An “inverse approach” to the treatment of amblyopia

Recent laboratory finding suggest that short-term patching the amblyopic eye (i.e., inverse occlusion) results in a larger and more sustained improvement in the binocular balance compared with normal controls. In this study, we investigate the cumulative effects of the short-term inverse occlusion in adults and old children with amblyopia. A prospective cohort study of 18 amblyopes (10-35 years old; 3 with strabismus) who have been subjected to 2 hours/day of inverse occlusion for 2 months. Patients who required refractive correction or whose refractive correction needed updating were given a 2-month period of refractive adaptation. The primary outcome measure was the binocular balance which was measured using a phase combination task, the secondary outcome measures were the best corrected visual acuity which was measured with a Tumbling E acuity chart and convert to logMAR units and the stereo acuity which was measured with the Random-dot preschool stereotest. The averaged binocular gain was 0.11 in terms of the effective contrast ratio (z = -2.344, p = 0.019, 2-tailed Related samples Wilcoxon Signed Ranks Test). The average acuity gain was 0.14 logMAR equivalent (t(17) = 0.13, p < 0.001, 2-tailed paired samples t-test). The averaged stereo acuity gain was 253 arc seconds (z = -2.689, p = 0.007). Based on more recent research concerning adult ocular dominance plasticity, contrary to current practice, patching the amblyopic eye makes more sense; comparable acuity benefits, better compliance, better binocular outcome and applicable to adults as well as old children.


46
Occlusion of the fixing eye has been the gold standard treatment for amblyopia ever since it was first introduced in 1743 by Conte de Buffon [1]. It has evolved over the years; partial 48 rather than fulltime occlusion is now preferred and filters (i.e. Bangerter filters) [2], lenses 49 (i.e. defocus, or frosted) and eye drops (i.e. atropine) [3,4] have been used instead of 50 opaque patches. It is effective in over 53% of cases in improving acuity in the amblyopic 51 eye by more than 2 lines of logMAR acuity [5]. It does however leave something to be 52 desired in a number of aspects. Compliance can be low[6] because it restricts school 53 age children to the low vision of their amblyopic eyes for part of the day and also because 54 of its psychosocial side-effects [7]. There is a relatively poor binocular outcome even 55 though the acuity of the amblyopic eye is improved [8]. Its effects are age-dependent; 56 effectiveness is much reduced for children over the age of 10 years old [9,10]. Finally, it 57 is associated with a 25% regression rate once the patch has been removed [11,12]. It is 58 effective but far from ideal. Interestingly, the basis of this widely accepted therapy is poorly 59 understood. An explanation is often advanced in terms of "forcing the amblyopic to work" 60 by occluding the fixing eye, which prompts the question, what is stopping the amblyopic 61 eye from working under normal binocular viewing? This suggests that the problem of 62 improving vision in the amblyopic eye, far from being simply a monocular issue, must 63 have an underlying binocular basis (i.e., involving the fixing eye). Occlusion of the fixing 64 eye must be, in some way, disrupting what is normally preventing the amblyopic eye from 65 working when both eyes are open. Within the clinical literature this is known as 66 suppression and one supposes that occlusion affects suppression in a way that is 67 beneficial to the acuity of the amblyopic eye. 68 69 Recent laboratory studies have shown that short-term occlusion (i.e., 2 hours) is 70 associated with temporary changes in eye dominance in normal adults. There are two 71 things that are particularly novel about this new finding; first, these changes occur in 72 adults and secondly, the eye that is patched becomes stronger in its contribution to the 73 binocular sum. In other words, the eye balance is shifted in favour of the previously 74 patched eye. This was first shown by Lunghi et al (2011)[13] using a binocular rivalry 75 measure to quantify eye dominance. Since then there has been a wealth of information 76 on this form of eye dominance plasticity in normal adults using a wide variety of different 77 approaches [13][14][15][16][17][18][19][20][21][22][23][24][25].  [25] were the first to show that adults with amblyopia 78 also exhibited this form of plasticity and that it tended to be of larger magnitude and of a 79 more sustained form. They made the novel suggestion that it could provide the basis of 80 a new therapeutic avenue for amblyopes in re-establishing the correct balance between 81 their two eyes. Such a suggestion rests on the assumption that serial episodes of short-82 term occlusion can lead to sustainable long-term improvements in eye balance. The 83 hallmark of this form of plasticity is that, once the patch has been removed, the patched 84 eye's contribution to binocular vision is strengthened. Zhou et al (2013)[25] suggested 85 that to redress the binocular imbalance that characterizes amblyopia, it is the amblyopic 86 eye that would need to be occluded, opposite to what has been in common practise for 87 hundreds of years to improve the acuity in the amblyopic eye. Such a therapy, in principle, 88 would be primarily binocular in nature (addressing the binocular imbalance as a first step), 89 it would be expected to have much less compliance problems since it is not affecting the 90 day to day vision of the patient and since it has been demonstrated in adults, it could be 91 administered at any age. While this is well and good from a purely binocular perspective, 92 the obvious question is how would occlusion of the amblyopic eye on a long-term basis 93 (e.g., 2 hours or more a day for months) affect the acuity of the patched eye? The ethical 94 basis for such interventions is not in doubt, as there is evidence indicating that such 95 treatment is likely to be benefit rather than harm the vision of the amblyopic eye (including 96 children). In the 1960s, so-called inverse occlusion was sometimes used in an attempt to 97 treat eccentric fixation, which accompanies amblyopia in its more severe form. A review 98 of these studies [26][27][28][29][30] leads to two conclusions; first, inverse occlusion did not make the 99 amblyopia worse and second, acuity improved in the amblyopic eye in a percentage of 100 cases. The percentage of patients whose vision improved was significantly less than that 101 of classical occlusion in most [26,29,30], but not all [27,28] studies, which could arguably 102 be a consequence of the fact that studies on inverse occlusion were restricted to the more 103 severe and resistant forms of amblyopia. Therefore, on the basis of recent laboratory 104 studies on ocular dominance plasticity resulting from short term monocular occlusion [13-105 25] and previous clinical studies, on inverse occlusion designed to treat eccentric 106 fixation[26-30], we have two expectations; first that inverse occlusion (i.e., occlusion of 107 the amblyopic eye) should improve the binocular balance in patients with amblyopia and 108 second, that improved acuity of the amblyopic eye should also be expected. Two 109 additional benefits of this approach would be the expectation of better compliance, as the 110 fellow eye is not occluded and its applicability to older children and adults, since ocular 111 dominance plasticity occurs in adults. 112

113
To determine whether this radical departure from what is in common practice has any 114 benefit, we studied the effects of inverse occlusion of 2 hours /day for 2 months on a 115 group of 18 anisometropic and strabismic amblyopic teens and adults (10-35 years old), 116 an age range where classical occlusion therapy has low compliance[31]. Our primary 117 outcome measure was the binocular balance or ocular dominance. The second outcome 118 measures were visual acuity and stereo acuity. The results suggest that this approach 119 results in modest gains in both binocular balance and visual acuity within this older age 120 group, no adverse effects were encountered.  Table 1. Observers wore their prescribed optical correction, if 128 needed, in the data collection. Written informed consent was obtained from all patients, 129 or from the parents or legal guardian of participants aged less than 18 years old, after 130 explanation of the nature and possible consequences of the study. This study followed 131 the tenets of the Declaration of Helsinki and was approved by the Ethics Committee of 132  combination task), visual acuity and stereo acuity were measured before and after two 154 months of occlusion of the amblyopic eye for 2 hours/day (i.e., the inverse occlusion). For 155 patients who required refractive correction or whose refractive correction needed 156 updating (n = 9), a 2-month period of refractive adaptation was provided prior to the 157 inverse occlusion study ( Figure 1

160
Eighteen amblyopes with (n = 3) or without (n = 15) strabismus participated in our experiment. Patients' 161 binocular balance (balance point in the binocular phase combination task), visual acuity and stereo 162 acuity were measured before and after two months of occlusion of the amblyopic eye for 2 hours/day 163 (i.e., the inverse occlusion). For patients who required refractive correction or whose refractive 164 correction needed updating (n = 9), a 2-month period of refractive adaptation was provided prior to the 165 inverse occlusion study.

167 168
Since this approach is different from that currently used (i.e., classical occlusion therapy), 169 we were careful to conduct follow-up evaluations in accordance with the regulations from 170 the Amblyopia Preferred Practice Pattern® guideline ("PPP" 2017), P124: "If the visual 171 acuity in the amblyopic eye is improved and the fellow eye is stable, the same treatment 172 regimen should be continued". In particular, we conducted weekly visits in the pilot study 173 (in S1 to S13), rather than the 2 to 3 months that "PPP" recommends (P124 in "PPP": "In 174 general, a follow-up examination should be arranged 2 to 3 months after initiation of 175 treatment ") to ensure that the acuity in the amblyopic eye did not deteriorate as a result 176 of patching ( Figure 2). 177

178
We quantitatively accessed the binocular balance using a binocular phase combination 179 paradigm [34,35], which measures the contributions that each eye makes to binocular 180 vision. The design was similar as the one we used in previous studies [36,37], in which 181 observers were asked to dichoptically view two horizontal sine-wave gratings having 182 equal and opposite phase-shifts of 22.5° (relative to the center of the screen) through 183 polarized glasses; the perceived phase of the grating in the cyclopean percept was 184 measured as a function of the interocular contrast ratio. By this method, we were able to 185 find a specific interocular contrast ratio where the perceived phase of the cyclopean 186 grating was 0 degrees, indicating equal weight to each eye's image. This specific 187 interocular contrast ratio reflects the "balance point" for binocular phase combination 188 since the two eyes under these stimulus conditions contribute equally to binocular vision. (1) 205 (2) 206 Where L0 is the background luminance; C0 is the base contrast in the nondominant eye; 207 f is the spatial frequency of the gratings, δ is the interocular contrast ratio and θ is the 208 interocular phase difference. Surrounding the gratings, a high-contrast frame (width, 0.11º; length, 6º) with four white 212 diagonal lines (width, 0.11º; length, 2.83º) was always presented during the test to help 213 observers maintain fusion. 214 215 2.5 Procedure: We used the same phase adjustment procedure as used by Huang et 216 al[35] for measuring the perceived phase of the binocularly combined grating. In each 217 trial, observers were asked firstly to align the stimuli from the two eyes; they were then 218 instructed to adjust the position of a reference line to indicate the perceived phase of the 219 binocularly combined grating. Since the gratings had a period of 2 cycles corresponding 220 to 180 pixels, the phase adjustment had a step size of 4 degrees of phase / pixel (2 cycles 221 × 360 phase-degree / cycle / 180 pixels).

235
In the pilot study, we firstly conducted a 0.5-month of inverse occlusion (2 hours/day) in 236 S1 to S13. We found that amblyopic eye's visual acuity improvement in 5 of the 13 237 patients after 2 weeks of treatment, with no cases of acuity loss in the amblyopic eye. 238 Visual acuity of the fellow eye was stable in all cases. We then extend the occlusion 239 period to 1 month and 9 of 13 patients were found to exhibit small gains in visual acuity.

260
Since we could not have a control group who were denied any treatment, there is always 261 the possibility that improvements in visual acuity measured at different time points are 262 simply due to learning effects. To test this, we recorded the stability of acuity measured 263 for the untreated fellow eye, as a similar learning effect should apply. In Figure 3 Once we had shown that inverse occlusion can be undertaken in a safe fashion, we added 281 5 additional patients (S14 to S18) to the original study cohort of 13 (S1 to S13). These 282 additional patients followed the same protocol as the original thirteen (S1 to S13), but 283 visual functions were only measured before and after 2 months of treatment. A summary 284 of the main result for all the 18 patients is shown in Figure 4 for the measures of ocular 285 balance, visual acuity and stereo acuity. Measurements before and after 2-month of 286 treatment are plotted against one another. In term of ocular balance, the measure used 287 is the interocular contrast that is required to achieve a binocular balance. By binocular 288

319
Amblyopes exhibit a range of binocular balances ranging from less than 0.04 to 0.82 320 ( Figure 4A). Inverse patching of 2 hours/day for 2 months improves some more than 321 others. Six subjects showed no improvement, the other patients showed varying levels of 322 improvement, meaning that their amblyopic eye was contributing more to binocular vision. 323 Overall, the averaged improvement was a 0.11 change (0.30 ± 0.052 (Mean ± S.E.M.) to 324 0.41 ± 0.058) in the effective contrast ratio (Square symbol), which was significant based 325 a 2-tailed Related samples Wilcoxon Signed Ranks Test: z = -2.344, p = 0.019. Our 326 patients exhibited a range of acuity deficits ranging from less than 0.22 to close to 1.40 327 logMAR ( Figure 4B). As expected, the acuity improvements were of varying degrees. 328 Three patients showed no improvement at all, while all the other patients did exhibit 329 improvements to varying degrees (shaded area). The averaged improvement (solid 330 symbol) was 0.14 logMAR (from 0.70 ± 0.085 to 0.56 ± 0.070), which was significant 331 based on a 2-tailed paired samples t-test: t(17)=0.13, p < 0.001.. This magnitude of acuity 332 gain is similar to the results of a recent PEDIG study using classical occlusion of the same 333 duration (i.e. 2 hours/day for 16 weeks) in patients of a similar age range[38]. The 334 averaged stereo acuity gain was 253 arc seconds (z = -2.689, p = 0.007, 2-tailed Related 335 samples Wilcoxon Signed Ranks Test). This is a very conservative estimate because 336 14/18 patients had stereo acuities outside of our measurement range and were 337 conservatively scored at 800 arc secs, the largest disparity tested. This means that the 338 true stereo acuity gain could be larger than 253 arc seconds. 339 340 These changes in binocular balance, visual acuity and stereo acuity are modest but still 341 impressive considering the fact that the period of occlusion was relatively short (2 hours), 342 the duration of the treatment limited to 2 months and it involved an older age group. One 343 interesting finding is that the improvements in balance and visual acuity are not 344 significantly correlated (p = 0.76, Spearman's correlation), so it is unlikely they have a 345 common basis. 346

347
These improvements are long lasting as we have followed four patients (S12, S14, S16 348 and S17) for 1 month and one (S9) for 5.5 months after finishing the 2-month of reverse 349 occlusion regime, which showed that the outcomes were sustained ( Figure 5).

358
In our study, patients' age ranged from 10 years old to 35 years old. Interestingly, all 360 patients who were younger than 14 years old had visual acuity gain. However, for patients 361 older than 14 years old, only 62.5% of them had a visual acuity gain. A Spearman 362 correlation analysis showed that there was a positive correlation between the 363 improvement in visual acuity of the amblyopic eye and the patients' age, i.e., the younger 364 the patients the more the visual acuity gain (Rho = 0.534, p = 0.022). The correlations 365 between patients' age and the binocular balance gain or the RDS stereo acuity gain were 366 not significant (p > 0.3). 367

368
The refractive correction needed updating in half of the patients (n = 9), and a 2-month 369 period of refractive adaptation was provided before inverse occlusion was commenced. 370 Even though the acuity gains from optical treatments have been shown to be modest after 371 5-6 weeks of refractive adaptation [39], since those observations were in a much younger 372 age group, there could still be an argument that our findings were due to the refractive 373 correction per se occurring after our 8-week period, rather than the inverse occlusion. To 374 assess this, we divided our patients into two subgroups, i.e., those who required refractive 375 adaptation (n = 9) and those who did not (n = 9). We found no significantly different of 376 visual outcomes in these two subgroups, in terms of the improvement of the amblyopic 377

384
The rationale for this study comes from the recent findings on ocular dominance plasticity 385 in normal and amblyopic adults [13][14][15][16][17][18][19][20][21][22][23][24][25], the finding that short term patching results in a 386 strengthening of the contribution of the previously patched eye to binocular vision. This 387 study, which applies this to amblyopia, raises three interesting issues that are relevant to 388 the treatment of amblyopia. First, it highlights just how poor our understanding of the basis 389 of classical occlusion therapy is. How is it that acuity improves in the amblyopia regardless 390 of which eye is occluded? This does not just come from this study; there is a literature on 391 the acuity improvements that occur as a result of inverse occlusion. While in most cases 392 these improvements are much less than that of classical occlusion, there are studies [27, 393 28], where it is comparable to that of classical occlusion. The standard explanation of 394 occluding the fixing eye to "forcing the amblyopic eye to work" is untenable. What is 395 preventing the brain using information from the amblyopic eye under normal viewing 396 conditions? Whatever it is, occlusion must be preventing (i.e., disinhibiting) it from 397 operating. The problem must be essentially binocular in nature, which is why it is not 398 critically dependent on which eye is occluded to disrupt the anomalous interaction. We 399 would normally think about this anomalous binocular interaction as a suppression of the 400 amblyopic eye by the fellow eye, but on the basis of the occlusion of either eye being 401 effective, it may be better to think of suppression as simply a reflection of a binocular 402 imbalance. Recent psychophysics [40] and animal neurophysiology [41] suggest that the 403 problem is not because the inhibition from the fixing to the amblyopic eye is greater but 404 because the matching inhibition from the amblyopic eye is less. It is due to a net 405 imbalance in interocular inhibition. The resulting net imbalance can be disrupted by 406 occluding either eye and it's the duration of relief from this imbalanced binocular inhibition 407 that may result in an acuity benefit for the amblyopic eye. 408 409 Ocular dominance plasticity in normals is an all-or-none, homeostatic process and would 410 not be expected to have accumulated effects over time [42]. In amblyopes, ocular 411 dominance plasticity has different dynamics, being much more sustained [25]. The present 412 results suggest also that it can exhibit accumulated effects in amblyopes that result in 413 long lasting changes in eye balance. These sustained changes are however modest in 414 size and it will be necessary to explore how the magnitude of this effect can be increased The finding that the binocular balance and the monocular acuity improvements from 422 inverse patching are not correlated suggests that a simple explanation in terms of reduced 423 suppression is not viable. The two visual improvements are likely to have separate causes 424 and possibly involving different sites in the pathway. The acuity improvement for the 425 amblyopic eye is not dependent on which eye is occluded, as shown here ( Figure 4B), 426 but the direction of the binocular balance change is dependent on which eye is 427 occluded [13,25]. This distinction between binocular balance and monocular visual acuity 428 is an important one and should be incorporated into future clinical treatment studies. 429 Finally, apart from the additional benefit of a better binocular balance, its applicability to 430 older children and adults should not be underestimated, nor should the better compliance 431 that should follow from the patching of the amblyopic rather than the fixing eye. 432 Application to younger children would necessitate weekly visits to ensure that the acuity 433 in the amblyopic eye did not deteriorate as a result of patching. 434 435

Relevance of a recently published study 436
During the writing up of this paper, another study was posted on bioRxiv that is highly 437 relevant and supportive of the present approach (Lunghi et al (2018); doi: improvements when treating amblyopia. This in turn was based on their previous finding 445 that exercise can enhance plasticity in normal adults ([18], but also see [23]). This 446 published study and the current one both suggest that inverse occlusion can provide long 447 term benefits in visual acuity, stereopsis and sensory balance. Lunghi et al find that six 448 short treatment duration in this older age group, future studies would need to assess this 470 for longer treatment durations and younger age groups.