**Purpose**:
To assess whether transcorneal electrical stimulation (TcES) current-dependently slows progressive loss of visual field area (VFA) in retinitis pigmentosa (RP).

**Methods**:
Data from 51 patients with RP who received monocular TcES treatment once weekly over 1 year in an interventional, randomized study have been analyzed a posteriori. Current amplitudes were 0.1 to 1.0 mA in the TcES-treated group (*n* = 31) and 0.0 mA in the sham group (*n* = 20). VFA was assessed in both eyes (semiautomatic kinetic perimetry, Goldmann targets V4e, III4e). Annual decline rate (ADR) of exponential loss and model-independent percentage reduction of VFA at treatment cessation were correlated to current amplitude.

**Results**:
For V4e, mean ADR was −4.1% in TcES-treated eyes, −6.4% in untreated fellow eyes, and −7.2% in placebo-treated eyes; mean VFA reduction in TcES-treated eyes was 64% less than in untreated fellow eyes (*P* = 0.013) and 72% less than in placebo-treated eyes (*P* = 0.103). Individual VFA reductions correlated with current amplitude (*P* = 0.043) and tended toward zero in patients who received 0.8 to 1.0 mA. For III4e, there was a marginally significant current-dependency of interocular difference in reduction (*P* = 0.11). ADR and VFA reduction did not significantly correlate with baseline VFA.

**Conclusions**:
Loss of VFA (V4e) in patients with RP was significantly reduced in treated eyes compared to untreated eyes by regular use of TcES in a dose-dependent manner. No dependence of effects on the initial extent of VFA loss was found.

**Translational Relevance**:
TcES provides potential for preservation of visual field in patients with RP.

^{1}It seriously impacts the quality of life of patients with RP.

^{2}Typically, RP begins with a degeneration of rods in the peripheral retina followed by a loss of cones. The visual field area (VFA) and the corresponding retinal area decrease exponentially over time,

^{3}

^{,}

^{4}with an annual decline rate (ADR) between 5% and 17%, depending on the genetic cause, measurement of VFA, and definition of ADR.

^{3}

^{–}

^{7}

^{8}stem cell therapies,

^{9}optogenetic therapies,

^{10}and electronic implants

^{11}are under development. However, with few exceptions, such as gene therapy for RPE65-associated retinal dystrophy,

^{12}no such method is yet available for routine clinical use.

^{13}

^{,}

^{14}A physical treatment approach is transcorneal electrical stimulation (TcES), which has been described as a promising strategy for RP.

^{15}

^{–}

^{19}TcES has a manifold body of evidence from large clinical trials and is available as therapy in Europe.

^{20}

^{–}

^{24}It aims to activate neuroprotective factors and pathways in the retina and retinal pigment epithelium to enhance survival or regeneration of photoreceptors and thereby halt or, at least, slow disease progression. In mathematical terms, TcES intends to increase the time constant of the exponential progression or, equivalently, to decrease the ADR.

^{15}

^{–}

^{17}). The cellular and subcellular effects in the retina elicited by electrostimulation depend on the stimulus intensity.

^{25}

^{–}

^{29}Clinical studies showed that TcES caused a significant increase in blood flow to the central retina,

^{30}

^{,}

^{31}increased oxygen consumption by retinal cells,

^{23}improved visual acuity,

^{22}

^{,}

^{30}

^{,}

^{32}slowed visual field loss

^{21}or improved visual field,

^{30}

^{,}

^{33}improved b-wave amplitudes, and shortened electroretinography latencies.

^{21}

^{,}

^{22}

^{,}

^{33}For a comprehensive summary of clinical studies of TcES in RP, see Liu et al.

^{24}

^{34}). To date, however, no dose–response curves have been evaluated that establish a relationship between the intensity of TcES and clinically relevant determinants such as visual field. We conducted an a posteriori analysis of VFA data of an earlier clinical study (EST2 trial

^{21}) to explore the hypothesis that TcES can slow the progression of the decline in VFA in RP and that the treatment effect depends on the current strength. Additionally, we investigated whether the TcES effect depends on the VFA at baseline and whether TcES shows a current-dependent effect on safety parameters.

^{2}. For further inclusion and exclusion criteria and details of patient population, study protocol, and methods, see the original publication.

^{21}

^{2}) were quantified using the built-in software algorithm.

_{e}VFA%). The log

_{e}VFA% values of the visits in the treatment period from weeks 1 to 52 were plotted as a function of time (Figs. 3A2–F2 [V4e] and Supplementary Fig. S4 [III4e]) and subjected to a linear regression analysis with the intercept fixed to zero. To investigate aftereffects of TcES treatment after discontinuation of stimulation, data from the follow-up visit in week 78 were included in a second linear regression of the log

_{e}VFA% values. The slope s (unit: 1/y) of the best-fit line was used to calculate the ADR of the group according to Equation (1). A negative ADR means a decrease of the VFA over time, and a positive ADR represents an increase.

_{e}VFA) and subjected to a linear regression (Fig. 2D) from weeks 1 to 52 and additionally from weeks 1 to 78. A selection of log

_{e}VFA courses is presented in Supplementary Figure S1. The slopes s of the best-fit lines and interocular differences Δs in the slopes of the individual log

_{e}VFA courses of the treated and untreated eyes were correlated to the mean current amplitudes (Fig. 4). As the slope has a negative sign if VFA decreases over time, a positive Δs means a slower decrease in the VFA of the treated eye than in the untreated eye.

_{1,abs}and R

_{0,abs}of the VFA from baseline to week 52 in the treated and untreated eyes for each patient (Fig. 2C) and the percentage reductions R

_{1}and R

_{0}by dividing R

_{1,abs}and R

_{0,abs}by the respective VFA value at visit 1 (baseline). We then analyzed the dependence of R

_{1}and R

_{0}and the interocular difference ΔR = R

_{1}– R

_{0}on the individual mean current amplitudes (V4e: Figs. 5D–F, III4e: Supplementary Figs. S5D–F). As R

_{1}and R

_{0}have a positive sign if VFA decreases over time, a negative ΔR indicates less loss of VFA in the treated eye than in the untreated eye after 1 year. With an ideal exponential course of the VFA progression, percentage reduction multiplied by –1 would correspond to the ADR. We further analyzed a possible linear dependence of the ADR and the reductions R on the initial VFA (Fig. 6 [V4e], Supplementary Fig. S6 [III4e]).

^{35}(Supplementary Figs. S2B, E). The percentage test–retest variability (TRV) was calculated for each eye using Equation (2)

^{36}:

_{.95}) as the 95th percentile of the distribution of the individual TRV values (Supplementary Figs. S2C, F). These measures have also been used to characterize the interocular variability between the treated and untreated eyes (Supplementary Fig. S3).

_{1}and R

_{0}were performed using the paired Wilcoxon signed rank test. Tests of a difference between groups of unequal size and unpaired samples (treated group versus sham group) were done using the Mann–Whitney

*U*test. To test for a linear effect of current strength, a general linear model for the slopes s of the log

_{e}VFA% courses, Δs, R

_{1}, R

_{0}, and ΔR with mean current strength as an independent continuous variable was fitted. To test for a (possibly nonlinear but monotone) effect of current range on visual field decline, a Jonckheere–Terpstra (nonparametric) test on R

_{1}, R

_{0}, and ΔR was employed.

*r*²), and

*P*value. To compare slopes of two regression lines, a

*t*-test was performed. The reductions R

_{1}and R

_{0}and difference ΔR are presented descriptively using the mean, standard deviation, median, and quartiles, as well as boxplots.

*r*² = 0.931, III4e:

*r*² = 0.886). Mean variability of the interocular difference in VFA (Supplementary Figs. S3C, F) was 12.9% ± 16.5% (III4e: 19.8% ± 19.3%). Test–retest variability using VFA values from screening and baseline visits were 8.1% for V4e and 13.9% for III4e. Further details of the analysis of test–retest variability and interocular variability are shown in Supplementary Table S1, Supplementary Figure S2, and Supplementary Figure S3.

*P*= 0.233). The distributions of the mean amplitudes of the administered stimulation currents in the 150% and 200% EPT groups (Table 2B, Fig. 1B) also were statistically not different (

*P*= 0.623). Aggregated from both groups, the mean current amplitudes span a continuum with almost equally distributed values from 0.3 to 1.0 mA (Fig. 1C).

*r*² = 0.122,

*P*= 0.012; target III4e, data not shown:

*r*² = 0.154,

*P*= 0.004). Consequently, also the mean current amplitude correlated weakly but significantly with the baseline VFA (target V4e, Fig. 1E:

*r*² = 0.198,

*P*= 0.012; target III4e, data not shown:

*r*² = 0.248,

*P*= 0.004). Within the groups T2, T3, and T4, the current amplitudes were scattered over the full range of VFA.

_{e}VFA% courses in the treated and in the untreated eyes (V4e: Figs. 3A2–F2, III4e: Supplementary Figs. S4A2–F2).

_{e}VFA% courses during the treatment period from weeks 1 to 52 and additionally to the courses from weeks 1 to 78 (V4e: Figs. 3A2–F2, III4e: Supplementary Figs. S4A2–F2). Within the limits of the 95% confidence intervals (Table 3), the slopes in both periods were essentially the same. Both for target V4e and for III4e and for both fitting intervals, the slope of the best-fit line for TcES-treated eyes (V4e: Fig. 3C2, III4e: Supplementary Fig. S4C2) was less steep than for placebo-treated eyes in the sham group (Fig. 3A2, Supplementary Fig. S4A2; slopes, see Table 3). In the most stimulated subgroup T4 (target V4e), there was no statistically significant linear relationship of log

_{e}VFA% of the TcES-treated eyes with time (weeks 1–52:

*P*= 0.34, weeks 1–78:

*P*= 0.15).

*P*= 0.043), also in the subgroup T4 (weeks 1–52:

*P*= 0.014, weeks 1–78:

*P*= 0.002). In T4, the slope for weeks 1 to 78 for the treated eyes was statistically significant different from the slope in the untreated fellow eyes (

*P*= 0.006). The slope for untreated eyes was statistically significant less steep in the TcES-treated group T than in the sham group (weeks 1–52:

*P*= 0.022, weeks 1–78:

*P*= 0.021). No significant differences in the slopes between groups and eyes were found for target III4e (Supplementary Fig. S4;

*P*values, see Table 4).

_{e}VFA% course in the fitting interval weeks 1 to 78 (group T in Table 3) was smaller than the ADR of the placebo-treated eyes, both for target V4e (−4.1% vs. −7.2%) and for target III4e (−7.8% vs. −8.6%). It was also smaller than the ADR for untreated fellow eyes (V4e: −4.1% vs. −6.4%, III4e: −7.8% vs. −8.0%). In the subgroup T4, the V4e ADR of the treated eyes was −1.4% compared to −6.0% in the untreated eyes (III4e: −6.5% vs. −5.2%). ADRs for fitting interval weeks 1 to 52 can be seen in Table 3.

_{e}VFA courses for V4e showed a slight, statistically not significant linear dependence on the mean current strength (Figs. 4A–C) for both the treated eyes (

*r*² = 0.05,

*P*= 0.12) and the untreated eyes (

*r*² = 0.03,

*P*= 0.25). No tendency for a correlation of slopes with current amplitude was found for III4e (Figs. 4D–F, treated eyes:

*r*² = 0.0,

*P*= 0.67; untreated eyes:

*r*² = 0.0,

*P*= 0.83).

_{1}and R

_{0}were 7.5% ± 10.5% and 9.7% ± 13.3% (mean ± SD), with a mean of the interocular difference ΔR of −2.2% ± 9.5% and a zero median of the difference (Table 5 and Fig. 5A). In the treated group T, the mean R

_{1}and R

_{0}were 2.1% ± 7.7% and 5.8% ± 10.3%, with a mean difference ΔR of −3.7% ± 11.6% and a median of the difference of −4.8% (Fig. 5B). The two distributions R

_{1}and R

_{0}were statistically significantly different (

*P*= 0.013), and the difference between R

_{1}in group T and in the sham group missed statistical significance (

*P*= 0.103). On average, R

_{1}was 63.8% less in the TcES-treated eyes than R

_{0}in the untreated fellow eyes and 72.0% less than R

_{1}in the placebo-treated eyes. In the sham group, R

_{1}was 29.3% less than R

_{0}. In the subgroup T4, R

_{1}was −1.0% ± 8.1%, R

_{0}was 4.9% ± 9.0%, and ΔR was −5.9% ± 10.3%, with a median difference of −8.2% (Fig. 5C). The difference in the two distributions R

_{1}and R

_{0}missed statistical significance (

*P*= 0.098), whereas R

_{1}was statistically significantly different from R

_{1}in the sham group (

*P*= 0.036).

*r*² = 0.078,

*P*= 0.047) between the reduction R

_{1}in the treated eyes and the current amplitude (Fig. 5D) for target V4e. The line of best fit indicated zero reduction (R

_{1}= 0) at stimulation with 0.9 mA. No significant linear correlation with current amplitude was found for the reduction R

_{0}in the untreated eyes (

*P*= 0.46) and for the difference ΔR (

*P*= 0.38). In the ordinal model (Figs. 5G, I), the nonparametric Jonckheere−Terpstra test confirmed a significant decrease in R

_{1}with increasing current strength (

*P*= 0.043) and a tendency for ΔR (

*P*= 0.052). The number of pairs of eyes in which the reduction of VFA (target V4e) in the treated eye was smaller than in the untreated eye (ΔR < 0 in Fig. 5I) was 50% in the placebo-treated sham group (10 of 20); 77% in the aggregated TcES-treated groups T1, T2, and T3 (17 of 22); and 89% (8 of 9) in subgroup T4.

_{1}and R

_{0}could be found (Table 5, Supplementary Figs. S5A–C). The linear correlations between the reductions and the current strength were statistically not significant (

*P*values, see Supplementary Figs. S5D–F). The Jonckheere−Terpstra test revealed a marginally significant correlation of ΔR with increasing current ranges (

*P*= 0.11, Supplementary Fig. S5I).

_{1,abs}, sham:

*P*= 0.002, T:

*P*= 0.054) and untreated eyes (R

_{0,abs}, sham:

*P*= 0.003, T:

*P*= 0.001) with the baseline VFA in the respective eyes (Fig. 6A). The slope m of the best-fit lines for R

_{1,abs}and R

_{0,abs}was smaller in the TcES-treated group than in the sham group, and the slope for the interocular difference was the same in both groups (slopes, see Supplementary Table S3). For target III4e (Supplementary Fig. S6A), no interocular difference in the TcES-treated group was found. No significant correlation with the baseline VFA was found for the percentage reductions (Fig. 6B, Supplementary Fig. S6B), the ADR (Fig. 6C, Supplementary Fig. S6C), and their respective interocular differences for both V4e and III4e (

*P*values, see Supplementary Table S3). The slopes of the log

_{e}VFA courses and the percentage reductions also did not correlate significantly with the interocular difference in VFA at baseline (

*P*> 0.05, data not shown).

*n*= 58). The relationship to the device was “certain” in 38 cases (33 of them were dry eye symptoms), “probable” in 3 cases, “possible” in 6 cases, and “unlikely” in 7 cases, and no relationship to the device was assigned to 7 AEs. The outcome was described as “recovered” in 42 cases, “improved” in 18 cases, and “unchanged” in 1 case (mild ocular discomfort in the sham group).

^{21}and included both the sham group and fellow-eye controls. This allowed both interocular comparison of the treatment effects of the monocularly applied current and comparing the effects in the TcES-treated group with the natural changes in the sham group.

^{4}

^{–}

^{6}The ADR of the untreated fellow eyes in the TcES-treated group was smaller than the ADRs of untreated and placebo-treated eyes in the sham group, indicating a contralateral attenuated slowing effect of unilateral stimulation. This could explain the flat time course of VFA of the untreated eyes in subgroup T4 (Fig. 3F1, Supplementary Fig. S4F1). It could also explain the weak, nonsignificant current dependence of the slopes of the individual log

_{e}VFA courses in untreated eyes (Fig. 4B). Mean test–retest variability was 8% for the V4e target and 14% for the III4e target, lower than published values for the semiautomatic kinetic perimetry.

^{36}The interocular differences contributed more to the statistical variability than the intervisit differences. At baseline, interocular variability was 13% for target V4e and 20% for III4e. The distinctly larger variabilities for target III4e may explain the lack of significance of the results.

_{e}VFA courses in treated eyes showed only a weak, statistically nonsignificant correlation with current intensity (Fig. 4A). In addition to the large variation of the individual values, the marginal significance may also be due to a possibly delayed onset of the treatment effect on the progression course. Assuming an exponential decline, the slope of the best-fit line from the linear regression analysis of the log

_{e}VFA progression is the suitable measure to describe the longitudinal course of the VFA, as shown by the semi-log plots in Figure 3 and Supplementary Figure S1. However, if stimulation causes the progression to gradually slow down, the curve of visual field decline will over time deviate more and more from the course of a simple decreasing e-function. In the best case, the visual field might even increase again after some time, as indicated by the course of the VFA of the TcES-treated eyes shown in Figure 3 C. If so, calculating the ADR from the slope from the regression line over the entire study period will result in an underestimation of the final effect size. This argument is supported by the model-independent VFA reduction within a year (R

_{1}in Table 5) that was smaller than the annual decline derived from the log

_{e}VFA% courses of TcES-treated eyes (Table 3).

_{e}VFA has the advantage of using more measurements and is thus less sensitive to variability, but it has the disadvantage of assuming an exponential decline, which may not be valid.

_{1}in the treated eyes and the current strength yielding zero reduction at 0.9 mA (Fig. 5F). In the range of 0.8 to 1.0 mA, 6 of 9 treated eyes (66%) even had an increase of the VFA, compared to 1 of 9 (11%) untreated fellow eyes and 5 of 20 (25%) placebo-treated eyes. This suggests that TcES treatment is most effective above 0.8 mA. However, this study had a small sample size, and larger trials are needed to confirm the effective dose for TcES. Not all patients with RP should be treated indiscriminately with 0.8 to 1.0 mA. In practice, the stimulation dose must be adjusted to the individual tolerance level at which patients can withstand TcES for 30 minutes. Limits for safe current densities at the ocular surface must also be considered.

^{25}

^{,}

^{26}

^{,}

^{29}An important role in the neuroprotective effects of ocular stimulation has been attributed to the Müller cells (MCs). Enayati et al.

^{25}found in cultured MCs from a mouse model that electrical stimulation enhanced the MC proliferation and expression of photoreceptor progenitor cell markers via calcium signaling, which is mediated by voltage-gated calcium channels. As the triggering of Ca

^{2+}channels is linked with the activation of subcellular biochemical cascades related to neuroprotective pathways in the retina,

^{25}we postulate that TcES dose-dependently activates subcellular pathways, which leads to cellular and thus potentially clinically relevant protective effects in the retina.

^{33}found in a randomized and controlled trial with 24 patients a significant increase of the mean VFA (target III4e) by 9% after 6 weeks of consecutive stimulation with 150% EPT. The improvement was still present at the follow-up visit 11 weeks after termination of the stimulation. Bittner and Seger

^{32}reported on the longevity of effects for three patients with RP who received 0.75 mA. They were positive responders from their previous randomized controlled trial of TcES,

^{30}in which four of seven patients showed improvement in VFA (target III4e) and corresponding retinal area, respectively, after six stimulations (current amplitude 0.75 mA) at 1-week intervals. Sinim Kahraman and Oner

^{22}found a similar result in a large observational trial with 101 treated and 100 untreated patients with RP. They observed a significant improvement by 1.67 dB of the mean deviation of the visual field from the age-corrected norm in the central 30° of the visual field after only 4 weeks of stimulation with 150% EPT. The improvement attenuated partially 4 months after cessation of stimulation but remained at a higher level than at baseline.

^{23}found an increased oxygen consumption by retinal cells, suggesting an increased metabolism by TcES. This may indicate an initial boost leading to measurable improvements of VFA and visual acuity.

^{22}that the protective effects of TcES are transient and suggests that the degenerative processes resume when stimulation is ceased. Hence, chronic TcES is required to permanently delay photoreceptor degeneration. However, whether the peculiar pattern in the averaged VFA% time courses is indicative of a treatment effect or due to variability in the data remains to be shown.

_{1}, the ADR calculated from the slopes of the log

_{e}VFA courses, and the interocular differences in ΔR and in ADR (represented by Δs) did not correlate significantly with the VFA at baseline (Supplementary Table S3, Fig. 6). This result is consistent with an exponentially decreasing function as it is used to describe the VFA progression in RP. For exponential decline over time, the absolute annual decline depends on the initial value at the beginning of the observation period, whereas the percentage decline is independent of the baseline value. However, since we only evaluated data from the kinetic perimetry, no generally valid statement can be made as to whether the efficacy of TcES treatment depends on the severity of the disease.

^{36}Thus, there is a possibility that the evaluated VFA values do not reflect the true size relations of the corresponding retinal area, especially of peripheral areas.

^{37}In our case, however, this is unlikely to cause error because we did not evaluate the absolute VFA but considered relative longitudinal changes of the visual field with respect to the baseline value.

^{20}

^{–}

^{22}

^{,}

^{33}The EPT is a subjective measure that varies greatly between patients (Fig. 1A) and spans a wide range of values.

^{34}It depends on many individual factors, including the cause and the state of degeneration of the retina.

^{34}

^{,}

^{38}Thus, defining individual stimulation strength as a factor of individual EPT necessarily results in a broad distribution of current amplitudes. If cellular effects of TcES depend on the stimulus strength and grouping patients according to a multiple of EPT, group differences may be lost due to the averaging of measured dose-dependent variables within groups. The effects are further attenuated when the current amplitude is adjusted to EPTs several times, as has been done in the EST2 trial. This may at least partially explain the high variability of the obtained data and the inconsistent results from the past studies.

^{39}In the trial presented here, the observed side effects of TcES (Table 6) were generally mild and transient in nature. The most frequent side effect was dry eye symptoms during treatment, which could be resolved by artificial tear application in less than 1 day.

^{21}The frequency of occurrence did not depend on the stimulus intensity. These findings are in line with the results from previous trials.

^{20}

^{,}

^{22}In the EST2 trial, artificial tears were not used in asymptomatic cases. However, in regular therapeutic practice, it has since been recommended to use artificial tears immediately before and during TcES treatment to reduce the occurrence of dry eye symptoms.

**A. Stett**, Okuvision GmbH (E);

**A. Schatz**, None;

**F. Gekeler**, None;

**J. Franklin**, Okuvision GmbH (F)

*Prog Retin Eye Res*. 2018; 66: 157–186. [CrossRef] [PubMed]

*Eye*. 2017; 31: 741–748. [CrossRef] [PubMed]

*Adv Biosci*. 1987; 62: 29–58.

*Clin Vis Sci*. 1990; 5: 1–26.

*Invest Ophthalmol Vis Sci*. 2020; 61: 56. [CrossRef] [PubMed]

*Ophthalmology*. 1996; 103: 398–405. [CrossRef] [PubMed]

*Ophthalmology*. 1997; 104: 460–465. [CrossRef] [PubMed]

*Neural Regen Res*. 2021; 16: 1751–1759. [PubMed]

*Cell Death Dis*. 2020; 11: 793. [CrossRef] [PubMed]

*Frontiers in Neuroscience*. 2020; 14: 570909. [CrossRef] [PubMed]

*Clin Exp Ophthalmol*. 2021; 49: 418–429. [CrossRef] [PubMed]

*Mol Ther*. 2021; 29: 442–463. [CrossRef] [PubMed]

*Asia Pac J Ophthalmol (Phila)*. 2020; 9: 159–179. [CrossRef] [PubMed]

*Clin Exp Ophthalmol*. 2021; 49: 270–288. [CrossRef] [PubMed]

*Prog Retin Eye Res*. 2018; 65: 50–76. [CrossRef] [PubMed]

*Int J Ophthalmol*. 2016; 9: 914–919. [PubMed]

*Am J Pathol*. 2016; 186: 2783–2797. [CrossRef] [PubMed]

*Exp Rev Ophthalmol*. 2016; 11: 325–327. [CrossRef]

*Graefes Arch Clin Exp Ophthalmol*. 2015; 253: 171–176. [CrossRef] [PubMed]

*Ophthalmic Res*. 2020; 63: 234–243. [CrossRef] [PubMed]

*Invest Ophthalmol Vis Sci*. 2017; 58: 257–269. [CrossRef] [PubMed]

*J Ocul Pharmacol Ther*. 2020; 36: 609–617. [CrossRef] [PubMed]

*Graefes Arch Clin Exp Ophthalmol*. 2020; 258: 79–87. [CrossRef] [PubMed]

*Neuromodulation*. 2021; 24(6): 992–1002. [CrossRef] [PubMed]

*Cells*. 2020; 9: 18. [CrossRef]

*Exp Neurol*. 2009; 219: 439–452. [CrossRef] [PubMed]

*Exp Eye Res*. 2010; 90: 285–291. [CrossRef] [PubMed]

*Invest Ophthalmol Vis Sci*. 2020; 61: 5. [CrossRef] [PubMed]

*Invest Ophthalmol Vis Sci*. 2007; 48: 4725–4732. [CrossRef] [PubMed]

*Acta Ophthalmol*. 2018; 96: e366–e376. [CrossRef] [PubMed]

*Clin Ophthalmol*. 2010; 4: 1441–1446. [PubMed]

*Graefes Arch Clin Exp Ophthalmol*. 2018; 256: 299–306. [CrossRef] [PubMed]

*Invest Ophthalmol Vis Sci*. 2011; 52: 4485–4496. [CrossRef] [PubMed]

*Invest Ophthalmol Vis Sci*. 2012; 53: 7440–7448. [CrossRef] [PubMed]

*Lancet*. 1986; 327: 307–310. [CrossRef]

*Transl Vis Sci Technol*. 2019; 8: 36. [CrossRef] [PubMed]

*Invest Ophthalmol Vis Sci*. 2011; 52: 8042–8046. [CrossRef] [PubMed]

*Invest Ophthalmol Vis Sci*. 2006; 47: 4966–4974. [CrossRef] [PubMed]

*Restor Neurol Neurosci*. 2020; 38: 239–250. [PubMed]