gms | German Medical Science

Objective: Subretinal implants (PRIMA) in patients with atrophic AMD provided letter acuity of 20/420, matching its 100 μm pixel pitch. The formed vision provided by these implants, unlike epiretinal prostheses, might be attributed to preservation of many aspects of the retinal signal processing in the network-mediated activation. Direct stimulation of the ganglion cells would result in indiscriminate activation of the ON and OFF pathways, causing conflicting signals in the higher visual processing centers. To assess selectivity of the bipolar cells’ stimulation vs. ganglion cells, we measured their stimulation thresholds with PRIMA implants and the next-generation devices having 20 μm monopolar pixels.

Materials and Methods: Photovoltaic arrays were implanted subretinally in Long Evans rats, leading to degeneration of the photoreceptors above the device. The implants were activated by 880 nm laser at pulse durations varying from 0.5 to 10 ms, and a digital micromirror display was used for pattern formation. Visually evoked potentials were measured with and without the intravitreal application of synaptic blockers.

Results: Strength-duration curves with the PRIMA implant (100 μm bipolar pixels) demonstrated 6-fold difference between the stimulation thresholds of the retinal ganglion cells vs. bipolar cells: rheobase without blockers was 0.2mW/mm², but with blockers –1.34mW/mm². Chronaxie without blockers was 2.8 ms, and with –0.9 ms, consistent with the previous retinal recordings based on patch clamping. Rheobase with full-field stimulation using monopolar 20 μm arrays was 0.01mW/mm² with chronaxie of 3.6 ms.

Conclusions: The 6-fold difference between the stimulation thresholds of bipolar and ganglion cells provides a wide dynamic range for selective network-mediated retinal stimulation with PRIMA implants, which preserves many aspects of the retinal signal processing. The next-generation implants with smaller 3-dimensional pixels should be designed to match such selectivity.