.485) after 10 weeks and to 0.420 (95% CI: 0.280 to 0.561) after 22 weeks. In the fluoxetine group, the mean visual acuity changed after 10 weeks from 0.645 (logMAR; 95% CI: 0.55 to 0.760) to 0.481 (95% CI: 0.345 to 0.618) and to 0.474 (95% CI: 0.331) to 0.617) after 22 weeks.

Binoculars, contrast sensitivity and crowded visual acuity

Similar to visual acuity, improvements in binocular vision, contrast sensitivity, and crowded visual acuity were observed in both the control group and the fluoxetine group. Binocular vision was rated for both near vision (33 cm) and distance vision (4 m). 22 subjects [9 subjects in the control group (45.0%), 13 subjects in the fluoxetine group (59.0%)] had suppression or ARC in the 4 m test at the start of the study (4A). After 10 weeks, the number of patients with suppression or ARC had dropped to 16 subjects [5 (31, 3%) subjects in the control group, 11 (55, 0%) subjects in the fluoxetine group]. This change persisted for 22 weeks in 11 subjects [3 (21, 4%) subjects in the control group, 8 (40, 0%) subjects in the fluoxetine group]. The results in the 33 cm test were similar for both groups (data not shown).

Binocular changes, contrast sensitivity, crowded visual acuity. ( A. ) Number of patients per group who had a change in binocular vision (suppression, abnormal retinal correspondence (ARC), or normal fusion) from baseline at weeks 10 and 22. The test results for striped Bagolini glass at a distance of 4 meters are displayed. ( B. ) Mean contrast sensitivity (logarithmic value) according to the Pelli-Robson diagram. The normal contrast sensitivity (1, 70) is indicated by a hatched line. Only two patients in the total patient population (n = 42) had a significant impairment in contrast sensitivity at the start of the study (ie 0 log). Both patients received fluoxetine and improved to near normal contrast sensitivity. ( C. ) Crowded near visual acuity as measured by Landolt C-ring diagrams. Normally crowded visual acuity was rated ≥0.7 (hatched line).In fields B and C, the direction of improvement is indicated by an arrow. ( D. ) Number of patients per group who improved from baseline in a densely populated visual acuity test at weeks 10 and 22.

Full size image

At baseline, contrast sensitivity was normal in all but two fluoxetine subjects (4.8%) with severe amblyopia (both had 0 log values ​​for contrast sensitivity at baseline). After fluoxetine treatment and perception training, both subjects had improved to an almost normal contrast sensitivity (after 10 weeks: 1, 20 and 1.95 log, after 22 weeks: 1.95 and 1.65 log; Fig. 4B).

Overcrowded areas near visual acuity were assessed using Landolt C-ring charts and were considered normal when optotypes were found to be 0, 7 or less. The mean of the baseline crowded near visual acuity in the amblyopic eye was 0.171 ± 0.116 in the control group and 0.165 ± 0.114 in the fluoxetine group. Improvements in these scores were seen after treatment in observed in both study groups (4C). At the primary endpoint (10 weeks), the control group had increased by 0.181 ± 0.027 (95% CI: 0.126 to 0.236) and the fluoxetine group by 0.148 ± 0.026 (95% CI: 0.095 to 0.201) improved, both p <0.001). After 22 weeks, the control group had increased by 0.221 ± 0.036 (95% CI: 0.150 to 0.293) and the fluoxetine group by 0.197 ± 0.033 (95% CI: 0.113 to 0.262, both p <0.001) improved. 4D shows the distribution of patients with normal (≥ 0.7) and impaired (<0.7) crowded visual acuity at baseline and at 10-week and 22-week time points. The primary and secondary outcome data for the primary efficacy endpoint (10 weeks) are summarized in Table 3.


A total of 66 adverse events (AEs) were reported after starting study treatment. Fifty-eight (87.9%) AEs occurred during treatment and eight AEs (12.1%) occurred after treatment. Only 16 AEs (24.2%) were associated or potentially associated with study treatment, and all were reported during the treatment period. Eleven (16.7%) of these side effects occurred in the fluoxetine group and 5 (7.6%) in the placebo group. None of the treatment-related side effects were reported after the 10 week treatment period, and none of the side effects resulted in withdrawal from the study. One subject in the fluoxetine group had a transient mild diplopia that resolved spontaneously. Other reported side effects were unrelated to visual function. One serious adverse event (benign ovarian cyst of moderate severity) occurred during the study but was unrelated to study treatment.


Amblyopia is a complex brain disorder that can interfere with everyday life due to the visual limitations associated with it. Despite good screening programs and effective childhood treatments, amblyopia remains a common cause of lifelong visual impairment, regardless of geographic location or ethnic origin 3, 5, 6, 38 . The combination of adequate refractive correction and occlusion therapy (patching of the non-amblyopic eye) was the main therapy for amblyopia of all etiologies. However, the benefits of various forms of occlusion therapy are greatest when therapy is started at an early age (<8 years). Therefore, early detection and treatment of amblyopia is the most important factor for successful visual results. Physiologically, the brain has the greatest plasticity in the critical phase of early postnatal life. However, the latest findings strongly suggest that the primary sensory cortex can remain plastic in adulthood (14, 15, 16, 17, 18, 19, 20, 21, 24, 25, 26, 27 ) . This finding suggests that there is a physiological basis for the treatment of amblyopia in adulthood that offers an opportunity to alleviate this global public health problem.

The current study looked at whether fluoxetine, an SSRI known to affect visual cortical plasticity in adult rats 15, May improve the effects of combination therapy of patching and computer-aided awareness training in adults with amblyopia. The response to treatment was good in both the fluoxetine and placebo groups. However, the change in visual acuity after 10 weeks of study medication / cognitive training therapy did not differ significantly between subjects who took fluoxetine and subjects who took a matching placebo. It is possible that 20 mg of fluoxetine, the dose commonly used at the beginning of treatment for depression, was too low a dose to modulate neuroplasticity. It may also be that the training paradigm (new glasses, patches, and computer-aided awareness training) was so effective that the 20 mg dose of fluoxetine did not provide any additional benefit. Larger doses of fluoxetine (up to 80 mg / day) are often used in depressed patients if the initial dose does not have the desired therapeutic effect. In addition, the 10 week treatment period may have been too short to maximize the benefits of fluoxetine and it is possible that a difference between treatment groups may have occurred after the longer treatment period.

Although there is evidence that some amblyopia treatments may be additive (optical correction combined with patching or atropine) 29, it has been documented in some reports that not all amblyopia treatment effects may be additive. The effects of multiple amblyopia treatment paradigms were not synergistic in rodent models of amblyopia 39 . In addition, environmental enrichment and treatment with fluoxetine have been shown to induce similar regression of amblyopia in rodents 25 . Therefore, it is possible that perceptual training alone (with refractive correction) promotes maximum visual cortical elasticity and that further treatments have no additive benefits. The current study should not determine this, and future studies should include a group of subjects treated with fluoxetine only (no cognitive training). However, one rodent study found that fluoxetine alone had no effect on eyesight 15 . In addition, there have been no previous reports of visual benefits of fluoxetine monotherapy in amblyopic patients, although millions of patients and probably thousands of amblyopic patients have used the drug over the past three decades. In addition, a recent placebo-controlled, double-blind clinical study showed that fluoxetine 20 mg / day for 19 days did not significantly affect visual perception learning in humans 40 . These earlier studies support the theory that fluoxetine 20 mg / day may not be large enough to effectively modulate visual cortical plasticity in adult humans.

The subjects of the current study received new glasses with the correct refraction correction at the start of the study. Initiating the use of appropriate Corrective glasses can amblyopia in children 41 and adults 42, 43 improve . Therefore, the use of new corrective lenses may have contributed to the visual gains observed in the current study. In addition, the variability of the test repetition should be taken into account. This was low in the present study because visual acuity was measured by the same observers under the same conditions and in the same locations. In addition, the regression to the mean must be taken into account; Since our study was placebo-controlled, the regression to the mean decreased because both groups were most likely to show the same tendency.

The subjects in this study had an average increase of approximately 2 lines of sight (0.2 logMAR), which is what was published with others Visual training protocols observed improvements was similar 35, 36 . Our results also agree with those of Li et al. 37who found that 33% of adult amblyopic subjects had a significant improvement in visual acuity after video game-based awareness training. A recently published randomized clinical trial using a video game with falling blocks and dichopic contrast offset versus a placebo game (no dichopic display) found only modest visual acuity gains (<0.1 logMAR) in older children, adolescents, and adults with amblyopia 44 . In this study in particular, the compliance requirements for the prescribed gaming behavior were rather low (> 25% of the prescribed minimum dose or at least 10.5 hours after 6 weeks), which could affect the results. In our study, the subjects completed> 85% of the computer-aided awareness training (at least 29 hours during the 10-week treatment period), which suggests that the intensity of the training can correlate with the degree of visual improvement.

Our study had several limitations. First, we didn't have a real control group with no treatment. Future clinical trials that combine multiple interventions should include multiple control groups to study the effects of each interventions and their combined effects. In addition, the fluoxetine dose and duration of use should be varied in future studies. Second, the response to treatment was remarkably different in the two treatment groups. This could be due to the large variability in the severity of amblyopia and etiology in our study population. Twenty-two subjects (52.4%) had abnormal binocularity at baseline, of which 10 had binocularity improvement during study treatment. In addition, only 2 subjects with severe amblyopia had low contrast sensitivity at the start of the study. Occasionally, contrast sensitivity deficits correlate with the visual acuity of the amblyopic eye 45, 46, 47 . Both subjects showed remarkable improvements in visual test results. Our results agree with those of Zhou et al. 48who showed that perceptual learning can improve visual acuity and contrast sensitivity in adult amblyopia patients. Nine of our subjects (21.4%) had improved near overcrowded visual acuity. Hussain et al. 49 . found a significant association between crowd reduction and visual acuity improvement in amblyopic adults. However, no correlation was observed between improvement in visual acuity and overcrowding, contrast sensitivity, or binocularity.

In summary, it can be said that both fluoxetine and software-based perception training were safe and well tolerated, with fluoxetine treatment not offering any further advantages over perception training. The software-based training tool developed for this study has proven useful for monitoring compliance with training guidelines and could be used in the future to personalize visual training in a clinical setting.

Materials and methods

All studies were carried out in accordance with the principles of the Declaration of Helsinki and in compliance with Good Clinical Practice. This study has been reviewed and approved by the Regional Ethics Committee of Tampere University Hospital (central process for all centers in Finland) and the Research Ethics Committee of the University of Tartu (Estonia). Written informed consent was obtained from all subjects prior to any investigation or procedure. The study was registered on October 1, 2010 in the European Clinical Trials Database (EudraCT) under the number 2010-023216-14.

This multicenter Phase 2 clinical study was a randomized, double-blind, placebo-controlled (drug) parallel group study to assess improvement in visual acuity of the amblyopic eye as measured by the Early Treatment of Diabetic Retinopathy Study Table (ETDRS) after 10 weeks of intake of Medication (20 mg fluoxetine or placebo) and computer-based training (with the dominant eye patch). The following assumptions were made to calculate the sample size: comparison of two groups of equal size, a difference between the groups (fluoxetine vs. placebo) in the change in logMAR visual acuity of at least 0.15, a standard deviation (SD) of 0.15 and a subject dropout rate of 10%. Thirty-four subjects had to be randomized to bring the study to 80%, assuming a bilateral Type I error rate of 5%.

Subjects of study

Between June 2011 and April 2013, 42 subjects were enrolled in four eye clinics in Finland and Estonia. The inclusion and exclusion criteria for the study are fully described in Table 1. Briefly, adult patients with monocular amblyopia without other ocular or neurological abnormalities were considered for enrollment. The subjects included were 19 to 57 years old and had moderate (0.3–0.6 logMAR difference) to severe (> 0.6 logMAR difference) amblyopia due to or a myopic or hyperopic anisometropia (≤ 4.25 D) congenital esotropia. The lower limit for anisometropy was not specified in the study protocol. The researchers considered amblyopia to be anisometropic if strabismus was not diagnosed in childhood and the childhood refractive error was at least 1 D anisometropia, determined as the spherical equivalent. Patients with other primary forms of strabismus, extrafoveal (eccentric) fixation, or on antidepressants in the past 6 months were excluded.

Study exams

Screening assessed suitability, demographics, medical history, relevant medications, vital signs, physical exam, blood and urine samples (including urine pregnancy test for fertile women), and amblyopia. Amblyopia was confirmed on screening and defined as the interocular ETDRS best-corrected visual difference of at least two lines and / or logMAR visual acuity between 0.30 and 1.10 in the amblyopic eye and 0.110 or better in the dominant eye. Before randomization, the patients received new glasses based on non-cycloplegic refraction in order to ensure the best possible vision correction during the study.

During the 26-week study period, a thorough ophthalmological examination was carried out at each of the seven planned visits (in weeks -2 (screening), 0 (randomization), 2, 6, 10, 14 and 22; Fig. 1). The eye tests included assessments of binocularity, visual acuity, overcrowding acuity, and contrast sensitivity and were performed with refractive error corrected. In addition, presbyopic correction was used to test closely spaced visual acuity in presbyopic subjects.

Binocularity was assessed using the Bagolini strip glass test 50 examined before the monocular test. Lens strips were placed using lorgnette frames at 135 ° in front of the right eye and 45 ° in front of the left eye. With this test set-up, each eye receives the same enamel pattern, with each fixation strip oriented perpendicular to the strips and 90 ° away from the other eye. The test enables the assessment of simultaneously perceived images with a minimal dissociative effect and was carried out under normal light conditions at close proximity (33 cm) and at a distance (4 m). Binocularity was categorized as suppression (1 light and only 1 line were seen), normal fusion (binocular single vision, BSV; 2 lines were seen as X and 1 light in the middle), abnormal retinal correspondence (ARK; harmonic if 1 light and 2 lines were seen but one of the lines was broken or out of harmony due to foveal suppression when 1 light and 2 lines were seen but the lines did not cross in the center where the light was lines were seen). None of the subjects in the current study had diplopia.

Visual acuity was measured under standardized lighting conditions (self-calibrated test lighting with a constant light level of 85 cd / m 2 ) using a large format standardized ETDRS light box (ESV3000 with LED lights, VectorVision, Greenville, OH) placed 4 m, rated by the subject. Three different ETDRS diagrams (Diagrams R, 1 and 2) were used to prevent subjects from memorizing eye diagrams. Visual acuity was first assessed on the amblyopic eye and measured as the number of correctly identified letters. A clinically relevant improvement in visual acuity was defined as a decrease in the logarithm of the minimum angle of resolution (logMAR) of visual acuity by 0, 2 or more (ie 2 lines or 10 characters in the ETDRS table).

Contrast sensitivity was determined under standardized lighting conditions using a Pelli-Robson chart at a distance of 1 m (charts A and B), using previously established age-dependent normative values 51 . Crowded near visual acuity was assessed using a specific crowded Landolt C ring chart booklet at a distance of 40 cm 52 . Crowded near visual acuity was defined by the smallest line in which the subject correctly identified at least 8 of 12 letters (≥66.7%). The right eye was tested first in all tests requiring charts. The contralateral eye was occluded during testing and charts were switched between eyes.All eye examination test charts and Bagolini striated glasses tests were standardized and validated for trial endpoint measurement. All staff involved in vision testing were masked to subject group assignment.

Treatment safety was assessed using ophthalmoscopy, biomicroscopy, intraocular pressure (IOP) measurement, laboratory safety tests [hematology (hemoglobin, hematocrit, erythrocyte count, leukocyte count, platelet count), clinical chemistry (alanine aminotransferase, alkaline phosphatase, aspartate aminotransferase, creatinine, gamma-glutamyl transferase, potassium, sodium, urea), urine analysis (blood, glucose, ketones, protein, pH)], vital signs, and physical examination performed at screening and at each study visit. Adverse events and changes in concomitant medications were recorded at each study visit.

Study medication

Fluoxetine capsules were manufactured by Orion Corporation (Espoo, Finland) and the matching placebo capsules were manufactured by Corden Pharma GmbH (Plankstadt, Germany). Study subjects were randomly assigned to receive either 20 mg fluoxetine each day (hard capsule) or a matching placebo. Randomization was done in a 1: 1 fashion in blocks of 4 and was stratified by site. Randomization was also stratified by amblyopia severity, determined using interocular visual acuity difference (moderate: 0.3-0.6 logMAR difference, severe:> 0.6 logMAR difference). The randomization structure was designed by a biostatistician and the final randomization list was generated by an independent person who had no contact with study subjects or study data. Medication was pre-packed and serially numbered so that subjects were assigned to a study group by giving them the next available medication number in the sequence. A drug accountability log was maintained by study-authorized personnel. The receipt, dispense and return of study medication was recorded in this log. Patients were instructed to return dispensed medication bottles at the next visit, even if the bottles were empty. The number of capsules dispensed and returned was reconciled against the number of days between the visits and any discrepancies were accounted for. After 10 weeks of receiving study medication, subjects were weaned off the daily medication (1 capsule every other day for the next 2 weeks, Fig. 1B).

Perceptual training

All subjects were prescribed daily computerized training with eye patching during the 10-week period of receiving study medication (Fig. 1A). The principle underlying the perceptual training software developed for this study is fully described in the Electronic Supplementary Materials and is illustrated in Fig. 2 and Supplementary video 1. All subjects received new spectacles before randomization and were instructed to wear an eye patch over their dominant eye while performing daily computerized perceptual training. The training software was used to track training compliance, which was calculated by dividing the total accomplished training time with the total prescribed training time. Training compliance was automatically reported to the study site prior to each scheduled visit.

All subjects were given an eye patch and were instructed to wear it over the non-amblyopic eye for 1 hour each day. Subjects were also instructed to complete approximately 30 minutes of the computer-based training each day while they were wearing the patch and their spectacles.

The training period was divided into ten 1-week segments and each subject played an identical composition of games each week. The maximum total training time over the 10-week training period was 35 hours. The training program was made up of seven different games wherein the performance was primarily determined by visual acuity and contrast sensitivity and secondarily by attention and mental effort. Thus, the training was primarily focused on visual acuity and contrast sensitivity and was aimed at their improvement. A schematic illustration of the training game task design and composition is shown in Fig. 2. In addition, the computerized training setup, training protocol structure, and individual training game design are described in detail in the Electronic Supplementary Materials. Data on behavioral performance were collected on a per-game basis. For each game type, the corresponding weekly test outcome measures were obtained by pooling the data from all individual games of that type played in that week (see Electronic Supplementary Materials).

Study results

The primary outcome of the study was an improvement in visual acuity in the amblyopic eye, as measured by the ETDRS chart, from baseline (week 0, randomization) to the 10-week visit (end of treatment). Secondary outcomes included the change from baseline in binocularity, contrast sensitivity and crowded near visual acuity to week 10 week. The persistence of changes observed at 10 weeks was also evaluated at the end of the follow-up period (week 22). Treatment safety was assessed using adverse event incidence and ophthalmological examination findings throughout the study. Exploratory outcomes included changes from baseline in training measures at each study visit.

Data analysis

The primary analysis population was the full analysis dataset (FAS), which included all randomized patients who had received at least one dose of study medication (intention-to-treat principle). A last observation carried forward (LOCF) imputation was applied up to week 10 for subjects who did not complete the study and those who were non-compliant with the treatment.

Differences in visual acuity and crowded near visual acuity between the two treatment groups were evaluated using the repeated measurements of analysis of covariance (RM ANCOVA) method with baseline values ​​as a covariate. The model included the study center, treatment and time point (visit) as main effects, and treatment by time point (visit) as an interaction effect. With regard to the primary endpoint, differences between the treatment groups with regards to change in the logMAR visual acuity at 10 weeks (and a 95% CI for the difference) were estimated using RM ANCOVA models with a contrast. A secondary RM ANCOVA analysis was used to compare the least square means between the treatment groups at the end of the follow-up (22 weeks) period to determine if treatment effects were maintained. Contrast sensitivity was not analyzed with RM models because of low variability in the dataset. Pearson's chi-square test was used to compare categorical variables. Some binocularity categories contained a low number of subjects. Therefore, differences between treatment groups in binocularity were evaluated using Fisher's exact test.

All statistical analyzes were performed using SAS software version 9.3 (SAS Institute Inc., Cary, NC, USA). P values ​​of less than 0.05 were considered to indicate statistical significance.


The authors thank prof. Lamberto Maffei for his input to the design of the study. We would also like to thank all participants for their time and efforts. The trial was funded by Herantis Pharma Plc, Helsinki, Finland.

Additional electronic material

  1. Further information

  2. Additional video 1

  3. CONSORT 2010 Checklist


By submitting a comment, you agree to our terms of use and community guidelines. If you find something abusive or that does not comply with our terms or guidelines, please mark it as inappropriate.