Photobiomodulation for Dry AMD: What the Trials Show

Photobiomodulation has not proven that it slows dry age-related macular degeneration (AMD). The first meta-analysis to pool randomized trials found no statistically significant advantage over sham treatment for visual acuity, drusen volume, geographic atrophy area, central retinal thickness, or microperimetry. Yet one later randomized trial in people with early-stage dry AMD reported improvement across several measures of visual function. The honest answer is not that light therapy works, or that it has failed forever. It is that the evidence remains too inconsistent to support routine treatment—and specific enough to justify a better trial in early disease.

That distinction matters. Dry AMD can progress to irreversible loss of retinal pigment epithelium and photoreceptors. A treatment that helps stressed but intact cells would be very different from a treatment that restores cells already gone. Biology allows that possibility. Clinical evidence has not established it.

The pooled evidence is negative, with a large asterisk

Chen et al.'s 2025 systematic review and meta-analysis combined six randomized controlled trials, involving 360 participants and 477 eyes. The trials compared multiwavelength photobiomodulation (PBM) with sham treatment and assessed best-corrected visual acuity (BCVA), macular drusen volume, geographic atrophy area, central subfield thickness, and microperimetry. PBM did not produce a statistically significant benefit on any of those prespecified endpoints.

That is the current starting point for anyone considering PBM for dry AMD. Better vision was not established. Less drusen burden was not established. Slower geographic atrophy was not established.

But the pooled BCVA result came with very high heterogeneity: I² was 83%. In ordinary English, the included studies differed so much that their results may not describe one common treatment effect. They used different devices, wavelength combinations, dosing schedules, visit structures, outcome windows, and AMD populations. One trial assessed outcomes after two weeks; others followed patients for months or longer. The review also combined active clinic-based treatment with a nightly home-worn OLED mask. Calling all of that one intervention is convenient for a spreadsheet and awkward for a patient.

High heterogeneity does not turn a null result into a positive one. It does mean the pooled estimate cannot settle whether one particular protocol might work in one particular stage of disease. An earlier independent synthesis, Rassi et al.'s 2024 meta-analysis, also found a mixed-to-inconclusive efficacy picture. The 2021 Cochrane review by Henein et al. likewise concluded that evidence was insufficient to support PBM for non-exudative AMD. Three evidence syntheses point in the same direction: PBM has not met the standard for an established dry-AMD therapy.

What the LIGHTSITE trials add—and what they cannot answer

LumiThera's Valeda system delivers light at 590, 660, and 850 nm. It is the platform used in the LIGHTSITE studies and has a CE mark in Europe. Commercial availability, however, does not prove that a treatment improves outcomes. The peer-reviewed evidence discussed here does not document FDA 510(k) clearance for an AMD indication.

LIGHTSITE II was a nine-center randomized controlled trial in people with intermediate non-exudative AMD. Participants received nine-session series of triwavelength PBM every four months over 10 months, or sham treatment. The study reported directional visual-improvement signals and possible disease-modifying effects. COVID-19 disruptions, however, compromised the protocol enough that the signals cannot be cleanly attributed to PBM rather than altered follow-up and care patterns.

That limitation carries unusual weight because trial size and follow-up alone do not repair compromised treatment delivery. LIGHTSITE I, one of the smaller studies in the literature, enrolled 46 eyes at a single center. No published power calculation is available in the evidence summarized here, and none of the individual PBM trials reached the scale usually needed to reliably detect a clinically meaningful difference of five ETDRS letters.

LIGHTSITE III's 24-month analysis, published in 2026, provides the longest follow-up among the cited PBM randomized trials for dry AMD and used the same triwavelength Valeda platform. Its timing matters: the full 24-month report appeared after the January 2025 search cutoff for Chen et al.'s meta-analysis. Longer follow-up gives researchers a better chance to see whether early visual signals persist and whether structural outcomes change. It does not erase the earlier pooled null findings, and the available summary does not provide numerical 24-month efficacy estimates to judge clinical size or statistical precision.

The early-stage trial is the reason the question remains open

The most interesting newer evidence comes from Küçükerdönmez et al.'s 2025 randomized trial. Unlike the mixed-stage LIGHTSITE populations, this study prospectively enrolled people with early-stage dry AMD and tested multiwavelength PBM. It reported improvement in multiple aspects of visual function in that early-stage population.

This finding deserves more weight than a post-hoc subgroup claim because the trial defined early disease before treatment began. It was also published after Chen et al.'s search cutoff, so it was not part of the six-trial pooled estimate. Still, one positive randomized trial without an adequately powered, multicenter replication does not establish clinical practice.

The stage-specific hypothesis has a coherent biological rationale. Red and near-infrared wavelengths may act through cytochrome c oxidase in mitochondria, increasing ATP production and reducing reactive oxygen species. Proposed downstream effects include reduced NF-κB-related inflammatory signaling, lower complement activation, and increased neuroprotective factors such as BDNF and NGF. These mechanisms remain hypotheses in this clinical setting, not proof that PBM prevents vision loss in humans.

They do explain why disease stage could matter. In early AMD, retinal cells may be metabolically impaired while still structurally present. In geographic atrophy, retinal pigment epithelium and photoreceptors have already died. Providing energy support to absent cells would be a difficult business model, even for mitochondria.

Safety is reassuring, but it is not efficacy

Chen et al. found fewer adverse events in the PBM group than in the sham group at six months: relative risk 0.48, with p=0.007. The review did not provide a confidence interval for that estimate in the available report. This result supports a generally favorable short-term safety profile in the studied settings, but it should not be read as evidence that PBM protects the retina. Differences in discomfort from sham exposure or in symptom reporting could also contribute to the imbalance.

Nor should PBM displace better-supported care. For intermediate AMD, AREDS2 supplementation remains the evidence-based standard, supported by a much larger multicenter randomized trial program of roughly 4,000 participants. For geographic atrophy, complement inhibitors have substantially larger phase III evidence bases than the entire PBM literature. A patient with geographic atrophy should not prioritize an unproven light protocol over treatment decisions with a retinal specialist.

Verdict: promising question, unproven treatment

PBM for dry AMD is not ready for routine clinical adoption. The best pooled evidence finds no statistically significant benefit on visual acuity, drusen volume, geographic atrophy area, retinal thickness, or microperimetry, while the extreme BCVA heterogeneity makes broad conclusions about every device and protocol less secure.

The early-stage randomized trial identifies a plausible target population worth studying, not a green light for self-treatment. The next useful study would prospectively enroll early-stage AMD, use a standardized device and dosing schedule, include enough patients to detect a clinically meaningful visual difference, and follow structural and functional outcomes long enough to matter. Until then, people considering PBM should discuss it with a retinal specialist, treat it as experimental, and continue established AMD care.