Neuroprotection with rasagiline in patients with macula-off retinal detachment: A randomized controlled pilot study

We aimed to evaluate the neuroprotective efficacy of rasagiline in pseudophakic patients who had surgery for macula-off rhegmatogenous retinal detachment (RRD). This was a 6-month, prospective, randomized, double-blind, placebo-controlled pilot study. Patients presenting with acute macula-off RRD were recruited and randomized 1:1 to receive rasagiline 1 mg/day or placebo for 7 days. Best-corrected visual acuity (BCVA) and optical coherence tomography were acquired 1 day before as well as 2 days, 3 weeks, 3 months and 6 months after surgery. We screened 26 patients with RRD whereof 23 were eventually included and randomized. The primary outcome was final BCVA. Secondary outcomes included central retinal thickness (CRT) and adverse events (AE). We evaluated photoreceptor cells (prc) recovery through morphological measurements. The baseline characteristics were comparable between groups. BCVA significantly improved in both groups (letters gained: rasagiline 61.5 ± 18.1 vs placebo 55.3 ± 29.2, p = 0.56), but no significant inter-group difference was found at any visit. CRT was stable 3 weeks after surgery onwards, with no inter-group difference. No treatment-emergent AE occurred. Significant prc restoration was observed from 3 weeks to 6 months after surgery, without inter-group difference at either visit. Ellipsoid zone integrity (β = 0.517, p = 0.008) and foveal bulge (β = 0.387, p = 0.038) were significant predictors of good final BCVA. In conclusion, perioperative oral treatment with rasagiline 1 mg/day for 7 days did not show significant benefits on visual or anatomical outcomes in macula-off RRD patients.


Randomization and masking. Randomization took place after informed consent and confirmation that
inclusion criteria were met. The trial was double-blinded. The study medication, i.e., placebo or rasagiline (Teva Pharmaceutical Industries Ltd, Netanya, Israel) was prepared by the Inselspital hospital pharmacy in identical pills and stored in neutral vials.
Procedures. Study participants received perioperative oral treatment with either rasagiline (1 mg) or placebo once daily for 7 days, which was initiated at the time of hospital admission (baseline visit), usually one day before planned surgery the next day. Standard 3-port pars plana vitrectomy was performed with 20-or 23-gauge instruments, and sometimes an encircling band was added for external support. Hence, the specific choice of the detailed surgical procedure was at the discretion of the operating surgeons (S.W., M.S.Z., M.N.M., A.E.). Four follow-up visits took place at 2 days, 3 weeks (±5 days), 3 months (±10 days) and 6 months (±10 days) after the surgery. At each visit, a thorough ophthalmic exam was completed, including best-corrected visual acuity (BCVA) assessed with the Early Treatment Diabetic Retinopathy Study (ETDRS) chart, slit-lamp Goldman applanation tonometry, pupil dilation, color fundus photographs including autofluorescence, and spectral-domain Optical Coherence Tomography (SD-OCT; Spectralis, Heidelberg Engineering, Germany). Investigators questioned patients about medication change or intercurrent adverse events (AE) and instructed patients to report any such future events.
Outcomes. The primary outcome was BCVA 6 months after surgical repair. Secondary efficacy outcomes were central retinal thickness (CRT) at month 6, measured in the ETDRS grid's central subfield. Safety endpoints included the type and frequency of reported AEs and serious AEs. Additionally, thickness of four different layers at week 3 and month 6 (visits 2 and 4) was measured using a commercial software (Orion, Voxeleron LLC, San Francisco, USA): ganglion cell complex (GCC), inner nuclear layer-outer plexiform layer complex (INL-OPL), outer nuclear layer (ONL) and external limiting membrane (ELM)-RPE. To evaluate prc recovery, we graded the integrity of the outer retinal bands, namely ELM, ellipsoid zone (EZ) and cone interdigitation zone (CIZ) 12 . The integrity was graded at the fovea in a 1-mm-diameter area on a 5-point scale as follows: (1) line not visible; (2) line disruption >500 μm; (3) line disruption >200 μm, but <500 μm; (4) line disruption <200 μm; (5) continuous line. Lastly, we graded the presence of the foveal bulge at the end of the study, an indicator of good foveal microstructure. A foveal bulge was defined as an EZ-RPE thickness at the central fovea >10 μm greater than the average EZ-RPE thickness at 250 μm temporal and nasal to the central fovea 13 . Statistical analysis. Surgical success and failure rates were calculated. The failure rate was expressed in two ways 14 . The primary failure rate was the percentage of eyes that had a recurrent detachment or a complication after the initial procedure requiring an additional surgery but no silicone oil. There was a second group of patients who developed severe proliferative vitreoretinopathy (PVR) that was managed with silicone oil tamponade left in place for at least six months. These eyes had silicone oil remaining at the end of the study and their final visual outcomes pending. These eyes were not included in outcome analysis.
Analyses were performed on three sets. The safety set consisted of all randomized participants who took the perioperative oral medication for 7 days and underwent at least 1 safety assessment after enrolment. The full analysis set (FAS) consisted of all participants who completed the trial and did not develop PVR managed with silicone oil tamponade and had reliable visual acuity data at study end 6 months after initial repair. The uneventful set consisted of all randomized participants completing the study without the need for any additional intraocular surgery to address complications (Fig. 1).
Continuous data were summarized using mean and standard deviation. We used Student's t-test to test the null hypothesis of no difference between the means of the two randomized groups. Categorical data were summarized using percentages, and the groups were compared using Fisher's exact test. We constructed linear regression models to evaluate the effect of variables on study outcomes. Statistical testing was two-sided with a 5% significance level.

Results
Baseline characteristics. A total of 26 participants signed informed consent and entered screening. Three participants did not meet inclusion criteria. Hence, the safety set included 23 patients (rasagiline, n = 12; placebo, n = 11).
Baseline characteristics were well balanced for demographic parameters, BCVA, and duration of symptoms before presentation group ( Table 1). The average age was 68.0 ± 10.5 years, 78% were male. There was no significant difference between groups for any of the parameters (p > 0.05).
Surgical success rate. SF6 gas tamponade was used in all 23 primary surgeries. In total, 16 participants were successfully repaired by single surgery, achieving a single operation success rate (SOSR) of 70% (rasagiline, 67%; placebo, 73%). Five recurrent detachments (rasagiline, n = 3; placebo, n = 2), one full thickness macular hole (FTMH; rasagiline group), and one severe epiretinal membrane (ERM; placebo group) occurred during the study period, leading to a primary failure rate of 30%. Two of the re-detachments, one in each arm, received two additional surgeries and were excluded from the FAS due to remaining silicone oil and severe PVR at the end of study. Thus, the PVR rate was 9%. The other cases were managed by one additional surgery. Therefore, 21 participants (91%) completed the trial and were included in the FAS (rasagiline, n = 11; placebo, n = 10). 16 participants, 8 in each arm had an uncomplicated postoperative course, forming the uneventful set. Details of the additional surgeries during the study follow-up are listed in Table 2.
Visual outcome. BCVA significantly improved in both groups after surgery (letters gained in 6 months, FAS: 61.5 ± 18.1 in rasagiline and 55.3 ± 29.2 in placebo group; Uneventful set: 68.8 ± 9.1 in rasagiline and 50.8 ± 30.6 in placebo group), but there was no significant difference of final BCVA between groups (Rasagiline 74.9 ± 9.1 letters vs. placebo 72.7 ± 19.5 letters, Table 3). The most significant improvement in BCVA was noted during the first three weeks after the surgery (FAS: 43 letters in rasagiline and 36 letters in placebo group; Uneventful set: 51 letters in rasagiline and 39 letters in placebo group), with no significant difference between groups. There was no difference at any visit between the rasagiline and placebo groups in BCVA (Fig. 2). Final BCVA of the two PVR cases with silicone oil tamponade in situ were finger counting and 20 letters, respectively. The other eyes requiring additional surgery reached a mean final BCVA of 63.8 letters (range: 45-81), and 75% of the eyes had 60 ETDRS letters or better BCVA. www.nature.com/scientificreports www.nature.com/scientificreports/ Anatomical outcome. CRT was stable from week 3 to month 6, and there was no difference between the rasagiline and placebo groups for the CRT at either visit (FAS: p = 0.6 at week 3, p = 0.9 at month 6; Uneventful set: p = 0.2 at week 3, p = 0.9 at month 6, Fig. 3).
Thicknesses of GCC, INL-OPL and ONL were stable from week 3 to month 6, and there was no difference between the rasagiline and placebo groups for CRT at either visit (regardless whether in the FAS or in the uneventful set (Fig. 4)).
ELM-RPE thickness significantly increased from week 3 to month 6, but there was no difference between the rasagiline and placebo groups at either visit, regardless of fovea (ETDRS central 1 mm subfield) or macula (ETDRS 6 mm area) (Fig. 5). Table 4 summarizes the grading result related to prc recovery. Integrity of ELM and EZ bands significantly improved in both groups without inter-group difference. The placebo group had a slightly better CIZ band integrity at week 3 (rank: 1.0 vs 1.1, p = 0.029). The foveal bulge was seen in 6 eyes at month 6

Safety.
A total of 8 participants (35%) reported side effects (Table 6). No serious AE was reported. There was no difference in the number of AE or side effects between the groups. Thus, rasagiline was well tolerated in patients with RRD.

Discussion
The surgical approach is currently the only means to repair RRDs and preserve or restore vision. However, visual outcomes vary, even in patients who have had successful primary surgical repair. Prc death supposedly is the ultimate cause of vision loss after a RRD. Therefore, pharmacotherapy reducing or slowing degeneration of prcs would be a potential add-on therapy enhancing functional outcomes. Several studies targeting these pathways provide evidences that a neuroprotective strategy might potentially result in prc rescue after RRD 3,4 .
Selegiline [L-deprenyl] is an earlier selective irreversible MAO-B inhibitor used in Parkinson's disease, and showed some neuroprotective effect in retinal cells, such as lower apoptosis and higher viability in cultured RPE that led to increased survival of retinal ganglion cells after retinal damage in rats 15 . Later, rasagiline replaced selegiline in the clinic because of 15-fold higher potency. Studies demonstrated that rasagiline influences multiple pathways related to prc death, including Caspase-3 activation and Bax/Bcl-2 expression that are involved in apoptosis, while Beclin-1expression is involved in autophagy, and Caspase-1 activation and glial fibrillary acidic protein expression are involved in inflammation 7,8 .  www.nature.com/scientificreports www.nature.com/scientificreports/ Despite these promising results in the animal model, no pharmacotherapies have been clinically tested yet. We, therefore, carried out this clinical pilot trial to evaluate the efficacy of pharmacotherapy as an adjunct to surgical RRD repair. This study prospectively assessed rasagiline 1 mg/day treatment for 7 days in pseudophakic macula-off RRDs. We found no measurable effect of rasagiline compared to placebo. Our findings thus fell short of identifying a clinical neuroprotective effect of rasagiline at this dose on patients who had surgical repair within 96 hours (maximal time interval between vision loss and presentation plus waiting time to start of operation for surgical repair) of onset of symptoms.
Failure of rasagiline to show any neuroprotective effect in our study could be due to several reasons: First, the dose tested or the treatment duration might have been insufficient. We decided to apply this conservative regimen in the first clinical trial of rasagiline in retinal disease based on two facts: (1) the standard 1 mg dose given to Parkinson's disease patients is well tolerated; (2) 7-day rasagiline administration was regarded as minimal duration for treatment 16 . However, with regard to the treatment dose, the two preclinical studies showing neuroprotection of rasagiline on prc used 2 μg/g BW/d in Prph2/rds mice and 15 μg/g BW/d in rd10 mice 7,8 . This would be equivalent to a daily dose of 10 mg and 75 mg for a 60-kg human. These doses are much higher than commonly used in clinic, and the second dosage would even exceed the reported tolerability range in human 17 . With respect to treatment duration, experience from the treatment of Parkinson's disease suggests that rasagiline is limited in efficacy, but has a consistent benefit with chronic administration 18 . Hence, 7 days of treatment might not have been long enough.
Moreover, the follow-up intervals might have been unsuitably defined to evaluate pcr rescue. Differences might have been present around the 3 weeks time point. Prc death in human occurred as early as 12 h, peaked at around 2-3 days and dropped to low levels by 7 days after RRD 19 . Meanwhile, rasagiline achieves 80% MAO-B inhibition after a single standard dose and increases to 99% after 7-day administration. After treatment withdrawal, the inhibitive effects decline time-dependently and lasts for utmost 3 weeks 20 . Taken together, more benefit from neuroprotection might result from faster availability. In addition, Eigeldinger-Berthou et al. stated in their preclinical study that prc death was only delayed but not absent after rasagiline treatment 7 . In our study, prc evaluation through OCT images was only available 3 weeks after surgery when the SF6 gas had disappeared from the macular area. Therefore, our failure to find difference between groups from 3 weeks to 6 months after surgery might be due to bad timing, imposed by technical constraints. Novel in vivo evaluation would be helpful to reveal the prc change in the early postoperative phase. Ahn et al. demonstrated that swept-source OCT performed significantly better than SD-OCT in macular visualization in gas-filled eyes at days 1 and 3 after surgery 21 . However, their research only focused on the visualization of macular configuration, but not of the outer retinal bands. Fluorescence Lifetime imaging ophthalmoscopy (FLIO) might be a promising future option. FLIO measures lifetimes of endogenous retinal fluorophores after excitation using a picosecond pulsed blue laser light 22 . Lipofuscin is a major endogenous retinal fluorophore, and accumulates after RPE cells aging, which originates from incomplete degradation of prc outer segment 23 . Therefore, FLIO might be valuable to visualize prc death after RRD. No study has covered this topic and further research is needed.
Furthermore, other pathways might outweigh the beneficial effects attributed to rasagiline (i.e., MAO-B inhibition, effects on oxidative stress, mitochondrial dysfunction and antiapoptotic properties) in the context of RRD. In addition to prc death, RRD also causes complex cellular remodeling of neuron synapses and Müller cells, which hamper visual function and prc regeneration, respectively 24 . Furthermore, prc survival and death pathways are trigged at the same time after detachment, when some mechanism works as a scroll bar pulling prc toward Figure 5. ELM-RPE thickness change from week 3 to month 6 of the participants in the FAS. ELM-RPE thickness significantly increased from week 3 to month 6 in both groups (p < 0.05). There was no difference between the rasagiline and placebo groups at either visit (P > 0.05). *p < 0.05, **p < 0.01.

Rasagiline (n = 11)
Placebo (n = 10)  www.nature.com/scientificreports www.nature.com/scientificreports/ survival or death, such as calpain activation. Such activation was found to be a key step in triggering prc to shift from survival to death, which peaks 7 days after detachment 25 .
The duration of macular detachment influences visual outcome greatly, because proapoptotic factors and inflammation cytokines become significantly upregulated with increasing duration of separation. Such upregulation leads to more disruption of the EZ and worse BCVA after surgery 26,27 . In this study, we only included pseudophakic macula-off RRD patients with detachment duration less than 72 hours to avoid time bias and influences of lens opacity. This could explain why our visual outcome was better than reported in other reports for patients with longer detachment duration. In our study, the mean BCVAs were 74.9 and 72.7 letters in the rasagiline and placebo group, respectively, and more than 76% patients achieved a BCVA of 65 letters or better at month 6. In a recent study, the final mean BCVA was reported to be merely 62 letters and only 61.1% of the eyes achieved 65 letters or better after a 24-month recovery after vitrectomy. Although the initial reattachment rate of 94.5% was high, their pseudophakic macula-off RRD participants had a long detachment duration of 11 days (range, 3-30 days) 28 . Of note, our regression analysis showed that there was no influence on final BCVA whether time before presentation was less than 24 hours or between 24 to 72 hours (regression coefficient: −0.320, p = 0.158). With respect to visual acuity, previous work has shown that best outcomes were achieved in patients with macular detachment of less than 7 days duration. There was no difference between patients with macular detachment of 0-3 days duration and those with central detachment of 4-7 days 1 . Our analysis suggests that there was no difference between patients with less that 24 h macular detachment and those with 2-3 days of macular involvement.
The height of macular detachment was previously reported to influence visual recovery. Lower height of macular detachment correlates with better visual outcome 29,30 . This current study observed a consistent negative association between the macular detachment height and the final BCVA, but this trend was not statistically significant (regression coefficient: −0.466, p = 0.069).
We achieved a similar SOSR of 70% in pseudophakic macula-off RRD as the rate of 63% reported in an earlier study 31 . However, the primary failure rate was 26% in our study, and was higher than the corresponding rate of 10.5% reported in previous reports on pseudophakic RRD 14 . In the current study, two PVR cases result in a PVR rate of 9%, which is rather high but still within the normal range of PVR incidence after RRD surgery cases (5-10%) 32 . Enders et al. studied 2457 patients with RRD and found macular detachment was not only associated with worse visual outcome but also with a higher recurrent detachment rate 33 . Therefore, we speculate that macula-off status might have contributed to our relatively high failure rate. Recurrent detachment and interventions to treat recurrent detachment have in general a worse final visual outcome than primarily successful cases. In our study, 33% of the recurrent detachment cases reached final BCVA of 70 letters or better when the detachment was managed by only one additional surgery. However, no patient achieved this level of visual acuity when  www.nature.com/scientificreports www.nature.com/scientificreports/ more interventions were needed. This finding is in accordance with the data from Enders et al. that the share of good BCVA decreased from 16.9% in cases with one recurrence to 7.6% in cases with 2 recurrences and to 0% for patients that had 3 interventions 33 .
We observed significantly increased ELM-RPE thickness during our follow-up. To reveal more details of prc restoration, we graded the integrity of the ELM, EZ and CIZ, and noticed significant improvement of integrity of ELM and EZ in both the rasagiline and the placebo groups. Our finding is in line with a previous report where the prc layer demonstrated a time-dependent thickness increase after successful surgery 34 . This increase might indicate prc regeneration after detachment-induced disruption of prc outer segments, which can be visualized on OCT as dropouts of the EZ layer 35 . Taking a closer look at ELM-RPE thickness, the inner segment, ELM-EZ thickness usually achieves fully-restoration at month 6. The outer segment, EZ-RPE thickness, however, stayed thinner than that in the healthy fellow eyes after 12-month recovery 36 . The recovery of EZ-RPE thickness is the basis of the reconstitution of the foveal bulge, a sign of favourable foveal configuration 36 . We observed a foveal bulge in 29% of the eyes at month 6, in line with the study from Hasegawa et al., who reported a bulge in 28.6% of patients 5-8 months after a macula-off RRD 13 . Increased EZ-RPE thickness, presence of a foveal bulge and better integrity of EZ band indicate good regeneration of prc outer segments and predict better visual outcome 36,37 . Our multiple regression model showed the same result, namely that EZ integrity and presence of foveal bulge are factors predicting better BCVA.
There are many limitations to the present study. First, this clinical trial tested a very conservative regimen. This might be one major reason for our non-significant results for efficacy of rasagiline. Secondly, the sample size is relatively small, which results in low statistical power. Since this was a pilot trial, sample size calculations were not rigorous but rather pragmatic. Whitehead et al. suggested a pilot trial sample size per treatment arm of 20 subjects for small standardized differences (0.1-0.3) to aim for a power of 80% in a subsequent main trial 38 . In the original pilot trial protocol, we planned to enroll 20 patients in each arm. However, the trial was stopped early due to slow enrollment, resulting in a sample size of around 10 patients in each arm, which should still be appropriate to detect medium standardized differences (0.3-0.7) as suggested in above paper by Whitehead et al.. Thirdly, the maximum time of foveal detachment in this study was 96 hours, and we didn't see any difference in the primary outcome. However, this doesn't mean that one might see a neuroprotective effect in patients with longer waiting periods between foveal affection and surgical repair. Since the detachment duration is relevant for the level of prc death in RRD, the therapeutic/anti-apoptotic effect of being re-attached may have exceeded that of rasagiline in our setting. Hence, the design of the study might just have been inappropriate in this regard. Last but not least, the integrity of the outer retinal bands and the presence of a foveal bulge were manually determined, with possible bias. Despite these limitations, this study provides some prospective information on the effects of rasagiline as pharmacologic adjunct treatment for surgical RRD repair.

conclusion
Perioperative treatment with rasagiline 1 mg/day for 7 days did not show significant benefits for either visual or anatomical outcomes 6 months after surgery surgical repair of a macula-off RRD. EZ integrity and the presence of a foveal bulge were significant predictors of good final BCVA.

Data availability
The dataset generated and analysed during the current study is available from the corresponding author on reasonable request.