Engineering the Eye IV Restoring Vision
29th Symposium: August 22-24, 2014
All talks & discussion sessions are in Goergen Hall
All breaks and lunches are in the Munnerlyn Atrium
Thursday, August 21
7:00 - 9:00 pm—Registration & Welcome Reception, Flaum Eye Institute
Friday, August 22
8:00 am—Registration & Breakfast
Talk session I: Visual deficits (Moderator: Mina Chung)
The most suitable means of restoration of vision will depend upon the nature of the disorder, the state of the retina and the history of visual loss. If profound visual loss occurs in the first year of life, restoration is unlikely unless initiated quickly given the influence of amblyopia. There is also concern that prolonged blindness might be associated with loss of neural organization that would limit potential recovery.
Clinical experience is limited to gene therapy and the use of growth factors.
Gene therapy: If there are viable photoreceptor cells gene therapy may be successful in modulating the metabolic defect and recovery of vision. Whether the mutation causes loss or gain of function will determine the genetic approach. Thus it is important to determine the state of the retina, and the gene causing visual loss and the metabolic influence of the mutation. Experience to-date implies the RPE 65 causes loss of function and treatment induces limited recovery of vision but it is not known how long this will last and it may not alter the rate of photoreceptor loss.
Growth Factors: CNTF reduces the rate of cell loss in several animal models of retina dystrophies. However they cause loss of function and loss of DNA compaction. Human experience shows similar responses. Explantation of the slow release device is followed by recovery of function to the levels of untreated eyes.
Cell replacement therapy using stem or pluripotential cells is being developed for gose with no surviving photoreceptor cells. Other forms of potential therapy have been explored for humans. Optico-electrical devices have given rise to some visual input allowing tasks to be undertaken in those who are totally blind but its usefulness in life is still in doubt. Gene therapy whereby light sensitive channels are expressed in second or third order retinal neurons is being explored to induce photoreception in these cells.
Amblyopia is a developmental disorder of vision that results from abnormal visual experience during an early postnatal sensitive period. Conditions that disrupt binocular vision are most likely to result in amblyopia when left untreated during visual development. However, There are multiple critical periods for visual development. Different visual functions have different developmental time courses. Those with long, late developmental profiles are higher-order perceptual abilities that rely on spatial localization and integration, and are more vulnerable in amblyopia. Neural investigations into amblyopic vision loss show comparatively mild deficits in striate and extrastriate cortex, that are not sufficient to explain the vision loss. We interpret these findings to mean that there is substantial information available in the amblyopic visual system that could potentially be tapped to improve visual function. Although amblyopia has traditionally been regarded as treatable only during early development, recently developed perceptual learning approaches show some promise for vision restoration in adults. We have used a perceptual learning paradigm designed to improve higher-order visual abilities in adult amblyopic macaques and found some improvement on untrained as well as trained tasks. An important open question regarding these approaches is the persistence of the improvement beyond the training period.
10:20 - 10:50 am—Break
Visual performance and the underlying neural processing in the eye and brain vary dramatically across the visual field. Furthermore, foveal and peripheral locations often play very different roles in supporting performance, depending on the specific visual task. Thus, to develop and evaluate methods of restoring vision it is important to understand how neural processing varies across the visual field and how performance in different tasks is related to that processing. In this talk I will first summarize how performance varies across the visual field for various tasks. I will then describe a quantitative model for predicting the detectability of arbitrary targets presented at arbitrary retinal locations in arbitrary gray-scale backgrounds, under photopic viewing conditions. The major factors incorporated into the model include: the optics of the eye, the sampling array and receptive field properties of midget retinal ganglion cells, and the orientation and spatial-frequency tuning properties of neurons in primary visual cortex. This model, based directly on anatomy and physiology, is able to account quite well for threshold measurements across the visual field, for a wide range of targets, in simple (uniform) and complex backgrounds, including natural scenes. I will argue that such models may be of practical value in estimating the expected effect of a vision-restoration treatment, and in designing sensitive behavioral tests for whether the treatment is having the expected effect.
12:10 - 1:10 pm—Lunch
Talk session II: Evaluating Restored Vision (Moderator: Krystel Huxlin)
Retinal prostheses are designed to artificially elicit electrical activity in retinas that have been damaged by disease. Current devices, however, produce limited visual function. Some of the reasons for this can be understood based on the high spatial and temporal resolution and cell-type specificity of visual signals in the retina. These considerations suggest that future devices may need to operate at single-cell, single-spike resolution in order to mediate naturalistic visual function. I will show data indicating that in some cases such resolution is possible. Using multi-electrode recording, visual stimulation, and electrical stimulation in isolated peripheral macaque retina, we show that low-amplitude (1 uA) current pulses delivered through small (10 um) electrodes can reliably trigger a single retinal ganglion cell to fire a single spike with high temporal precision (0.1 ms) and without activating other nearby cells. This precision extends to spatiotemporal patterns of stimulation of small groups of cells, and to the five major cell types in the primate retina. I will discuss the factors that limit spatial and temporal resolution, including axon stimulation and the higher density of cells in the central retina, and the approaches we are taking to address them.
Some approaches to vision restoration respond to degeneration in outer retina by providing visual information directly to surviving neurons with optoelectronics or optogenetics or by replacing degenerated neurons using stem cells. This technique has been shown to be effective in generating visual response in blind rodent by measurements of retinal and visual pathway electrophysiology and by visually driven behavioral responses. Our laboratory is working to add a new approach for evaluating vision restoration by in vivo imaging of the light responses of retinal neurons before and after interventions to restore vision. We use adaptive optics (AO) imaging of calcium responses in retinal neurons in which fluorescent calcium indicators have been placed by viral vector methods. This method offers repeated measures of the light responses of many neurons with resolution to track the responses of individual cells. We are currently deploying this approach to examine optogenetic restoration of function using the light-gated ion channel channelrhodopsin2. Our method is to insert a joint vector that contains channelrhodopsin and the red-shifted calcium indicator R-CaMP into inner retinal neurons in animal models of outer retina degeneration. For mice studies we are using blind triple knockout mice, a transgenic model lacking rod, cone and melanopsin cell function, and for macaques we use focal regions of blindness produced by small laser lesions of photoreceptors and RPE cells. This approach, if successful, can be used to examine which features of cell responses are restored by channelrhodopsin and to predict visual benefit.
2:30 - 3:10 pm—Break
The premise of sensory substitution for blindness is that the “visual brain” , once deprived of its input, will accept information about the environment that is coming from alternative sensory channels. Sensory substitution devices (SSD) for blindness use a video camera to capture visual scenes, and couple them to intact, tactile our auditory interfaces s a surrogate for sight. SSD have been shown to enable some increase in simple visual tasks and navigation ability in both congenitally and acquired blind subjects with varying pathologies. In addition, neuroimaging studies have shown differential activation of visual brain areas between blind and sighted subjects engaging in both haptic and auditory sensory substitution tasks. This lecture will cover an introduction to the types of SSD being developed, methods for outcomes assessments and the types of vision loss for which they are suitable. Our findings on the neural responses to sensory substation use will also be reviewed.
I. Consequences of constraints inherent in current electrical stimulation approaches
Electrical stimulation constrains achievable grain in potential field distribution, both spatially and temporally. Calculations on possible local field gradients reveal marked constraints on local resolution, as will be demonstrated by simulation results. Further, the resolution is, obviously, limited by electrode count. Thus spatial aliasing needs to be addressed: In human eyes, the optical resolution – as constrained by both refractive aberrations and diffraction due to the finite pupil aperture – harmoniously matches the receptor mosaic's resolution. With prosthetic devices, it will be necessary to match optics to stimulation resolution, possibly by purposefully degrading the optics. Finally, temporal aliasing is also expected: rods have their CFF at 20 Hz, cones at 50 Hz, while typical chip sampling currently occurs below 10 Hz. This poses obvious limitations on flicker resolution, but also on motion perception, especially with high-frequency targets, and can lead to strong illusions in real-world scenes.
II. Quantitative approaches to evaluating restored vision
It is currently debated whether to use real-world scenarios for vision assessment (e.g. obstacle courses) or easier-to-standardize schematic approaches (e.g. huge optotypes). I will argue for tailored standardized approaches. At the very low end of vision, the BaLM test (“Basic quantitative assessment of light, motion, etc.”, IOVS, 2010) provides a monotonous measure of visual function starting at Light Perception. It is based on basic visual dimensions, derived from texture segregation, taking into account that there is more to vision than acuity. In the region of Hand Movement and above, BaLM overlaps with the FrACT test (Freiburg Acuity and Contrast Test, http://michaelbach.de/fract/). Together these two tests provide continuous measures of vision from light perception up to normal acuity.
4:30 - 5:30 pm—Overall discussion session on goals for vision restoration, Moderator: Donald MacLeod
5:30 - 7:30 pm—Poster session and dinner, Munnerlyn Atrium
Saturday, August 23
Talk session III: Comparison of available methods (Moderators: Tatiana Pasternak & Jennifer Hunter)
Background/Aims: There exists no cure yet for blindness caused by hereditary retinal degeneration of the photoreceptors (e.g. retinitis pigmentosa (RP) but restoration of vision by various electronic retinal implants has made rapid progress in recent years with remarkable results concerning visual localization of objects, mobility, even reading and face recognition in some cases. There are, in principle, three different approaches (fig. 1, from Zrenner Sci. Transl. Med. 2013): a) Epiretinal implants are positioned on top of the neuroretina, placing electrodes, controlled by a camera outside the body, at the functional output of the retina, i.e. near the RGC fibers; b) subretinal implants that utilize light sensitive photodiodes, each connected to an electrode, both placed beneath the retina, contacting the functional input of the retina on the photoreceptor side, and c) the suprachoroidal approach that inserts electrode arrays from the back of the eye, positioning them on top of the choroid.
The interim results of a prospective mono-& multicenter clinical trial on safety & efficacy of subretinal implants for partial restoration of vision in patients blind from hereditary retinal dystrophies performed since 2010 will be reported here (registered at http://www.clinicaltrials.gov, Registration number: NCT00515814, NCT01024803, NCT01497379). The aim was to assess the 12-month visual and safety outcomes following implantation of a 1500-pixel subretinal implant Alpha IMS (Retina Implant AG Reutlingen) in patients blind from retinitis pigmentosa (RP).
Methods: Alpha-IMS subretinal implants (Retina Implant AG, Reutlingen, Germany) were positioned beneath the foveal region of 14 male and 12 female RP-patients (mean age 53.2 ± 8.2). Each of the 1500 subfoveal photodiodes within a 11° by 11° field controls an amplifier that, depending on the strength of the light, emits currents to stimulate overlying bipolar cells. Power and control signals are supplied inductively via a subdermal, retroauricular coil from which a subdermal cable leads to the eyeball. Function was tested by 4 procedures - 1: Monitor-based standardized tests for light perception, light localization, movement detection, grating acuity and visual acuity with Landolt C-rings (2- or 4 alternative-force-choice-tests); 2: Detection, localization and identification of objects placed on a table; 3: Reading letters; 4: Visual experiences during outdoor and daily-life activities.
Results: 1: Implant-mediated light perception was possible for 22 (85%) patients; light localization for 15 (58%); movement detection (up to 35 cycles/degree) for 6 (23%); measurable grating acuity (up to 3.3 cycles/degree) for 14 (54%); and measurable visual acuity (up to 20/546) for 4 (18%). 2: On a visual ability scale from 0 (worst) to 4 (best) for 4 white geometric figures presented on a black table, patients averaged 3.12 ± 0.31 for detection, 2.94 ± 0.3 for localization and 1.06 ± 0.28 for identification at month 2 which was significantly better (p< 0.012) than with chip power switched off. Similar results were obtained with activities of daily living. Losses of approximately 1-1.5 units occured over 9-12 months. 3: Four patients (18%) could read letters 4-8 cm in size. 4: Twelve patients (46%) reported useful visual experiences including recognition of details and 7 patients (27%) could localize objects in daily life without details. 5: Besides two treatable serious adverse events there were no safety concerns.
Conclusion/Summary: The Alpha-MS implant has meanwhile received a CE mark for commercial use in Europe. Psychophysical testing and self-reported outcomes show restoration of useful vision in a majority of patients. Subretinal implantation surgery is safe and a multicenter study is continuing with a slightly modified implant built for long term durability. It has turned out that the foveal localization of implants is of particular importance, concerning optimization of visual cognitive abilities in these patients.
Three patients with bare light perception from long-term retinitis pigmentosa were implanted with an electrode array in the suprachoroidal space, connected to an external stimulator via a percutaneous connector.
Thresholds and other characteristics of artificial visual percepts or phosphenes were measured over an 18 month period to produce phosphene maps for each patient. Reasonably stable phosphenes were obtained with all three patients, within the safe range for charge-balanced biphasic stimulation. The lowest thresholds (and therefore the largest dynamic ranges) occurred for monopolar stimulation with the anodic phase first and for relatively high pulse rates. Phosphenes varied in shape and brightness as a function of electrode position, and there were complex temporal patterns including onset and offset flashes as well as persistent percepts lasting several seconds. The patients have used interleaved stimulation of phosphenes to perform visual tasks successfully with the laboratory vision processor and with a semi-portable vision processor for improved mobility.
The results provide proof of concept that a suprachoroidal electrode array can provide potentially useful visual percepts for blind patients, and justification for further investigation and development of a wearable vision processor.
10:20 - 10:50 am—Break
Epiretinal implant patients can use spatial information from the stimulator to detect motion and report the orientation of lines. The ability to perceive forms has been inconsistent in the implant patients. Form vision requires the ability to combine percepts from individual electrodes into shapes. A first step is limiting the size of each individual percept to an area near the electrode. We explored the shape of retinal responses created by single and multi-electrode stimulation, in both animal models and human implant patients. In vitro retinal responses to epiretinal stimulation were imaged using genetically encoded calcium indicators. Using calcium imaging to record cell activity allowed simultaneous visualization of multiple retinal ganglion cells. Biphasic pulses in the range typically used for retinal stimulation (0.5-2 ms) resulted in elongated areas of activation, suggesting activation of axons passing the electrode. Stimulation with longer pulses, including 20 Hz sine waves and square waves, result in stimulation focused around electrode, suggesting that axonal stimulation is avoided with these types of stimuli. Patient reports are consistent with the in vitro results. An epiretinal prosthesis patient reported elongated percepts with 1 ms/phase, biphasic pulses but round percepts when a 20 Hz sinewave was applied to the same electrode. These results are consistent with other studies that have examined axonal stimulation. Long pulse width stimuli can be used to avoid axonal stimulation with epiretinal electrodes. The use of calcium imaging to visualize retinal responses is a powerful tool for investigating creation of complex response patterns in the retina.
It has been hypothesized that a vision prosthesis capable of evoking useful visual perceptions can be based upon electrically stimulating the primary visual cortex (V1) of a blind human subject via penetrating microelectrode arrays. We examined several spatial and temporal characteristics of microstimulation using an array of 100 penetrating microelectrodes chronically implanted in V1 of a behaving macaque monkey. Acutely, we were able to evoke visual percepts using microstimulation via single electrodes. After two years implantation we were able to evoke visual percepts via microstimulation across the spatial extent of the array using groups of contiguous electrodes. Consistent responses to stimulation were evoked at an average threshold current per electrode of 204 ± 49 µA (mean ± std) for groups of four electrodes and 91 ± 25 µA for groups of nine electrodes. Saccades to electrically-evoked targets using groups of nine electrodes showed that the animal could discriminate spatially distinct percepts with groups having an average separation of 1.6 ± 0.3 mm (mean ± std) in cortex and 1.0 ± 0.2 degrees in visual space. These results demonstrate chronic perceptual functionality and provide evidence for the feasibility of a cortically-based vision prosthesis for the blind using penetrating microelectrodes. Future work will focus on directly comparing cortical stimulation using intra-cortical and epi-cortical electrodes, and undertaking a feasibility trial in profoundly blind human patients.
12:10 - 1:10 pm—Lunch
Converting inner retinal neurons to photosensitive cells by expressing genetically encoded light channels offers a novel approach for treating blindness caused by retinal degenerative diseases. As a new strategy for treating blindness caused by retinal degeneration, a number of studies reported the feasibility of restoring light sensitivity to photoreceptor-deficient retinas in rodents by expressing channelopsin-2 and other light gated channels in inner retinal neurons. Rodent studies have demonstrated the long-term stability and safety of the expression of ChR2 in inner retinal neurons and the restoration of vision-driven behavior. These studies used recombinant adeno-associated virus (AAV) vectors, to deliver the cDNA encoding the light gated channels to the retina. AAV-mediated gene delivery through subretinal injection has been successfully used to target transgenes to photoreceptor and retinal pigment epithelial cells both in nonhuman primates and humans. However, for restoring retinal light sensitivity after photoreceptor degeneration, it is necessary to target the surviving inner retinal neurons. Studies in rodents have shown that the intravitreal injection of AAV2 driven by ubiquitous promoters can mediate the robust expression of ChR2 in the inner retinal neurons. However, there have been a limited number of detailed studies of AAV-mediated transgene expression in inner retinal neurons in nonhuman primates through intravitreal injection. Though intravitreal injections present benefits from the surgical standpoint, they lead to transduction of a limited zone in the primates and evidence suggests they are more likely to lead to immune response. Therefore the evaluation of the AAV-mediated microbial opsin delivery in the retina of nonhuman primates through intravitreal administration would be an important step in developing ChR2-based gene therapy in humans. We examined the efficacy, safety and functionality of the AAV-mediated expression a ChR2 variant using AAV in the retina of macaques. The preliminary results of this translational study will be discussed from the point of view of light intensity requirements, immunological safety and efficacy.
Advances in stem cell technologies are now enabling production of a variety of cell types that are beneficial for vision research and the development of therapeutics. Human pluripotent stem cells, which can generate all cell types in the body, produce cell types present in the eye, including retinal neurons, photoreceptors and retinal pigment epithelial (RPE) cells. These in vitro products are being used to model disease or be a source of healthy cells to use in cell transplantation strategies. The recent induced pluripotent stem cell (iPSC) approach enables production of cells from an individual, and the generation of cell products that are genetically matched to the patient. Currently, allogeneic RPE transplantations are leading the clinical application of pluripotent stem cells, and results are eagerly awaited. In addition, several adult or tissue specific therapies are being developed that use the adult RPE stem cell, fetal neural stem cells or umbilical stem cells, to replace or substitute for retinal cell loss. In the future, autologous treatment using iPSC products and RPE stem cells is an exciting prospect, although personalized therapy for a single individual raises challenges such as ensuring patient safety and to the economy of healthcare delivery, which will be discussed.
2:30 - 3:10 pm—Break
Together with rodent and other models, the dog has played a pivotal role in developing and testing different gene therapy approaches for translation to the clinic. Success has been achieved in RPE65/LCA, CNGB3/achromatopsia, RPGR/XLRP and VMD2/Best disease, and these promising outcomes raise several issues and opportunities for future translational studies.
Among these are:
- gene therapy of RPE65/LCA patients and dogs treated during the degeneration phase indicate that there is stable functional recovery even though progressive degeneration of photoreceptors continues. It is important to determine if this is common to other retinal diseases, and if combinatorial therapies directed both at correcting the underlying gene defect and promoting photoreceptor survival are effective.
- in CNGB3/achromatopsia we have shown that the therapeutic transgene and protein can be expressed in older mutant cones, yet functional recovery fails. However, transient 'deconstruction' of the cone photoreceptors, either immediately prior to or after gene therapy, results in proper assembly of the cGMP-gated channel subunits in outer segments, and stable recovery of function. Approaches to modifying the target retinal cells should be considered, if necessary, as a means of improving functional and structural outcomes.
- in both RPGR/XLRP and VMD2/Best, treatment at early, mid and late disease stages arrests further disease progression. In the former, normal photoreceptor structure is restored and post-synaptic remodeling is reversed. In VMD2/Best, there is reversal of large psudohypopyon lesions with treatment, and the photoreceptor/RPE interface becomes normal.
- general issues for optimizing translational applications will require modifying vector pseudotypes/promoters to enhance high efficiency transduction of outer retinal cells when delivered intravitreally, and minimize trauma of fragile, diseased photoreceptors. As well, assessing promoter or promoter combinations that are effective both in degenerating cells, and once gene therapy correction has reversed the disease, will be important to ensure long-term stability of correction.
4:50 - 6:15 pm—Poster session
6:30 - 8:30 pm—Dinner, Colgate Divinity School
Sunday, August 24
Talk session IV: Central Visual Pathways (Moderator: David Williams)
Retinal ganglion cells (RGCs) transmit the only information that the brain receives about the visual world. When photoreceptors die as a result of inherited retinal degenerative diseases, the synaptic circuitry between photoreceptors and RGCs undergoes substantial morphological and functional alterations. Any visual system stimulation-based vision rescue strategy must overcome the challenge imposed by the functionally altered retinal circuitry. I will discuss research on the physiological properties of RGCs in the rd1 mouse model of retinal degeneration in the context of the limitations imposed on signal transmission from eye to brain by ongoing RGC activity. After photoreceptor death, RGCs exhibit continuous aberrant burst firing at ~10 bursts per second. Evidence from many labs now suggests that this activity is driven by oscillatory synaptic input arising from depolarization of A2 amacrine / ON bipolar cell networks. Using in vivo optical recording methods, we have begun to investigate the effects of aberrant RGC firing on downstream visual system function, including ongoing activity in primary visual cortex (V1). One of our goals is to track changes in functional properties before and after photoreceptor death using an inducible model system. The results will provide fundamental information about the functional state of neural circuits in the early visual system, information that will be critical for the rational design of vision rescue strategies.
Visual cortex contains multiple topographic maps of the visual field, with a disproportionately large region devoted to processing central vision. Therefore, overlapping lesions in the central retina of both eyes will deprive a relatively large region of cortex from feedforward inputs. When central retinal lesions are present at birth, the region of deprived cortex can be reallocated to represent more peripheral parts of the visual field. However, when lesions occur later in life, e.g. through macular degeneration, such large-scale visual remapping is not apparent and the region of deprived cortex remains largely unresponsive to feedforward input. Moreover, long-term retinal deficits can lead to a volumetric reduction in structures along the visual pathways, suggesting anterograde degeneration occurs in the absence of visual input. In contrast, there is also evidence that deprived cortex can respond in some patients under certain conditions, e.g. when a visually relevant task is performed, possibly mediated through cortical feedback connections. Furthermore, numerous studies report that deprived visual cortex can be recruited in other sensory tasks (somatosensory, auditory), which may also depend on the strengthening of existing crossmodal connections. In spite of studies indicating possible structural and functional changes in visual cortex, when retinal inputs are restored in patients with wet age-related macular degeneration, occipital cortex can resume visually driven activity in parallel with clinical measures indicating recovery of visual function. This lends reassurance that topographic maps in adult visual cortex are robust and can remain viable to process restored inputs successfully.
10:20 - 10:50 am—Break
Until recently, patients with Retinitis Pigmentosa (RP) faced an inevitable decline in vision, often resulting in total blindness with no hope of treatment. Now, multiple groups worldwide are developing and testing retinal prostheses – “bionic eyes” that can restore some vision. The Argus II Retinal Prosthesis System was the first prosthesis to be commercially approved in Europe, and the only system approved for sale in the United States.
The implanted portion of the System includes a receiving antenna and an electronics case implanted on and in the eye, and a 6x10 electrode array that is tacked over the macula epiretinally. A glasses-mounted camera collects visual information and sends it to a small processing unit, which down-samples and processes the image. Data and power are sent wirelessly from a transmitting antenna on the glasses to the internal receiving antenna. The electrodes in the array emit pulses of electricity whose amplitude correspond to the brightness of the scene in that location. Stimulation of the remaining retinal cells induce cellular responses that travel through the upstream visual system, resulting in visual percepts.
The Argus II has been implanted in almost 80 patients, including 30 in the clinical trial and more than 45 in the post-approval setting. We have developed an eligibility assessment for Argus II patients, including the evaluation of patient motivation and expectations. Once implanted, patients undergo a unique rehabilitation program in order to learn to use their restored vision to enhance their functional vision and well-being. These factors are critical to ensure the best outcome for each patient.
1:00 pm—Closing Lunch at the home of David and Inger Williams