The Role of Dopaminergic Agonists in Geographic Atrophy: Converging Evidence From Clinical and Experimental Data

Abstract

Purpose

Levodopa (L-DOPA) is an emerging drug repurposing candidate for age-related macular degeneration (AMD). Basic and clinical studies identify a potential role for L-DOPA to prevent and treat neovascular AMD. However, only one prior study has been performed in geographic atrophy (GA).

Methods

We performed a retrospective cohort analysis from the Sight Outcomes Research Collaborative database, including eyes with a diagnosis of early- or intermediate-stage AMD in 1 eye. We compared eyes exposed to any dopamine agonist (DA), dopamine receptor D2 agonist, and L-DOPA with eyes with no DA exposure using propensity score matching and survival analysis with multivariable Cox proportional hazard models with new GA development as our primary outcome. Mice were treated with intraperitoneal NaIO₃ to induce GA-like pathology. We compared L-DOPA with phosphate-buffered saline–treated controls in both pigmented and albino mice. We quantified retinal thickness by optical coherence tomography, and performed immunofluorescence to quantify choriocapillaris and retinal pigment epithelium density.

Results

Any DA, dopamine receptor D2, and L-DOPA exposure were not associated with new-onset GA. NaIO₃ treatment caused retinal thinning, choriocapillaris loss, and retinal pigment epithelium degeneration, which were not rescued by L-DOPA treatment in pigmented nor albino mice.

Conclusions

In both retrospective database analysis and preclinical models, DA had no effect upon GA pathology.

Translational Relevance

Further investigation of levodopa may be more appropriately directed toward neovascular age-related macular degeneration, rather than geographic atrophy, where preclinical, retrospective, epidemiologic, and prospective data suggest a benefit.

Introduction

Geographic atrophy (GA), the advanced atrophic form of age-related macular degeneration (AMD), is one of the leading causes of irreversible central vision loss in the aging population and represents a significant and growing public health burden.¹ Although anti-vascular endothelial growth factor (VEGF) therapies have transformed the management of neovascular AMD, comparable disease-modifying treatments for GA have been explored recently and offer modest benefits. Complement inhibitors such as pegcetacoplan² and avacincaptad pegol³ have demonstrated the ability to slow lesion growth; however, they do not restore lost vision, show variable efficacy among patients, and require chronic administration. Due to these limitations, alternative GA treatments are an unmet clinical need. Drug repurposing—identifying new indications for previously approved and safe medications—is an attractive option for GA. Multiple database studies have identified several drug repurposing candidates in AMD.⁴

Levodopa (L-DOPA), a well-established dopamine precursor widely used in the treatment of Parkinson's disease, has recently emerged as a molecule of growing interest in retinal research due to its potential protective effects in AMD. Beyond its canonical role in replenishing dopamine, L-DOPA may exert neuroprotective and cytoprotective effects in the retina through modulation of dopamine signaling pathways, activation of G protein–coupled receptor cascades, and maintenance of retinal pigment epithelium (RPE) health and function.⁵ Increasing evidence from database studies suggest that L-DOPA exposure is associated with a reduced risk and delayed onset of AMD, highlighting a possible disease-modifying role.⁶ These results were replicated for neovascular AMD in three distinct database studies.^7–9 However, only one database study has investigated L-DOPA in GA, finding a protective effect.¹⁰ Mechanistically, dopamine signaling has been implicated in reducing oxidative stress,¹¹ decreasing exosome release,¹² and dilating the choroidal vasculature,¹³ all of which could reduce GA development and/or progression. Based on these data, additional database and animal model studies are needed to investigate the role of L-DOPA in GA.

We performed a follow up database study in the Sight Outcomes Research Collaborative (SOURCE) database mined from academic medical institutions. This database is intentionally nonoverlapping with the Vestrum database from private practice retina specialists that we previously investigated.¹⁰ Additionally, we performed animal model studies using the sodium iodate (NaIO₃) mouse model of GA to further clarify the role of L-DOPA in GA.

Methods

SOURCE Analysis

A retrospective analysis was conducted of patients in the SOURCE Ophthalmology Data Repository. SOURCE contains electronic health record data of all patients receiving any eye care at academic health systems across the United States participating in this consortium from the time each site went live on the electronic health record until the present. SOURCE captures information on patient demographics, diagnoses identified based on International Classification of Diseases billing codes, and eye examination findings from every clinic visit, along with data on medications, laser and surgical interventions. The data in SOURCE are completely deidentified. However, privacy-preserving software (Datavant Inc., Phoenix, AZ) permits researchers to follow patients longitudinally over time and across SOURCE institutions, while still protecting patients’ identities. Researchers have published other studies using SOURCE data. The study adhered to the tenets outlined in the Declaration of Helsinki. The Northwestern University Institutional Board of Review determined this deidentified retrospective study was exempt from review (STU00223846), and informed consent was not applicable to this study given the dataset was deidentified.

Inclusion criteria comprised patients with a diagnosis of non-neovascular intermediate stage AMD in at least one eye, via International Classification of Diseases, 10th edition, coding. Intermediate stage AMD was included because this population is associated with a higher risk for GA. If the patient had non-neovascular early- or intermediate-stage AMD in the fellow eye, it was also included in the study. The fellow eye was included because both eyes are exposed to dopamine agonist (DA) effects. The cohort was divided into three groups for analysis. Study 1 included eyes exposed to any DA (dopamine receptor D2 [DRD2] agonist or L-DOPA), which were compared with eyes with no history of DA exposure. Study 2 included eyes exposed to DRD2 agonist (pramipexole, ropinirole, apomorphine, rotigotine, bromocriptine, piribedil, lisuride, cabergoline, and/or pergolide), which were compared with eyes with no history of DA exposure. Study 3 included eyes exposed to L-DOPA, which were compared with eyes with no history of DA exposure. Exclusion criteria included eyes newly exposed to L-DOPA or a DRD2 agonist after a non-neovascular AMD diagnosis or when a variable of interest was unknown.

Information regarding age, sex, race (non-White vs. White), tobacco exposure (never vs. former vs. active), AREDS2 use, AMD stage (early vs. intermediate), AMD stage in fellow eye (non vs. GA vs. neovascular AMD), follow-up duration, and conversion to GA was collected. Follow-up duration was calculated based on the difference between date of non-neovascular AMD diagnosis and date of last follow-up with an ophthalmic provider in the SOURCE dataset. To account for confounders and generate balanced cohorts for analysis, propensity score matching (PSM) was conducted via the “MatchIt” package in R to perform nearest-neighbor matching between DA-exposed and control eyes. PSM was conducted using the following covariates: age, sex, race, tobacco exposure, AREDS2 use, AMD stage, and follow-up duration. Results of PSM were validated via Wilcoxon signed-rank or χ² analysis.

After PSM, a survival analysis was performed to investigate the effect of DA exposure on time to GA conversion within 5 years of non-neovascular AMD diagnosis (Table 1). Multivariable Cox proportional hazard models were created to generate adjusted hazard ratios (HRs) and 95% confidence intervals (CIs), with covariates in the models including: age, sex, race, tobacco exposure, AREDS2 use, AMD stage, neovascular AMD or GA in the fellow eye, and DA exposure. To account for within-subject correlation due to possible inclusion of two eyes from the same patient in the Cox regression analysis, patients were clustered via use of the cluster() term. These steps regarding PSM, survival analysis, and Cox regression analysis were replicated for the three studies as outlined (Table 2). Statistical significance was set at a P value of less than 0.05. Statistical analysis was performed with R (version 4.4.0).

Table 1.

Demographic and Clinical Characteristics of Propensity Score–Matched Eyes Exposed to any DA vs. no DA Exposure, DRD2 Agonist vs. no DA Exposure, and L-DOPA vs. no DA Exposure

	Any DA vs. No DA			DRD2 Agonist vs. No DA			L-DOPA vs. No DA
Variable	Any DA (n = 1785)	No DA (n = 5355)	P Value	DRD2 Agonist (n = 1046)	No DA (n = 3138)	P Value	L-DOPA (n = 571)	No DA (n = 1713)	P Value
Age, years	78.0 ± 7.5	78.0 ± 7.5	0.93	77.4 ± 7.8	77.2 ± 7.7	0.36	79.5 ± 6.6	79.6 ± 6.8	0.85
Sex
Male	683 (38.3)	1974 (36.9)	0.30	271 (25.9)	777 (24.8)	0.48	327 (57.3)	985 (57.5)	0.96
Female	1102 (61.7)	3381 (63.1)		775 (74.1)	2361 (75.2)		244 (42.7)	728 (42.5)
Race
White	1656 (92.8)	4982 (93.0)	0.75	997 (95.3)	2978 (94.9)	0.65	501 (87.7)	1528 (89.2)	0.38
Non-White	129 (7.2)	373 (7.0)		49 (4.7)	160 (5.1)		70 (12.3)	185 (10.8)
Tobacco exposure
Never	967 (54.2)	2910 (54.3)	0.99	579 (55.4)	1786 (56.9)	0.64	297 (52.0)	869 (50.7)	0.87
Former	757 (42.4)	2262 (42.2)		415 (39.7)	1194 (38.0)		267 (46.8)	822 (48.0)
Active	61 (3.4)	183 (3.4)		52 (5.0)	158 (5.0)		7 (1.2)	22 (1.3)
AREDS2 use
Yes	364 (20.4)	1072 (20.0)	0.76	211 (20.2)	657 (20.9)	0.63	104 (18.2)	295 (17.2)	0.63
No	1421 (79.6)	4283 (80.0)		835 (79.8)	2481 (79.1)		467 (81.8)	1418 (82.8)
AMD stage
Early	757 (42.4)	2278 (42.5)	0.94	483 (46.2)	1433 (45.7)	0.80	216 (37.8)	670 (39.1)	0.62
Intermediate	1028 (57.6)	3077 (57.5)		563 (53.8)	1705 (54.3)		355 (62.2)	1043 (60.9)
Follow-up duration (years)	1.1 ± 1.3	1.1 ± 1.3	0.062	1.1 ± 1.4	1.1 ± 1.4	0.28	1.0 ± 1.2	1.0 ± 1.2	0.81
Lost to follow-up at year 1	1006 (56.4)	3043 (56.8)	0.75	598 (57.2)	1807 (57.6)	0.84	328 (57.4)	990 (57.8)	0.92
Lost to follow-up at year 2	281 (15.7)	760 (14.2)	0.21	138 (13.2)	385 (12.3)	0.46	112 (19.6)	284 (16.6)	0.26
Lost to follow-up at year 3	150 (8.4)	517 (9.7)	0.092	100 (9.6)	294 (9.4)	0.60	41 (7.2)	130 (7.6)	0.67
Lost to follow-up at year 4	131 (7.3)	375 (7.0)	0.69	69 (6.6)	235 (7.5)	0.85	45 (7.9)	103 (6.0)	0.055
Lost to follow-up at year 5	60 (3.4)	186 (3.5)	0.73	40 (3.8)	119 (3.8)	0.28	10 (1.8)	35 (2.0)	0.88
GA
Incidence	55 (3.1)	144 (2.7)	0.43	35 (3.3)	102 (3.3)	0.96	13 (2.3)	54 (3.2)	0.35

AREDS2, Age-Related Eye Disease Study 2.

Values are mean ± standard deviation or number (%).

Table 2.

Multivariable Cox Regression Analysis

	Any DA vs. No DA		DRD2 Agonist vs. No DA		L-DOPA vs. No DA
Variable	Adjusted HR (95% CI)	P Value	Adjusted HR (95% CI)	P Value	Adjusted HR (95% CI)	P Value
Age, years
<70 (reference)	—	—	—	—	—	—
70–74	0.65 (0.29–1.46)	0.3	0.53 (0.18–1.54)	0.2	1.81 (0.20–16.3)	0.6
75–79	1.54 (0.80–2.97)	0.2	1.76 (0.80–3.88)	0.2	4.51 (0.54–37.5)	0.2
80–84	2.13 (1.15–3.96)	0.017	2.19 (1.03–4.66)	0.041	6.34 (0.82–48.9)	0.076
≥85	2.16 (1.01–4.58)	0.046	3.14 (1.28–7.69)	0.012	5.24 (0.62–44.3)	0.13
Sex
Male (reference)	—	—	—	—	—	—
Female	1.27 (0.87–1.84)	0.2	1.40 (0.82–2.40)	0.2	1.36 (0.71–2.60)	0.4
Race
White (reference)	—	—	—	—	—	—
Non-White	0.67 (0.28–1.61)	0.4	1.70 (0.64–4.50)	0.3	1.02 (0.38–2.73)	>0.9
Tobacco exposure
Never (reference)	—	—	—	—	—	—
Former	1.19 (0.84–1.69)	0.3	1.03 (0.66–1.61)	0.9	0.93 (0.48–1.78)	0.8
Active	1.45 (0.48–4.36)	0.5	1.64 (0.44–6.08)	0.5	—	—
AREDS2 use
No (reference)	—	—	—	—	—	—
Yes	1.36 (0.95–1.96)	0.092	1.71 (1.08–2.72)	0.022	1.14 (0.57–2.29)	0.7
AMD stage
Early (reference)	—	—	—	—	—	—
Intermediate	3.29 (2.04–5.28)	<0.001	2.91 (1.69–4.99)	<0.001	5.46 (1.84–16.2)	0.002
Stage in fellow eye
None (reference)	—	—	—	—	—	—
GA	5.42 (2.23–13.2)	<0.001	4.62 (1.21–17.7)	0.025	—*	—*
Neovascular AMD	1.99 (0.90–4.41)	0.091	1.49 (0.55–3.99)	0.4	1.28 (0.16–10.3)	0.8
DA exposure
No DA (reference)	—	—	—	—	—	—
DA Type	1.11 (0.76–1.63)	0.6	1.00 (0.62–1.61)	>0.9	0.74 (0.36–1.55)	0.4

^*These values were empty due to insufficient number of events in the subgroup for analysis.

Animals

All experiments included both male and female mice. All data were investigated for sex-specific effects, and none were found. Wild-type C57BL/6J (#000664) and C57BL/6J Tyr^c-2J/c-2J (#000058, henceforth referred to as Tyr^−/− or albino) mice were obtained from Jackson Laboratories (Bar Harbor, ME). Animals were housed in a pathogen-free barrier facility at the Center for Comparative Medicine, Northwestern University (Chicago, IL). Genotyping was performed by Transnetyx (Cordova, TN) to confirm the absence of the rd8 mutation (Crb1⁻). All procedures adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the Northwestern University Institutional Animal Care and Use Committee. Mice used in the experiments were 9 to 12 weeks of age.

NaIO₃–Induced GA Animal Model

To establish the GA model, NaIO₃ (Sigma-Aldrich, St. Louis, MO) was dissolved in phosphate-buffered saline (PBS; C8267, Sigma-Aldrich) to a final concentration of 1% (w/v). Mice received a single intraperitoneal injection of NaIO₃ at a dose of 20 mg/kg (2 µL/g body weight).

L-DOPA Administration

L-DOPA (D1066; Spectrum Chemicals, New Brunswick, NJ) and benserazide hydrochloride (B7283-16; Sigma-Aldrich) were freshly prepared daily in PBS. Mice received daily intraperitoneal injections of L-DOPA (30 mg/kg) combined with benserazide hydrochloride (12 mg/kg) or PBS as a control. Treatment was initiated 2 days before NaIO₃ administration and continued for 6 days post injection.

Optical Coherence Tomography (OCT) Imaging

Mice were anaesthetized and administered analgesia as previously described.¹⁴ Ocular lubrication was maintained using 2.5% hypromellose (Vista Gonio, Scottsdale, AZ). Retinal imaging was performed using the Spectralis OCT2 system (Heidelberg Engineering, Heidelberg, Germany). High-resolution scans were acquired with a 55 × objective centered on the optic nerve head. Imaging was conducted at baseline and 1 week after NaIO₃ administration.

Total retinal thickness, outer nuclear layer (ONL) thickness, and ellipsoid zone (EZ) thickness were measured at six eccentricities along a line perpendicular to the horizontal axis, located approximately 300, 500, and 600 µm from the optic nerve head in each direction. Measurements were converted from arbitrary image units to micrometers using a scaling factor of 59.44 µm per unit. Total retinal thickness was defined as the distance from the inner limiting membrane to the RPE. ONL thickness was measured as the distance spanning the outer plexiform layer to the external limiting membrane, and EZ thickness was measured from the inner aspect of the EZ to the RPE. Measurements were averaged between eyes for each thickness and location.

Immunofluorescence Imaging

Mice were euthanized 7 days after NaIO₃ injection. Albino RPE and choroidal wholemounts were imaged as previously described.¹⁵ Eyes from wild-type, pigmented mice were split for independent analysis: one eye for unbleached RPE wholemount immunofluorescence imaging and one eye for bleached choroidal wholemount immunofluorescence imaging. Bleached choroidal wholemount immunofluorescence imaging was performed as previously described.¹⁶

For unbleached RPE wholemount immunofluorescence, eyes were enucleated and fixed in 4% paraformaldehyde (#15713-S; Electron Microscopy Sciences, Hatfield, PA) for 30 minutes at room temperature. Eyes were dissected to include the RPE–choroid–sclera complex, washed in TBS, fixed for an additional 30 minutes at room temperature, and blocked in TBS (#BP24711, Thermo Fisher Scientific, Waltham, MA) with 5% Donkey Serum (#S30, Sigma-Aldrich) + 2.5% bovine serum albumin (A2153, Sigma-Aldrich) + 0.5% Triton X-100 (X100; Sigma-Aldrich) overnight at 4°C. RPE wholemounts were washed twice in TBS for 30 minutes and incubated in primary (in blocking buffer) (Table 3) overnight at 4°C. RPE wholemounts were next washed five times in TBS with 0.5% Tween-20 (#00777; Amresco, Solon, OH) for 30 minutes at room temperature. Secondary incubation was performed overnight in blocking buffer at 4°C. RPE wholemounts were again washed five times in TBS with 0.5% Tween-20 for 30 minutes at room temperature. RPE wholemounts were mounted onto HistoBond microscope slides (#16004–406; VWR, Batavia, IL) using Immu-Mount (#9990402; Thermo Fisher, Carlsbad, CA). Fluorescence imaging was performed using a Nikon SoRa Spinning Disk Microscope with Nikon NIS-Elements software at the Northwestern University Center for Advanced Microscopy.

Table 3.

Immunofluorescence Antibodies

Target	Fluorophore	Manufacturer, Product #, Dilution	Use
Goat anti-podocalyxin	—	R&D Systems, AF1556,1:250	Primary
Rat anti-podocalyxin	—	R&D Systems, MAB1556,1:200	Primary
Actin-Phalloidin	Alexa Fluor 555	Invitrogen, A34055, 1:100	Secondary
Donkey anti-goat (H+L)	Alexa Fluor 488	Invitrogen, A11055, 1:500	Secondary
Donkey anti-rat (H + L)	Alexa Fluor 488	Invitrogen, A21208, 1:500	Secondary

Image Analysis

To ensure unbiased quantification, animal identities were masked for all image analyses. The choriocapillaris (CC) area was assessed using Fiji (ImageJ; NIH, Bethesda, MD) with the Labkit plugin. Segmentation of vasculature from background was performed using Labkit's automated classifier, retrained roughly 10 times to optimize reliability and accuracy. Next, binary segmentation masks were imported into Fiji for analysis, and the CC area was calculated with the ‘Analyze Particles’ function. The percent area coverage was determined by dividing the segmented area by the total image area × 100, providing a quantitative metric of vascular density.

RPE density was assessed by first cropping a 650 × 650-pixel region. Individual RPE cell boundaries were then manually traced in Fiji using the polygon selection tool. Next, cell outlines were added to the ROI Manager for area measurement, and RPE density was computed as the total number of traced cells divided by the sum of all measured cell areas. Quantification of the RPE size distribution was done by binning RPE cell area into predefined size categories. For each bin, the number of RPE cells was quantified, and the percentage of cells in each bin was computed by dividing the number of cells in the bin by the total number of RPE cells analyzed.

Statistics on Animal Studies

Normality of the data was assessed using the Shapiro–Wilk test. All results are presented as mean ± standard error of the mean. For comparisons between two groups with parametric data, unpaired t-tests or Welch's t-test were applied based upon the standard deviations of each group. Nonparametric data were analyzed using the Mann–Whitney test. Analyses between three groups were performed using one-way analysis of variance followed by Tukey's multiple comparisons test. Analyses of RPE cell size distributions and OCT retinal thickness across experimental conditions and time points were conducted using two-way analysis of variance with Tukey's (OCT) and Sidak's (RPE) multiple comparisons test. Sample sizes for each experiment are provided in the corresponding figure legends. Statistical significance was defined as a P value of less than 0.05.

Results

Any DA Exposure

After PSM, we identified 1785 eyes exposed to any DA and 5355 control eyes. The cohort was balanced across variables of interest, and the incidence of GA was similar between the two groups (Table 1). The probability of conversion to GA between DA exposed and control eyes is depicted in Figure 1A. The incidence of GA development was 2.7% in the no DA group and 3.1% in the any DA group over 5 years (Table 1). Older age (80–84, ≥85), intermediate AMD stage, and GA in fellow eye were risk factors for conversion to GA (Table 2). Exposure to any DA was not significantly associated (HR, 1.1; 95% CI, 0.76–1.63; P = 0.60) with conversion to GA (Fig. 1A; Table 2).

Figure 1.

Probability of progression to GA during DA. (A) Probability of conversion to GA in any DA exposed vs. control eyes. (B) Probability of conversion to GA in DRD2 agonist-exposed vs. control eyes. (C) Probability of conversion to GA in L-DOPA–exposed vs. control eyes.

DRD2 Agonist Exposure

After PSM, there were 1046 DRD2 agonist exposed and 3138 control eyes matched across variables of interest; the incidence of GA was similar between the two groups (Table 1). The probability of conversion to GA between DRD2 agonist exposed and control eyes is shown in Figure 1B. The incidence of GA development was 3.3% in both the no DA group and DRD2 groups over 5 years (Table 1). Older age (80–84, ≥85), AREDS2 use, intermediate AMD stage, and GA in fellow eye were associated with increased conversion to GA (Table 2). Exposure to DRD2 agonist was not significantly associated (HR, 1.00; 95% CI, 0.62–1.61; P > 0.90) with conversion to GA (Fig. 1B; Table 2).

L-DOPA Exposure

After PSM, there were 571 eyes exposed to L-DOPA and 1713 control eyes that were balanced across variables of interest (Table 1). The probability of conversion to GA between L-DOPA–exposed and control eyes is demonstrated in Figure 1C. The incidence of GA development was 3.2% in the no DA group and 2.3% in the L-DOPA group over 5 years (Table 1). Intermediate AMD stage was the only risk factor for conversion to GA in this analysis (Table 2). Exposure to L-DOPA was not significantly associated (HR, 0.74; 95% CI, 0.36–1.55; P = 0.40) with GA development (Fig. 1C; Table 2).

NaIO₃ Mouse Model of GA

Because we have reported on two independent databases with differing results in GA development, we sought to clarify these conflicting findings using animal models. The gold standard mouse model in GA research is the NaIO₃ model of RPE damage. NaIO₃ is a potent oxidant that specifically targets the RPE, leading to patchy RPE degeneration, photoreceptor death, and complement overactivation, mimicking features of advanced dry AMD.^17,18 We treated 9- to 12-week-old wild-type C57BL/6J mice intraperitoneally with 20 mg/kg NaIO₃ (3 female, 5 male) or PBS (4 female, 4 male) control (Fig. 2A) and assessed the effects on day 7. Using OCT, NaIO₃ treatment caused the accumulation of hyper-reflective foci (yellow arrows in Fig. 2B) and subretinal deposits (yellow arrowheads in Fig. 2B). Quantitative analysis revealed that NaIO₃ treatment resulted in a significant decrease in retinal thickness (P < 0.0001 for all locations), ONL thickness (P < 0.01 for all locations), and EZ thickness (P < 0.0001 for all locations) compared with PBS-treated controls (n = 8 mice per group) (Figs. 2C–E). Using choroidal wholemount immunofluorescence with bleaching, NaIO₃ treatment resulted in a 33.4% absolute decrease in CC density (P < 0.01; n = 6–10 mice per group) (Figs. 2F–G). Using choroidal wholemount immunofluorescence without bleaching, we found that NaIO₃ causes a 28.8% reduction in RPE cell density (P < 0.01; n = 8–10 mice per group) (Fig. 2H) and a 15.3-fold increase in the proportion of enlarged RPE cells (>750 µm²; P < 0.001) (Fig. 2I). Collectively, these data demonstrate that NaIO₃ induces both CC degeneration and RPE structural abnormalities, recapitulating key features of GA.

Figure 2.

NaIO₃ induces GA-like retinal and choroidal pathology in C57BL/6J mice. (A) Experimental schematic showing NaIO₃ or PBS administration with analysis at day 7. (B) Representative OCT images illustrating retinal structural abnormalities in NaIO₃-treated vs. PBS control eyes. Red lines indicate locations of retinal thickness measurements at three eccentricities (300, 500, and 600 µm) in each direction from the optic nerve head. Yellow arrows denote hyper-reflective foci, and yellow arrowheads indicate subretinal deposits. (C) Quantification of total retinal thickness. (D) Quantification of ONL thickness. (E) Quantification of EZ thickness. Data were analyzed using two-way analysis of variance (ANOVA) followed by Tukey's multiple comparisons test. (F) Choroidal wholemount immunofluorescence stained with podocalyxin (green, CC) and phalloidin (red, RPE) to visualize CC and RPE morphology (G). Quantification of CC area (Mann–Whitney test). (H) Quantification of RPE cell density (Welch's t-test). (I) Quantification of the distribution of RPE cell size in NaIO₃-treated vs. PBS controls (two-way ANOVA followed by Šidák's multiple comparisons test). n = 6–10 mice per group. **P < 0.01, ***P < 0.001, ****P < 0.0001. ns, not significant.

L-DOPA Was Ineffective in the NaIO₃ Model of GA

To assess whether L-DOPA can mitigate GA-like pathology, we evaluated its effects in 9- to 12-week-old C57BL/6J mice after NaIO₃-induced injury. Mice were treated with either PBS (control, n = 10 mice, 6 male and 4 female) or L-DOPA (n = 8 mice, 5 male and 3 female) with benserazide hydrochloride (a decarboxylase inhibitor that decreases peripheral conversion of L-DOPA to dopamine) daily for 2 days before and daily for 6 days after NaIO₃ administration (Fig. 3A). This dose of 30 mg/kg L-DOPA and 12 mg/kg benserazide over 1 week was previously shown to reduce laser-induced choroidal neovascularization (CNV) area and spontaneous CNV number and area in Vldlr^−/− mice^8,15 in two independent laboratories. NaIO₃ + PBS administration resulted in a significant reduction in total retinal thickness (P < 0.0001 for all locations), ONL thickness (P < 0.001 for all locations), and EZ thickness (P < 0.0001 for all locations) compared with baseline (Figs. 3B–E, blue symbols). Similarly, NaIO₃ + L-DOPA treatment resulted in significant retinal degeneration, including decreased total retinal thickness (P < 0.0001 for all locations), ONL thickness (P < 0.01 for all locations), and complete loss of the EZ (P < 0.0001 for all locations) compared with baseline (Figs. 3C–E; red symbols). We detected no significant differences between NaIO₃ + PBS and NaIO₃ + L-DOPA groups in any measurement at any location (Figs. 3C–E). Choroidal immunofluorescence analysis demonstrated an equally severe reduction in CC density, RPE density, and RPE organization between groups (Figs. 3F–I). No sex-specific effects were identified (Supplementary Fig. S1). These data suggest that L-DOPA was ineffective at decreasing RPE damage and GA-like pathology in pigmented mice.

Figure 3.

L-DOPA does not attenuate NaIO₃ induced pathology in pigmented mice. (A) Schematic of the experimental design showing NaIO₃ administration, with PBS or L-DOPA treatment given 2 days before and continuing for 6 days after NaIO₃ exposure. (B) Representative OCT images demonstrating comparable retinal structural abnormalities in both treatment groups. Red lines indicate locations of retinal measurements. Yellow arrows denote hyper-reflective foci, and yellow arrowheads indicate subretinal deposits. (C) Quantification of total retinal thickness. (D) Quantification of ONL thickness. (E) Quantification of EZ thickness. Data were analyzed using two-way analysis of variance (ANOVA) followed by Tukey's multiple comparisons test. (F) Choroidal wholemount immunofluorescence stained with podocalyxin (green, CC) and phalloidin (red, RPE) to assess CC density and RPE morphology. (G) Quantification of CC area (unpaired t-test). (H) Quantification of RPE cell density (unpaired t-tests). (I) Quantification of distribution of RPE cell size (two-way ANOVA with Šidák's multiple comparisons test), n = 8–10 mice per group. ****P < 0.0001. ns, not significant.

Because pigmented mice produce L-DOPA endogenously as a byproduct of pigment synthesis and endogenous pigment affects the ability of L-DOPA to inhibit CNV development in the Vldlr^−/− mouse model of neovascular AMD,¹⁵ we performed parallel experiments in 9- to 12-week-old C57BL/6J Tyr^c-2J/c-2J albino mice (n = 9 mice per group) (Fig. 4A). Similar to pigmented mice, OCT imaging showed comparable retinal abnormalities, including hyper-reflective foci, subretinal deposits, and reduced retinal thickness in both PBS– and L-DOPA–treated groups (Fig. 4B). Compared with baseline measurements, NaIO₃ + PBS–treated albino mice exhibited significant reductions in total retinal thickness (P < 0.0001 for all locations), ONL thickness (P < 0.01 for all locations), and EZ thickness (P < 0.0001 for all locations) compared with baseline (Figs. 4C–E, blue symbols). Similarly, NaIO₃ + L-DOPA–treated mice showed significant decreases in all parameters (P < 0.0001) (Figs. 4C–E). There were no significant differences between NaIO₃ + PBS and NaIO₃ + L-DOPA treatment in total retinal and EZ thickness (Figs. 4C, 4E). However, L-DOPA treatment caused a significant further reduction in ONL thickness compared with PBS treatment (P < 0.01 for all locations) (Fig. 4D). Choroidal immunofluorescence analysis demonstrated that CC density was significantly reduced by 16% in NaIO₃ + PBS treated mice (P < 0.05) (Figs. 5A, 5B) compared with untreated controls. In NaIO₃ + L-DOPA–treated mice, CC density decreased by 26.3% compared with untreated controls (P < 0.001), which was significantly different and worse compared with NaIO₃ + PBS (P < 0.05) (Figs. 5A, 5B). RPE density was not significantly different between groups, although trends were present for reductions after NaIO₃ treatment (Fig. 5C). Finally, the population of enlarged (>750 µm²) RPE cells was significantly increased in NaIO₃ + L-DOPA–treated mice (Fig. 5D) compared with untreated albino controls. No sex-specific effects were identified (Supplementary Fig. S2) (male = 6 PBS vs. 5 L-DOPA, female = 3 PBS vs. 4 L-DOPA). Collectively, these results demonstrate that L-DOPA does not attenuate NaIO₃-induced CC degeneration, RPE loss, or retinal structural abnormalities in pigmented mice, and may even exacerbate degeneration in albino mice.

Figure 4.

L-DOPA does not prevent NaIO₃-induced retinal degeneration in albino mice. (A) Experimental schematic. (B) Representative OCT images. Red lines indicate locations of retinal measurements. Yellow arrows denote hyper-reflective foci, and yellow arrowheads indicate subretinal deposits. (C) Quantification of retinal thickness. (D) Quantification of ONL thickness. (E) Quantification of EZ thickness. Data were analyzed using two-way analysis of variance followed by Tukey's multiple-comparisons test.

Figure 5.

L-DOPA treatment doesn't preserve CC integrity or RPE morphology after NaIO₃-induced retinal injury in albino mice. (A) Representative choroidal whole-mount immunofluorescence images from untreated control, NaIO₃ + PBS, NaIO₃ + L-DOPA–treated albino mice. Samples were stained with podocalyxin (green) to visualize the CC and phalloidin (red) to label the RPE. (B) Quantification of CC area. (C) Quantification of RPE cell density. (D) Distribution of RPE cell size. Data were analyzed using one-way analysis of variance (ANOVA) followed by Tukey's multiple-comparisons test for (B) and (C), and two-way ANOVA followed by Šidák's multiple-comparisons test for (D). Data were obtained from 3 to 10 mice per group. Statistical significance is indicated as follows: P < 0.05, P < 0.01, P < 0.001, P < 0.0001; ns, not significant.

Discussion

In this study, we investigated the role of dopaminergic signaling in mitigating GA using complementary clinical and preclinical experimental approaches. Across propensity score–matched clinical cohorts, exposure to dopaminergic agents was not associated with a reduced risk of GA development (Fig. 1; Table 2) (n = 1713 control and 571 L-DOPA–exposed eyes). These findings contrast with our prior report (n = 2532 control and 844 L-DOPA–exposed eyes), showing a protective effect of L-DOPA on new-onset GA development.¹⁰ These discrepancies could reflect a reduction in sample size, but more likely reflect the challenges of database research with unknown confounders and analysis based on diagnosis coding. Thus, the replicability of database studies is essential in drug repurposing research and animal models can help to unravel these complexities.

Using the NaIO₃ mouse model of GA, L-DOPA did not confer RPE nor CC protection (Figs. 3, 4). The NaIO₃ model recapitulates key pathological features of GA, including RPE oxidative stress–induced degeneration and secondary CC atrophy.^17,18 Dopaminergic signaling reduces oxidative stress^11,19,20; however, this reduction was not enough to overcome the severe oxidative stress induced by NaIO₃ on the RPE. However, other mechanisms contribute to GA that are not tested by the NaIO₃ model, including exosome release and primary choroidal vasculature changes. Dopamine signaling has been implicated in reducing exosome release¹² and dilating the choroidal vasculature,¹³ which may still impact GA development and/or progression in patients with AMD.

It is possible that the 30 mg/kg L-DOPA given in this study is a subtherapeutic dose for GA. However, 30 mg/kg L-DOPA has previously been shown to reduce laser CNV and spontaneous CNV number and area in two prior mouse studies.^8,15 Additionally, we found that 30 mg/kg L-DOPA exacerbated ONL loss in albino mice (Fig. 4), suggesting it is not subtherapeutic and in fact potentially harmful. Further, in Parkinson's disease, L-DOPA is dosed at 400 to 1600 mg/d, which would be 5 to 20 mg/kg in an average 80-kg adult. Additionally, we previously showed that L-DOPA at 100 to 300 mg/d reduced neovascular AMD development.⁷ Based on these mouse and human studies, if 30 mg/kg L-DOPA were a subtherapeutic dose for GA, then this dose would be too high in patients for clinical utility.

In contrast with GA, previous work in our laboratory and by others has demonstrated reproducible benefits of dopaminergic signaling in neovascular AMD models and patients. Using the choroidal sprouting assay, two independent laboratories found that L-DOPA and DRD2 agonists reduce choroidal angiogenesis ex vivo.^8,15 Using in vivo models of choroidal angiogenesis L-DOPA and DRD2 agonists decreased the laser-induced CNV area⁸ and spontaneous CNV number and area in the Vldlr^−/− mouse.¹⁵ Consistent with these preclinical findings, database studies report associations between L-DOPA exposure and increased age of onset of neovascular AMD^6,8 and reduced number of anti-VEGF injections.^7,8 Further, L-DOPA exposure is associated with reduced new-onset neovascular AMD in three independent databases.^7,9 Finally, a small, nonrandomized prospective clinical trial found that L-DOPA therapy delayed the need for anti-VEGF treatment in new-onset neovascular AMD patients and reduced the anti-VEGF injection burden in previously treated neovascular AMD patients.²¹ Together, these data provide strong preclinical and clinical data to support a randomized, prospective clinical trial. In fact, the Levodopa for Neovascular AMD (LAMDARCT) trial in Europe is expected to provide prospective evidence regarding the efficacy and therapeutic potential of dopaminergic agents to reduce anti-VEGF injection burden in neovascular AMD.

Although our findings suggest that L-DOPA is ineffective in mitigating RPE damage and GA-like pathology in this model, some of its limitations should be acknowledged. First, the NaIO₃ model reflects an acute injury paradigm and does not fully recapitulate the chronic and progressive nature of GA. Moreover, variability in model induction, with approximately 50% penetrance, may reduce the sensitivity to detect potential therapeutic effects. Second, this study primarily focused on oxidative stress, whereas other key mechanisms implicated in GA pathogenesis, including inflammation, complement dysregulation, and mitochondrial dysfunction, were not evaluated. In summary, the evidence for L-DOPA in neovascular AMD includes three different animal models, four independent database studies, and one small, prospective clinical trial. By comparison, the evidence for L-DOPA in GA includes one positive database study, one negative database study, and one negative animal model. Further investigation of L-DOPA may be more appropriately directed toward neovascular AMD rather than GA. The results of the European LAMDARCT trial will be critical in clarifying the therapeutic potential of dopaminergic agents in neovascular AMD.

Appendix. Appendix: The SOURCE Consortium

Members of the SOURCE Consortium and their site principal investigators include the Henry Ford Health System: Sejal Amin, Paul A. Edwards; Johns Hopkins University: Divya Srikumaran, Fasika Woreta; Montefiore Medical Center: Jeffrey S. Schultz, Anurag Shrivastava; Medical College of Wisconsin: Baseer Ahmad; Northwestern University: Dustin D. French, Paul Bryar; Scheie Eye Institute: Brian L. Vanderbeek; Stanford University: Suzann Pershing, Sophia Y. Wang; University of Colorado: Jenna Patnaik; Prem Subramanian; University of Maryland: Saleha Munir, Wuqaas Munir; University of Michigan: Joshua Stein, Lindsey DeLott; University of Utah: Brian C. Stagg, Barbara Wirostko; West Virginia University: Brian McMillan; Washington University: Arsham Sheybani; Yale University: Soshian Sarrapour, Kristen Nwanyanwu; Ji Liu; University of California, San Francisco: Michael Deiner, Catherine Sun; University of Texas – Houston: Robert Feldman; University of Rochester: Rajeev Ramachandran; SUNY Upstate: Preethi Ganapathy; Mount Sinai School of Medicine: Louis R. Pasquale; Oregon Health Sciences University: Michelle Hribar, Merina Thomas. The SOURCE Data Center is located at the University of Michigan. The Chief Data Officer of SOURCE is Joshua Stein. The Lead Statistician of SOURCE is Chris Andrews. More information about SOURCE is available at https://www.sourcecollaborative.org/.