A first-in-human phase 1 trial has brought an intriguing new vision-restoration concept into the clinic: intravitreal photoswitch therapy for advanced retinitis pigmentosa (RP). Reported in Nature Medicine, the open-label ABACUS-1 study evaluated KIO-301, a small azobenzene-based molecule designed to confer light responsiveness to surviving retinal ganglion cells in eyes with end-stage photoreceptor loss. In a field dominated by gene-specific approaches, the appeal is obvious: this is a gene-agnostic strategy aimed at patients with profound vision loss, regardless of the underlying mutation.

KIO-301 was tested in six participants (12 eyes) with advanced RP, divided into two cohorts based on baseline visual function. Three participants had no light perception (NLP) or bare light perception (BLP), while three had hand-motion (HM) or count-fingers (CF) vision. The study used sequential unilateral then contralateral intravitreal dosing, escalating from 7.5 µg to 50 µg across cohorts. All participants completed the protocol and reached the primary day-30 safety endpoint.

The main result was reassuring: the primary safety outcome was met. No serious adverse events or dose-limiting toxicities were observed, and there were no drug-related systemic adverse events. Importantly, no participant developed intraocular inflammation, vitreous haze, macular edema, or treatment-related structural retinal changes on fundus examination, autofluorescence, or OCT. Reported ocular adverse events were mild and procedure-related, including transient eye pain, swelling, and one case of ocular hypertension that resolved with topical therapy. For retina specialists accustomed to balancing efficacy with inflammatory risk in some gene and cell therapy programs, that clean early safety signal is notable.

While this was not an efficacy trial, several exploratory signals still hint at biologic activity. In some cohort 1 eyes with NLP or BLP at baseline, light perception task performance varied after dosing, and one participant with longstanding NLP reported subjective light awareness within two days of treatment. Functional vision testing also showed time-dependent fluctuations, with the clearest gains appearing at early post-treatment visits before trending back toward baseline by day 30.

Perhaps most compellingly, functional MRI demonstrated stimulus-associated BOLD signal in occipital visual cortical regions after treatment. Activity was shown in V1 and adjacent extrastriate cortex at early post-treatment time points, again with a temporal profile consistent with short-lived pharmacodynamic engagement. The study authors are careful not to over-interpret these findings, but the alignment of subjective reports, task variation, and cortical activation provides an important proof-of-principle.

Mechanistically, KIO-301 differs sharply from optogenetic gene therapy. Rather than introducing new genetic material, the molecule is intended to enter surviving retinal ganglion cells and modulate endogenous ion channels in a light-dependent manner. That makes it potentially simpler to deliver and easier to repeat, though durability remains a key unanswered question.

This was a very small, uncontrolled, safety-focused study, and it cannot tell us whether photoswitch therapy will deliver meaningful clinical benefit. But it does something arguably just as important at this stage: it establishes that intravitreal photoswitch delivery is feasible and appears well tolerated in humans with advanced RP. For ophthalmologists following restorative retinal therapeutics, ABACUS-1 opens a new – and distinctly pharmacologic – path toward mutation-agnostic vision restoration.