Testing of Semichronically Implanted Retinal Prosthesis by Suprachoroidal-Transretinal Stimulation in Patients with Retinitis Pigmentosa

Abstract

Purpose.: To examine the safety and effectiveness of a retinal prosthesis that is implanted semichronically in two patients with advanced retinitis pigmentosa (RP).

Methods.: Two eyes of two patients with advanced RP had a retinal prosthesis implanted in a sclera pocket of one eye. The visual acuity of both eyes before the implantation was bare light perception. Phosphenes were elicited by suprachoroidal-transretinal stimulation (STS). The internal devices of the STS were implanted under the skin on the temporal side of the head, and the 49 electrode-array was implanted in the scleral pocket of one eye. Biphasic electrical pulses (duration, 0.5 ms; frequency, 20 Hz) were delivered through nine active electrodes. The threshold current was determined by currents ≤1 mA. Behavioral tasks were used to determine the functioning of the prosthesis.

Results.: The surgery was completed without a retinal detachment and retinal/vitreous hemorrhage. The implanted STS system remained functional for the 4-week test period. Phosphenes were elicited by currents delivered through six electrodes in Patient 1 and through four electrodes in Patient 2. The success of discriminating two bars was better than the chance level in both patients. In Patient 2, the success of a grasping task was better than the chance level, and the success rate of identifying a white bar on a touch panel increased with repeated testing.

Conclusions.: Semichronic implantation of a microelectrode-STS system showed that it was safe and remained functional for at least 4 weeks in two patients with advanced RP. (www.umin.ac.jp/ctr number, R000002690.)

Retinitis pigmentosa (RP) is one of the leading causes of blindness in developed countries and is characterized by a progressive degeneration of the photoreceptors. ^1,2 To restore some vision to these patients, stimulating the residual functional retinal neurons by electrical currents delivered through a retinal prosthesis is being extensively studied. ^{3 –5}  

Various types of retinal or optic nerve prosthesis have been developed, and these have been tested in animals ^{6 –14} and patients. ^{15 –20} A typical retinal prosthesis consists of an array of electrodes that is implanted on or beneath the retina, and it is used to deliver electrical current to the retina to stimulate functioning retinal neurons to send signals to the visual cortex, where they are perceived as light sensations called phosphenes. 

We have developed a new approach for stimulating the retina called suprachoroidal-transretinal stimulation (STS). ^9,21 In this method, the retinal prosthesis is placed in a scleral pocket, and the reference electrode is in the vitreous cavity. Although the distance between electrode array and the retina is not close compared with other types of retinal prosthesis, the transretinal currents can stimulate the retinal neurons effectively, and the threshold current to evoke electrically potentials from the visual cortex by the STS is comparable to that by other electrodes in animals. ²¹  

The ability of patients to discriminate objects visually with chronically implanted retinal prosthesis has been reported by several groups. Thus, Humayun's group reported that patients can recognize simple shapes with a 16-channel epiretinal electrode system. ¹⁸ More recently, the same group developed a chronically implantable retinal prosthesis made up of 60 electrodes. With this system, patients were able to recognize simple words (Humayun MS. IOVS 2010;51:ARVO E-abstract 2022). Zrenner's group implanted a 1500 channel electrode array subretinally, and the patient was able to recognize simple words or Landolt's Cs. ^19,20  

Although, the resolution of the image might be lower with STS prosthesis because the electrodes are some distance from the retina, the advantage of the STS prosthesis over the epi- or subretinal prosthesis is the safety of the surgical procedures because the electrodes do not touch the retina and are stably fixed in the scleral pocket. Based on the safety of this approach, the STS system has been adopted by several other groups. ^22,23 In an acute clinical trial with STS prosthesis, localized phosphenes were perceived with safety currents (≤1 mA) in two patients with advanced RP. ²⁴  

We have developed a microelectrode-STS system that can be chronically implanted, and the purpose of this study was to determine its safety and stability when it is implanted semichronically in patients with advanced RP. 

Patients and Methods

Retinitis Pigmentosa Patients

Two patients with RP were studied (Table 1). The diagnosis of RP was made by independent ophthalmological and eletroretinography (ERG) examinations. Patient 1 (Pt 1) was a 73-year-old woman who has had night blindness since the age of 55 years when she was diagnosed with RP. Her visual acuity decreased to hand motion in both eyes at the age of 68 years, and she had bare light perception (LP) in both eyes at 72 years at the time of these experiments. Transcorneal electrical stimulation (TES) ²⁵ elicited phosphenes that were perceived in the central visual field with a threshold current of 0.80 mA (pulse duration, 10 ms) in the right eye and 0.65 mA (pulse duration, 10 ms) in the left eye. The area of the phosphenes increased with an increase of the stimulating current in the left eye. 

Table 1.

 

Patients for the Retinal Prosthesis

Patient	Age (y)	Sex	Diagnosis	Visual Acuity (Right/Left)	Years with Lowest Visual Acuity
1	72	F	RP	LP/LP	1 y
2	67	F	RP	LP/LP	17 y

 F, female; RP, retinitis pigmentosa; LP, light perception.

Patient 2 (Pt 2) was a 67-year-old woman who has had night blindness since the age of 10 years and was diagnosed with RP at 26 years. Her visual acuity decreased to hand motion in both eyes at age 50 years and was bare LP in both eyes at the time of the surgery. TES elicited phosphenes that were perceived in the central visual field with a threshold current of 1.0 mA in the left eye and was not evoked in the right eye even with a current of 2.0 mA. The area of the phosphene did not increase with an increase in the stimulating current in the left eye. 

A full explanation of the purpose of this study and the procedures to be used were given to each patient, and each signed an informed consent form. They were also instructed that they were free to withdraw at any time. The procedures used in this study adhered to the Declaration of Helsinki and were approved by the Ethics Committee of Osaka University Hospital. 

Implant

The implanted electronic devices consisted of a secondary coil that receives signals from the external coil and a decoder that generates biphasic pulses to deliver to the individual electrodes sequentially (Figs. 1A, 1B). The electrode array (size, 5.7 × 4.6 mm; Nidek, Gamabori, Japan[b]) consisted of 49 electrodes made of 0.5-mm–diameter platinum wire, and the center-to-center separation of a pair of electrode was 0.7 mm (Fig. 2C). Each electrode protruded from the silicon base by 0.5 mm (Fig. 2D). The return electrode was a 0.5-mm–diameter, 6-mm–long platinum wire that was insulated except for 3 mm of the tip (Fig. 2E, 2F). 

Figure 1.

 

Diagram of retinal prosthesis system. (A) Lateral view of the skull XP of Pt 1 after implantation surgery. (A) Position of skin incision to insert and anchor the device. (B) Position of skin incision to fix the cable to the bone of the lateral orbital wall. (C) Return electrode. (D) Stimulating electrode. (E) Decoder. (F) Secondary coil. (B) The implanted devices, cable, and electrodes. Scale bar, 3 cm.

Figure 2.

 

Photograph of 49-channel stimulating electrode (A–D) and return electrode (A, B, E, F). Top (A) and side (B) views of stimulating and return electrode protected by protective silicone cover. Top (C) and side (D) views of 49-channel stimulating electrode. The diameter of each electrode is 0.5 mm, and the center-to-center electrode distance is 0.7 mm. The inset in (D) shows the protective cover that reinforced the junction between the electrode array and the cable. Top (E) and side (F) views of return electrode. The diameter of return electrode is 0.5 mm. Scale bars in (A–F), 5 mm.

Surgical Procedures

The subjective vision was not different between the right and left eyes in both patients. 

The left eye was selected for the implantation in both patients because the threshold current to elicit phosphenes by TES was lower in the left eye than in the right eye. 

Under local anesthesia, the lateral rectus muscle was dissected at its insertion, and transscleral monopolar stimuli were given to determine the scleral area that consistently evoked low-threshold phosphenes. ²⁴ After identifying and marking the low-threshold area, the patient was placed under general anesthesia. The area identified from the monopolar stimulation was relatively large and was posterior to the insertion of the inferior oblique (IO) muscle in Pt 1 and was very restricted to approximately 2 mm posterior to the insertion of the IO muscle at around 3 o'clock in Pt 2. 

The skin over the left temporal bone was incised to insert the electronic devices (Fig. 1A). A second skin incision was made over the left zygomatic bone to fix the cable (see also Besch ²⁶ ; Fig. 1B). The electrode array and the return electrode were protected with a silicone cover (Fig. 2A, 2B) and passed under the fascia of the temporal muscle from the first incision to the second incision through a trocar catheter (Medikit, Tokyo, Japan). 

The bone of the lateral orbital wall was drilled, and the electrode array, return electrode, and cable were passed into the periocular space using the trocar catheter. The cable with its protective cover was fixed by a titanium plate below the second incision. The electrode array and cable were circled around the equator passing under the four recti muscles. 

A scleral pocket of 6 × 5 mm was made at the temporal to lower-temporal scleral area where the phosphenes were elicited. A 49-electrode array was placed in the scleral pocket (Fig. 3) and secured with sutures that passed through the protective silicone cover around the junction of the electrode array and the cable (Fig. 2D). The return electrode was inserted into the vitreous cavity through the upper nasal pars plana area. 

Figure 3.

 

Photographs of surgical procedure to insert the electrode-array. (A) Creating a scleral pocket. (B) Holding the electrode array. (C) Grasping the electrode array for insertion. (D) Inserting the electrode array into the scleral pocket.

After suturing the conjuntival incision, the electronic device was fixed to the temporal bone, and the skin was sutured. At the end of the implant procedure, the system was tested to be certain that all electrodes were functioning. Five (Pt 1) to 7 (Pt 2) weeks after the implantation, the device and wire were surgically removed. 

Functional Testing of Each Electrode

From week one after the surgery, the wireless system was tested twice a week for 4 weeks. For the functional test of each electrode, 9 out of the 49 electrodes were tested as shown in Figure 4B. The distance between adjacent active electrode was 2.1 mm. An electronic stimulator was designed to deliver charge-balanced biphasic pulses to individual electrodes sequentially with a delay of 0.45 msec (Fig. 5). Cathodic-first biphasic pulses (duration, 0.5 ms; frequency, 20 Hz; interpulse delay, 0.5 msec; number of pulses, 20) were delivered through the selected channel or combination of multiple channels. The stimulating parameters were chosen based on the acute clinical experiment, which also used STS. ²⁴  

Figure 4.

 

Position of the stimulating electrode. (A) Anterior-posterior view of the skull XP of Pt 1 after the surgery. (B) A magnified view of (A). The insets show the position of the nine active electrodes. (C) The presumed position of the none active electrodes superimposed on the left fundus of Pt 1.

Figure 5.

 

Time sequence of stimulating current pulses. The first pulse was a cathodic current, and the second pulse was an anodic current to balance the charge. A pair of pulses was delivered sequentially from channel 1 to channel 9 electrodes.

The current was applied for 0.5 second after a conditioning buzzer signal. The threshold current that elicited a phosphene was determined by increasing the current intensity from 0.1 mA in 0.1 mA steps until the patients were able to recognize and localize the phosphenes correctly in >50% of the trials. 

To identify the position of the phosphene, a plastic board (65 × 65 cm) was set in front of a patient at a distance of 40 cm. The patient was instructed to put her right index finger on the position of the perceived phosphene while the left index finger was positioned on the pad glued to the center of the board (Fig. 6B). For safety, the maximum current was 1.0 mA. ²¹ The procedure was repeated with changes in the current to determine the threshold current. Care was taken not to influence the response of the patients. The experiments to map the perceived phophenes were repeated on different days. We also tested the effect of simultaneous activation of two electrodes. 

Figure 6.

 

(A) Map of the perceived phosphenes in response to the stimulation of individual electrodes. The estimated position of the phosphenes when each electrode is stimulated and normal topographical organization exists between the retina and visual cortex. (B) Method to record the position of the phosphene in relation to the center of the body. The left index finger is positioned at the center of the board, and the right index finger is placed at the position of the perceived phosphene. (C) The phosphene maps of Pt 1. The results of multiple trials are superimposed. The red, orange, green, dark blue, green, purple, and white X's indicate the position of the perceived phosphene in response to the stimulation of Chs 2, 3, 5, 6, 7, and 8 individually. The colored circles indicate the gravitational center of the responses to the stimulation of the individual channels. The bars indicate the standard deviations. (D) The phosphene map of Pt 2. The brown, red, orange, green, and purple X's indicate the position of perceived phosphene in response to the stimulation of Chs 1, 2, 3, and 7.

Functional Testing with Video Camera

For these experiments, the patients performed visual tasks using a commercial video camera as the detector of a visual object (QVR-13; Logitech, Tokyo, Japan). The camera was attached to a headband, and an eye mask was placed over the both eyes during the testing. Because the camera's field of view was approximately 16.7° of visual angle and the implant covered 14.3°, the visual angle subtended by an object on the retina was reduced by a factor 1.2. ¹⁸  

The object viewed by the camera was converted to a 3 × 3 square with 40 × 40 pixels, and if the light level was above the threshold, the square was expressed as white (on), and if the light level was below the threshold, the square was expressed as black (off). The information of the square was converted to an electronic signal and sent to the secondary coil through the external coil. The activated electrodes were channels (Chs) 2 to 8 (seven electrodes) in Pt 1, and Chs 1, 2, 3, 4, and 7 (five electrodes) in Pt 2. 

All tests were carried out with the patients sitting on a chair and a plastic board covered with black cloths set 40 cm from the patients. The white target was presented against a black background under regular room lightning. Head movements were allowed during all experiments except in Experiment 3. 

To eliminate the possibility that the patients reacted to clues other than the visual stimuli, e.g., acoustic stimuli, we performed the experiment with the electrical stimulator off but the buzzer on in each experiment. The sequence of presentation was randomized. 

Experiment 1: Object Detection

A white box (chopsticks box) that was 2.6 cm × 27 cm (3.7° × 34° visual angle) was set randomly at 15 cm (21°) to the left or right of the center of the board. The patients were asked where the white box was located. 

Experiment 2: Object Discrimination

Two white bars of different widths, 1 cm × 30 cm (1.4°× 37°) and 3 cm × 30 cm (4.3° × 37°), were presented at the center of the board, and patients were asked to tell the examiner whether the thicker bar was on the left or right. 

Experiment 3: Detection of Direction of Motion

Patients were asked to keep their head stationary. The rectangular white box (chopsticks box) was placed in front of the patients and was moved horizontally or vertically with speed 2 to 3 cm/s. The patients were asked to tell whether the bar moved horizontally or vertically. 

Experiment 4: Grasping Objects

A white object was set randomly either 15 cm (21°) to the left or 15 cm to the right of the center of the board. The patient was asked to grasp the object with her right hand. 

Experiment 5: Touch Panel

A white rectangular bar of 4.7 cm × 20 cm (6.7 × 27°) was presented randomly either 9.5 cm (13°) to the left or right from the center of a touch panel screen (Tyco Electronics, Menlo Park, CA) that was connected to the computer. The patient was asked to touch the white bar with her right index finger. The position touched was recorded and analyzed by the computer. Depending on whether the patient touched the correct position, a different sound was emitted by the computer. 

Statistical Analyses

The percentage of correct answers on each task was analyzed statistically by the binominal test, and the criterion for statistical significance was 0.05. We tested whether each patient's performance was better than the chance level (50%) on each task. These analyses were performed with commercially available software (JMP 8.0; SAS Institute, Cary, NC). 

Results

Surgical Results

After surgery, it was confirmed that the device, cables, and electrodes were implanted and connected as judged by the skull x-ray projections (XPs; Fig. 1). An anterior-posterior view of the XPs of the skull showed that the electrode array was positioned at the lower temporal ocular area, and the return electrode was positioned at the upper nasal area in both patients. This was consistent with the intraoperative placements (Fig. 4). 

From the fundus picture, fluorescein angiograms, and OCT images, neither retinal detachment nor hemorrhage was observed after both surgical procedures in both patients. The visual acuity remained at light perception after the removal of the device in both patients. Eye movements were slightly restricted in all directions after the initial surgery in both patients but recovered in 4 weeks. The eye movements remained normal after the second surgery. The connection between the device and electrodes remained functional during the 4 weeks of testing. 

Functional Testing of Each Electrode

Delivering electrical pulses from any one of the nine-electrode array elicited localized phosphenes, which were reproducible for each of the six channels (Chs 2, 3, 5, 6, 7, and 8) in Pt 1 and in four channels (Chs 1, 2, 3, and 7) in Pt 2 with current ≤1 mA (Table 2). The size of the phosphene varied from the size of a pea to a quarter coin at an arm's length distance depending on the channel stimulated in both patients. Two distinct phosphenes were perceived when the stimuli were delivered through two channels (No. 2-3, 2-8, 3-8) in Pt 1 (Fig. 6) but not in Pt 2. 

Table 2.

 

Threshold Stimulus Current to Evoke Electrical Phosphene

Electrode	Patient 1 (mA)	Patient 2 (mA)
Ch1	—	0.90
Ch2	0.35	0.80
Ch3	0.50	0.90
Ch4	—	—
Ch5	0.70	—
Ch6	0.60	—
Ch7	0.60	0.70
Ch8	0.60	—
Ch9	—	—

The phosphenes were perceived mostly in the upper nasal field, which is consistent with the position of the stimulating electrodes in the inferior temporal quadrant (Fig. 6). The gravitational center and the standard deviation (SD) of the perceived phosphene was plotted for both patients (Fig. 5). The median value of the SD in the horizontal and vertical directions was 13.9° and 7.7 ° in Pt 1 and 2.9° and 2.5° in Pt 2, respectively. The topographical correspondence between the gravitational center of the perceived phophene and each electrode was not always consistent in both patients. The relative position of the phosphenes evoked by simultaneously activating two electrodes was almost consistent with the position and distance of electrodes in Pt 1 (Fig. 7). 

Figure 7.

 

The position of the perceived phosphenes in response to simultaneous activation of two electrodes in Pt 1 (single trial). Circles show the position of phophenes in response to simultaneous stimulation of Chs 3 and 8, triangles to Ch 2 and 3, and squares to Chs 2 and 8.

Functional Testing Using a Video Camera

Both patients scored better than chance in the object detection and object discrimination tasks with head scanning. Pt 2 scored 90% better than chance in detecting the direction of motion task, but Pt 1 scored 60%, which was not significantly better than chance. 

The task of grasping objects was carried out by Pt 2 because the elicited phosphene was located close to the subjective center. The score (90%) was significantly better than chance (Fig. 8). 

Figure 8.

 

Success rate of behavioral tasks. The detection of objects was tested by 20 trials in Pt 1 and by 30 trials in Pt 2. The discrimination of object task and detection of direction were tested by 10 trials in both patients. Grasping task in Pt 2 was tested by 20 trials.

The success rate of behavioral tasks with the electrical stimulator off was less than the chance level in each task for both patients. 

The touch panel task was also applied to only Pt 2. The subjective phosphene was perceived shifted slightly to the right of the bar when presented on the right side and shifted to the left of the bar when presented on the left side. The success rate increased with repeated testing (Fig. 9). 

Figure 9.

 

Results of the touching the panel task in Pt 2. (A) The touched positions when the white bar was presented on the left side. (B) The touched positions when the white bar was presented on the right side. (C) The superimposed results of (A) and (B). (D) The success rate after repetition of examination. The dashed rectangular area in (A–C) represents the position of the white bar. Blue dots: first trial; pink dots: second trial; green dots: third trial.

Discussion

We implanted a retinal prosthesis in a sclera pocket of two patients with advanced RP using surgical procedures developed in experiments on dogs (Morimoto T. IOVS 2010;51: ARVO E-abstract 3023) and cadaver eyes. Neither a retinal detachment nor retinal bleeding was observed in both patients after the surgical procedures, confirming the safety of our surgical methods. 

The connection between the internal device and the electrodes remained functioning during the 4-week testing period, indicating that the system is able to withstand the surgical manipulations and the continuous eye movements. The silicone cover of the electrode array and the return electrode helped protect the tip of the electrodes (Fig. 2A, 2B). The silicone cover at the junction between the electrode array and the cable may have also protected the wire from being disconnected during eye movements (Fig. 2D). 

Eye movements were slightly restricted in all direction just after the surgery in both patients, suggesting the circumferential fixation of the cable may have affected the eye movements, which is similar to that after scleral buckling procedures. The restriction was reduced 2 weeks after surgery in both patients. A transient restriction of eye movements was also reported after an implantation of a subretinal prosthesis. ²⁶  

The position of electrode array of the STS system could not be identified directly because the electrodes were inserted in the scleral pocket and could not be observed by ophthalmoscopic examinations. However, the XP image identified the position of electrode array relative to the globe because the connecting cable was circled around the equator of the globe (Fig. 4). Gekeler et al. ²⁷ used computed tomography for identifying the subretinal implants. 

The number of electrodes that evoked phosphenes was greater in Pt 1 than in Pt 2, suggesting that more retinal neurons were preserved in Pt 1 than in Pt 2. This suggestion was supported by the fact that the threshold current determined by TES was lower in Pt 1 than in Pt 2, and the duration of the vision loss was longer in Pt 2 than in Pt 1 (Table 2). The better preservation of retinal neurons in Pt 1 is also supported by the observation that the area that could elicit a phosphene was much larger in Pt 1 than in Pt 2 during monopolar extraocular stimulation during surgery. 

The position of the perceived phophenes was at the upper-nasal visual field, which is consistent with the implantation of the electrode array in the lower-temporal quadrant of the eye (Fig. 6). In Pt 2 the phophenes elicited by stimulating electrodes were located around the subjective center of the patient, suggesting that the active electrode was situated at the scleral area close to the fovea. The position of phophene was scattered in Pt 1 but relatively concentrated in Pt2. The reason for this might be that the electrodes were positioned a slight distance from the fovea in Pt 1 and close to the fovea in Pt 2. This may also account for the better repeatability in Pt 2 than Pt 1. 

The position of phosphenes evoked by activating two-electrodes was consistent with the position of electrodes in Pt 1 (Fig. 7), suggesting that the localized excitation of retinal neurons was achieved by STS in this patient. 

Functional testing using the CCD camera revealed that the detection and discrimination of objects were possible by head scanning with a small number of active electrodes (Fig. 8), which is consistent with the findings of epiretinal stimulation. ¹⁸ The reaching and grasping task was possible only in Pt 2, in whom the electrodes were situated close to the fovea. In the first trial, the touched position tended to shift to an area ipsilateral to the position of target. Because the patient moved her head to identify the target position and a delay of 0.1 second existed between imaging the scenery by the CCD camera and stimulating the electrode, the patient might have shifted the position of the perceived phosphene lateral to the target (Fig. 9). The success rate of the touching the panel increased after repeated testing, suggesting that a training effect may have occurred during the testing (Fig. 9D). 

In summary, semichronic implantation of the electrode array–STS system showed that our approach for a retinal prosthesis is safe and feasible for artificial vision. Further improvements are necessary to achieve reading ability, and this may require increasing the number of functional electrodes. 

Footnotes

 Supported by Health Sciences Research Grants (H19-sensory-001) from the Ministry of Health, Labor and Welfare, and by the Strategic Research Program for Brain Sciences from the Ministry of Education, Culture, Sports, Science and Technology, Japan.

Footnotes

 Disclosure: T. Fujikado, None; M. Kamei, None; H. Sakaguchi, None; H. Kanda, None; T. Morimoto, None; Y. Ikuno, None; K. Nishida, None; H. Kishima, None; T. Maruo, None; K. Konoma, Nidek Company (F, I, E), P; M. Ozawa, Nidek Company (F, I, E), P; K. Nishida, None

The authors thank Yasuo Tano, Masahito Ohji, Hajime Sawai, Mineo Kondo, Tomomitsu Miyoshi, and Jun Ohta for advice and discussion and Koji Oosawa, Kenzo Shodo, Motohiro Sugiura, Akira Yabuzaki, Eiji Yonezawa, Yasuo Terasawa, Masayuki Shinomiya, Masamichi Fukasawa, Tohru Saitoh, and Masakazu Yoshida for technical support.