Vision Restoration with the PRIMA System in Geographic Atrophy Due to AMD
Full text via PMC open-access author manuscript. N Engl J Med. 2026;394:232–242. Published online October 20, 2025. DOI: 10.1056/NEJMoa2501396.
Authors: Holz FG (Univ. Bonn — corresponding), Le Mer Y, Muqit MMK (Moorfields), Hattenbach LO, Cusumano A, Grisanti S, Kodjikian L, Pileri MA, Matonti F, Souied E, Stanzel BV, Szurman P, Weber M, Bartz-Schmidt KU, Eter N, Delyfer MN, Girmens JF, van Overdam KA, Wolf A, Hornig R, Corazzol M (Science Corp.), Brodie F, Olmos de Koo L, Palanker D (Stanford; co-senior), Sahel JA (co-senior).
Funding: Science Corporation (previously Pixium Vision SA); Moorfields Hospital NIHR Biomedical Research Centre (NIHR203322).
Study design
Open-label, baseline-controlled, multicenter, prospective, single-arm confirmatory clinical trial — the PRIMAvera study (NCT04676854).
- Sites: 17 clinical sites in 5 European countries (France, Germany, UK, Italy, Netherlands).
- Population: 38 participants; mean age 78.9 ± 6.41 years (18M/20F); geographic atrophy (GA) due to AMD in both eyes; VA ≥ logMAR 1.2 (≤20/320 Snellen); fovea- involving atrophy > implant size (> 2.4 mm diameter) in the study eye.
- Exclusion: wet AMD, pre-existing epiretinal membrane likely to complicate surgery.
- Design authority: Science Corporation with advisory board; local ethics committee approvals; DSMB (Bakri, Wolf, Wolfensberger) reviewed all safety data.
Device description
The PRIMA system comprises:
- Subretinal implant: 2 × 2 mm, 30 µm thick crystalline silicon array with 378 pixels each 100 µm in size. Each pixel has a central active electrode, hexagonal return electrode mesh, and two photodiodes. Subretinally placed within the atrophic lesion; no transcutaneous wires (fully photovoltaic).
- PRIMA glasses: frame-mounted camera + NIR projector (880 nm, 30 Hz frame rate, pulse duration 0.7–9.8 ms, peak irradiance 3.5 mW/mm²). Transparent lens allows simultaneous natural peripheral + prosthetic central vision.
- Pocket processor: converts camera feed (with adjustable zoom 1×–12×, contrast, brightness) into projection pattern on implant.
- Mechanism: NIR light → photodiode array → bipolar cell stimulation → inner retinal processing → optic nerve → brain. Unlike epiretinal (Argus II) stimulating RGCs/axons, PRIMA stimulates bipolar cells, preserving inner retinal signal processing.
Endpoints
- Primary efficacy: proportion achieving VA improvement of ≥ logMAR 0.2 (≥10 ETDRS letters) at 12 months; null hypothesis (failure) = ≤50% of participants.
- Primary safety: number and severity of device/procedure-related serious adverse events (SAEs) at 12 months.
- Secondary: VA improvement at 6 months; mean VA at 6 + 12 months; IVI questionnaire (Impact of Vision Impairment); central visual perception at 12 months.
- Natural VA (without glasses) monitored as safety endpoint.
Results
Participants and follow-up
38 enrolled; 32 reached 12-month endpoint. Six did not complete: 3 deaths (unrelated to device/procedure), 1 withdrawal, 2 unavailable. Multiple imputation (MI) used to account for missing data.
Efficacy at 12 months
| Metric | Value |
|---|---|
| Primary endpoint (≥0.2 logMAR), observed (26/32) | 81.3% (95% CI: 63.56–92.79%; p<0.001) |
| Primary endpoint, MI estimate (all 38) | 79.9% (95% CI: 65.58–94.16%; p<0.001) |
| ≥0.3 logMAR improvement (post-hoc), observed | 78.1% (25/32; 95% CI: 60.03–90.72%) |
| Mean VA improvement with PRIMA glasses | logMAR 0.49 (95% CI: 0.35–0.63; ~24.5 letters) |
| Mean VA improvement, participant’s choice | logMAR 0.51 (95% CI: 0.39–0.64; ~25.5 letters) |
| Best individual improvement | logMAR 1.18 (59 letters) |
| Natural VA change (without glasses) | logMAR 0.00 (95% CI: –0.05 to 0.04) — no change |
At 6 months: 20/35 (57%) achieved ≥0.2 logMAR; mean +0.32 logMAR with glasses.
Prosthetic VA (direct stimulation, no camera): mean logMAR 1.32 ± 0.16 (Snellen 20/417) — exactly matches the theoretical sampling limit of 100 µm pixels (logMAR 1.32). 30 of 32 participants had measurable central perception.
Functional use: 84.4% (27/32) could read letters, numbers, and words using prosthetic vision at home; 68% reported medium-to-high user satisfaction. IVI questionnaire showed no significant behavioral change (study underpowered for this outcome; questionnaire may be insufficiently sensitive for this population).
Safety
26 device- or procedure-related SAEs in 19 participants:
| SAE Type | N participants | N events |
|---|---|---|
| Ocular hypertension | 6 | 6 |
| Peripheral retinal break | 5 | 5 |
| Macular hole | 3 | 3 |
| Subretinal hemorrhage | 3 | 4 |
| Choroidal neovascularization (CNV) | 2 | 2 |
| Choroidal hemorrhage | 1 | 1 |
| Choroidal fold | 1 | 1 |
| Proliferative vitreoretinopathy | 1 | 1 |
| Retinal detachment | 1 | 1 |
| Retinal hemorrhage | 1 | 1 |
| Thrombophlebitis | 1 | 1 |
- Timing: 21/26 (81%) within 2 months postoperatively; 20/21 resolved within 2 months. All are consistent with risks of vitrectomy + subretinal surgery.
- Severity: 22/26 (85%) mild or moderate; 4 severe (macular hole, ocular hypertension, retinal detachment, PVR — the latter three in one participant, resolved with surgery + silicone oil tamponade).
- No SAE was device-alone (all procedure-related, or procedure+device).
- CNV: 2 participants (5.2%), both treated with intravitreal anti-VEGF pro re nata; CNV occurs in GA without intervention and has been reported in 16% of vitrectomised dry-AMD eyes at 2.6 years.
- 5 peripheral retinal breaks — all treated intraoperatively, none became rhegmatogenous RD.
- SAE rate did not predict VA outcome: 79% of SAE-affected participants and 83% of unaffected achieved ≥0.2 logMAR improvement.
- Atrophy area increased 8.5 mm² in study eye vs 2.5 mm² in fellow eye at 12 months (attributed to post-surgical effects); inner retinal structure in the atrophic area unchanged on OCT.
- DSMB conclusion: benefits outweigh risks of implantation.
Discussion highlights
- Comparison to prior devices: Argus II best acuity logMAR 1.8 (20/1260); Alpha AMS subretinal up to ~20/546; suprachoroidal (44-ch BVT) logMAR ~3.0. PRIMA is the first to achieve form vision rather than only phosphenes.
- Advantage over epiretinal approach: PRIMA stimulates bipolar cells (subretinal), preserving inner retinal signal processing and retinotopic map. Argus II stimulated ganglion cell axons → elongated percepts / distorted maps.
- Eye movement advantage: PRIMA allows natural gaze-based scanning (unlike head- mounted camera systems); fixational eye movements improve resolution beyond the pixel sampling limit (super-resolution effect).
- Zoom enhancement: digital zoom enabled reading of fonts down to 20/42 — far smaller than the 100 µm pixel limit (equivalent Snellen ~20/400); zoom/contrast enhancement is therefore integral to the system’s functional acuity.
- Longevity signal: 4-year feasibility study (5 patients) showed stable anatomy with minimal inner retinal thinning over 36 months; mean VA logMAR 0.64 (32 letters) at 48 months — the device is not deteriorating.
- Upgrade pathway: wireless design enables in-situ replacement with higher-resolution next-gen chips (Bhuckory et al., Nat Commun, 2025, DOI: 10.1038/s41467-025-58084-y) or tiling multiple modules via minimal incision to cover larger atrophic areas.
- Natural vision preservation: no change in peripheral natural VA — the lens in PRIMA glasses is transparent, so prosthetic and natural vision coexist simultaneously.
Companion editorial
Duncan J. “Behind the science: restoring vision for patients with AMD and geographic atrophy.” N Engl J Med. 2026;394:298–301. DOI: 10.1056/NEJMe2514592.
Citation
Holz FG, Le Mer Y, Muqit MMK, et al. Vision Restoration with the PRIMA System in Geographic Atrophy Due to AMD. N Engl J Med. 2026;394:232–242. DOI: 10.1056/NEJMoa2501396. PMC7618305.