PRIMA at 3 years: Early surgical risk, durable vision gains in GA

Patients with geographic atrophy (GA) have had treatments that slow lesion growth but nothing that restores central vision once it is lost. That changed in July 2026, when Science Corporation received a CE mark for the PRIMA photovoltaic retinal implant, making it commercially available across 30 European countries. The US Food and Drug Administration (FDA) review is ongoing.

The 12-month results of the pivotal PRIMAvera trial were published in the New England Journal of Medicine in October 2025. In that trial, 26 of 32 participants who reached 12 months (81.3%) had a clinically meaningful gain in prosthetic visual acuity. The same report recorded 26 serious adverse events (SAEs) in 19 participants, most occurring within the first 2 months after implantation. As retina specialists begin fielding patient questions about the device, longer-term safety data and clear patient-selection criteria have become the central questions.

At the American Academy of Ophthalmology 2026 Annual Meeting, Ehsan Rahimy, MD, presented 36-month safety data from PRIMAvera. Rahimy is a retina specialist at the Palo Alto Medical Foundation in California and holds an adjunct faculty appointment at Stanford University.

In this Q&A, Rahimy discusses how early surgical complications changed over 3 years and how clinicians should interpret the SAE rate. He also covers which patients are appropriate referrals, the rehabilitation infrastructure practices will need, and how PRIMA could fit alongside complement inhibitors over the course of GA.

The Q&A

Note: Transcript lightly edited for clarity.

Your presentation extends follow-up of the PRIMAvera trial to 36 months, 2 years beyond the 12-month results published in the New England Journal of Medicine. Walk us through the key safety findings at this timepoint. How did the adverse events observed early after surgery evolve, and did any new device- or procedure-related events emerge with longer follow-up?

Ehsan Rahimy, MD: The most compelling takeaway from the 36-month data is that the safety profile of the PRIMA system emulates what I would expect from an intricate vitreoretinal surgical procedure rather than signaling a chronic, progressive biomaterial hazard causing long-term safety concerns. We observed a total of 29 serious adverse events (SAEs) in 38 implanted participants. The vast majority of these (83%) were concentrated entirely within the first 6 months post-implantation. These were primarily procedure-related events—such as transient IOP spikes, localized subretinal hemorrhage, or retinal tears—that are within the wheelhouse of an experienced retina surgeon to manage.

Notably, once the early postoperative healing window closed, new device-related events became exceedingly rare, with only 5 SAEs being reported between months 6 and 36. We are not seeing late-onset device extrusion, migration, or chronic foreign-body inflammation out to 3 years, which is further corroborated by histopathologic data we present showing excellent preservation of the inner nuclear layer and an absence of fibrous encapsulation.

Nineteen of the 38 patients implanted in PRIMAvera experienced a serious adverse event, most occurring early after surgery and most resolving. How should retina specialists interpret that figure in light of the 36-month data? Which events matter most when counseling patients, and which do you think reflect a surgical learning curve that should diminish with experience?

Rahimy: When counseling patients or reviewing this data with colleagues, context is everything. Hearing that 19 out of 38 patients experienced an SAE may appear daunting in a vacuum. However, you have to look at the functional outcomes of those specific patients. Our data clearly demonstrated that experiencing an SAE did not compromise the natural residual vision of these eyes regardless (when compared to their fellow eye). Furthermore, it did not permanently impede the patient’s ability to gain prosthetic vision—68% of patients in the SAE group still managed to achieve a ≥0.3 logMAR improvement, which was not statistically significantly different from the 83% responder rate in the non-SAE group (p=0.45).

Additionally, the willingness to accept these risks may depend on the personality type of a patient. It is important to be aware that the patients enrolled in this study had severely compromised central acuity (baseline Snellen equivalent of 20/460) at baseline. Some individuals may be more willing to accept an increased risk profile that may come with a complex vitreoretinal procedure as a trade-off in the hopes of improving their central vision. Each case is going to be unique, and I imagine our individual conversations we have with our patients will be tailored to their particular needs and personalities.

As for the surgical learning curve, subretinal device implantation in thin, atrophic tissue is technically demanding. As we hopefully move from clinical trials into wider real-world adoption, surgical techniques for bleb creation and retinotomy management will become more standardized, and I expect that early complication rate to diminish, especially as individual surgeons continue to gain reps by doing more procedures.

At 12 months, patients gained a mean of approximately 25 letters, yet the vision PRIMA provides is prosthetic: monochrome, limited by pixel resolution, and dependent on the glasses. What does the durability of that gain through 36 months mean for patients’ everyday function, such as reading, recognizing faces, or navigating their surroundings? What questions do the data not yet answer?

Rahimy: A mean gain of nearly 30 letters (0.58 logMAR) at 36 months is a significant leap for a patient who started with a central scotoma. While it is prosthetic form vision, the practical applications are still significant. At 12 months, 84% of participants reported being able to read letters and words at home. By 36 months, over half the cohort was using the system for advanced daily tasks like reading menus, solving crosswords, and checking train departures. I believe this type of feedback indicates a shift away from “laboratory/research vision” and more into functional, daily use vision.

What remains unanswered—and what we will continue to study—is the upper limit of neuroplasticity in terms of continued long-term improvements and refinements as the patient gains confidence and expertise in utilizing their device. We saw visual acuity continue to improve from month 12 (-0.49 logMAR) to month 36 (-0.58 logMAR). Understanding exactly how the visual cortex continues to adapt to these photovoltaic signals over a 5- or 10-year horizon is a very exciting frontier.

PRIMAvera enrolled patients with advanced, foveal-involving geographic atrophy and severely reduced visual acuity. Based on the 36-month findings, who is the ideal candidate for PRIMA? Conversely, which patients should clinicians not refer, whether because of anatomy, residual natural vision, cognitive factors, or ability to commit to visual rehabilitation?

Rahimy: The ideal candidate is someone with profound, foveal-involving geographic atrophy who has lost functional central vision but retains a structurally intact inner retina (the bipolar cells and ganglion cells). Just as importantly, they must have intact peripheral vision for ambient navigation, which the PRIMA system optimally preserves due to the transparent nature of the augmented reality glasses together with the microchip implant only covering the size of the atrophic macula.

However, meeting certain functional/anatomic metrics I would argue is only a small fraction of the equation. I would argue that personal, behavior, and social implications matter as much, if not even more, when deciding on the right type of patient to refer. I would not refer a patient with significant neurological cognitive decline or one lacking a robust social support system. This technology is not a passive fix; it requires immense dedication to the post-operative visual rehabilitation protocols. If a patient is unwilling or unable to commit to the post-implantation training necessary to interpret these new visual signals and utilize the pocket processor, they will not achieve the functional gains we saw in the trial. So, to that end, I am looking for highly motivated patients with good neurocognitive health and a robust support system to commit to the postoperative rigorous rehabilitation schedule.

Outcomes with PRIMA depend substantially on postimplantation training and device optimization. How should practices think about the rehabilitation infrastructure this approach requires? And how would you frame realistic expectations in a first conversation with a patient who has lost central vision to geographic atrophy?

Rahimy: From a practice management standpoint, adopting PRIMA means adopting a low-vision rehabilitation infrastructure. Just as we look at expanding clinical support staff to handle new procedures in the retina clinic (e.g. photobiomodulation device for intermediate-stage dry AMD), deploying PRIMA requires integrated occupational therapists and low-vision specialists.

In that first conversation with a patient, I frame it clearly: "This is not a magic switch that returns your vision to how it was in your 40s. We are building an entirely new visual pathway." I explain that the surgery, even though it is a critical step, is just step one. Step two is learning a new visual language. The fact that patient outcomes improved steadily between years one and three tells us that patience and training are the primary drivers of success once the hardware is in place.

Having the right type of team in place to help patients navigate through this critical period post-operatively will be paramount. I am sure our field will lean on assistance, resources, and support from Science Corp, while at the same time figuring out what works best in the real world setting. Clearly, retina specialists do not have the time or bandwith to handle this entirely on their own. I would anticipate integrating physician extenders (e.g., occupational therapists, physician assistants, and/or optometrists). I also anticipate that certain “centers of excellence” are likely to emerge around the country where the majority of these procedures are being referred locally and being performed. Such centers are likely to invest in the necessary framework to manage these patients during their post-operative training.

Complement inhibitors aim to slow the progression of geographic atrophy, whereas PRIMA aims to restore function after central vision has been lost. How do you see these approaches fitting together across the course of a patient’s disease? Could a patient treated with pegcetacoplan or avacincaptad pegol eventually become a candidate for PRIMA?

Rahimy: In my eyes, complement inhibitors and photovoltaic implants are potentially synergistic, not competitive. Medications like pegcetacoplan and avacincaptad pegol are our defensive line - we use them to slow the expansion of extrafoveal or early foveal lesions to preserve native photoreceptors for as long as possible.

PRIMA is potentially one of our restorative options. We know that complement inhibitors do not stop the disease entirely; they only modify the velocity of the growth curve. Therefore, a patient treated with intravitreal therapies who eventually suffers complete foveal burnout and profound central vision loss would theoretically then become am eligible candidate for PRIMA implant. That particular individual is just transitioning from one phase of their GA journey (preservation phase) to the next (rehabilitation and restoration phase).

PRIMA received CE mark approval in Europe in July 2026, and the sponsor is pursuing availability in the United States. What questions remain open after 36 months, and what would you most want to see in longer-term or US data before implantation becomes a routine referral option?

Rahimy: With the European CE mark secured, bringing this exciting technology to the United States becomes the crucial next step. Moving forward, as a clinician surgeon, I would want to see two things. First, as the sponsor pursues US availability, I am looking closely at the real-world evidence emerging from commercial rollout in Europe. I want to see how the safety profile and surgical times evolve as community retina specialists there - not just trial investigators - begin performing the procedure. Hopefully the safety profiles mirror what we saw and reported in PRIMAvera.

Second, I want to see the 5-year durability data. The stability of the microarray at 36 months is already quite reassuring, but seeing that the inner retina continues to tolerate the electrical stimulation over a half-decade without signal degradation or issues with the microchip will lend further confidence to an intrigued retina community on this device’s potential to be the definitive standard of care for end-stage geographic atrophy.