Effects of the Pulsed Electron Avalanche Knife on Retinal Tissue

RECENTLY, OUR group developed a new type of surgical instrument for precise microdissection of tissue in liquid media. This device, called the pulsed electron avalanche knife (PEAK), uses electrical pulses rather than laser light for the generation of plasma microstreamers in physiological medium or in tissue.¹ Nanosecond pulses of high voltage(2-7 kV) applied to an intraocular microelectrode produce plasma in a conductive fluid, which is confined to the immediate vicinity (several tens of micrometers) of the instrument's tip (Figure 1). High temperatures associated with this process cause explosive evaporation of the surrounding liquid that results in ablation or dissection of the adjacent tissue. Because of the short duration of the pulse, dissection takes place with no thermal damage to the surrounding tissue; the cuts are purely mechanical and can be very well controlled.

Lasers that generate plasma have been used within the eye for a number of years without evidence of toxic effects beyond the mechanical effects, and neither glass nor tungsten (which form the PEAK microelectrode) is known to have a toxic effect on ocular tissue.² Nevertheless, free radicals produced by plasma in liquid,³ UV light emitted by plasma,⁴ and metal etched from the electrode tip could, in theory, have a toxic effect on tissue.

In this work, we demonstrate the precision of the PEAK in cutting of retinal tissue, both in vitro and in vivo, and evaluate whether there are any toxic effects from free radicals, UV light, or etched metal associated with application of the PEAK.

Ten bovine eyes were obtained from a local slaughterhouse 3 hours after death and packed on ice during delivery. Retinal dissection with the PEAK was performed through the vitreous after removal of cornea, lens, and iris with the use of 100-nanosecond pulses at 50 µJ with a repetition rate of 20 Hz. This energy level is near the minimum required to produce a full-thickness retinal cut.

Fifteen albino rabbits weighing 3.0 to 3.5 kg were used in accordance with the Association for Research in Vision and Ophthalmology Statement Regarding the Use of Animals in Ophthalmic and Vision Research. Before retinal cutting, a partial vitrectomy was performed via a 1-mm sclerotomy made 3 mm behind the limbus. The PEAK probe (shown diagrammatically in Figure 1) was inserted through the same sclerotomy, and 2 linear cuts of about 4 mm in length were made parallel to the medullary ray with the use of 100-nanosecond pulses of 17 µJ with a repetition rate of 10 Hz. This energy level was just sufficient to cut the full thickness of the retina without producing hemorrhage. The eyes were enucleated immediately thereafter and fixed in glutaraldehyde-paraformaldehyde fixative, postfixed in osmium tetroxide, and embedded in epoxy resin (Epon 812; Electron Microscopy Sciences, Fort Washington, Pa). One-micrometer sections were cut on an ultramicrotome and stained with toluidine blue for light microscopy.

For scanning electron microscopy, retina was fixed in glutaraldehyde-paraformaldehyde fixative (1.25%/1% in 0.2M sodium cacodylate buffer) at room temperature for 30 minutes. The tissue was then fixed for an additional 18 hours at 4°C, rinsed in buffer, and postfixed in osmium tetroxide. The tissue was washed with distilled water, dehydrated in a series of ethanols, transferred to absolute acetone, and critical point dried. The dried tissues were then mounted on stubs and plasma coated with gold.

To test for possible toxic effects from the PEAK probe, we applied the discharges into the vitreous of 10 rabbit eyes. To maximize concentration of the substances released during the procedure, no infusion and no irrigation were attached to the eye. Electroretinography (ERG) was performed on 5 rabbits before, and at intervals after, the procedure. For ERG recordings, the pupils were dilated and the eyes dark adapted for 20 minutes. A jet contact lens electrode was positioned on the cornea with a reference wire placed under the eyelid and a ground attached to the ear. The stimulus flashes at setting 1 were applied with a flash lamp (PS22; Grass Instruments Co, West Warwick, RI). The signal was amplified with a preamplifier (SR530; Stanford Research Systems Inc, Mountain View, Calif) and recorded on an oscilloscope (TDS 620B; Tektronix Inc, Irvine, Calif). In 5 other rabbits, we applied the same PEAK treatment and enucleated the eyes 4 and 14 days after surgery for histologic evaluation with light microscopy.

Light emission from the PEAK was measured with a large-area silicon photoreceiver (model 2031; New Focus Inc, San Jose, Calif).

To compare the rates of metal erosion from the PEAK with that of conventional diathermy, we applied the Wet-Field Hemostatic Eraser (Mentor Ophthalmic Inc, a division of Goldman and Shurtleff Inc, Randolf, Mass) powered by a Wet-Field Coagulator (Mentor Ophthalmic Inc) at settings typical for retinal applications(25/70 on the instrument's scale).

In all the surgical applications of the PEAK in this study, pulses of 100-nanosecond duration and negative polarity were applied to the central wire of the PEAK probe, while the coaxial needle served as a return electrode(see Figure 1).

Cuts were made in bovine retina while the probe was held by hand and moved with a velocity of approximately 1 mm/s. Scanning electron microscopy(Figure 2) showed crisp cuts of about 50 µm in width. Some perforations from single PEAK pulses were seen at the edge of the cut. A histologic section of a similar cut (Figure 3) showed that the edges of the cut were remarkably abrupt and there were no signs of coagulation or cellular disruption beyond the edge.

In living rabbit eyes, a single PEAK pulse produced a small crater from which tissue was removed to about half retinal thickness (Figure 4). By applying pulses at a repetition rate of 10 Hz and scanning across the retina at a rate of 0.5 to 1.0 mm/s, we made full-thickness cuts through retina without choroidal damage. There was no bleeding from cuts through avascular rabbit retina, but the major retinal vessels bled if cut, since the device does not coagulate tissue. Typical cuts, as shown in Figure 5, were remarkably clean and sharp. The width of the dissected zone was generally less than 100 µm, and neither retinal detachment nor signs of thermal damage were observed near the cut. If multiple pulses were delivered to one location, there was deeper penetration into the choroid, resulting in hemorrhage (Figure 6). Although PEAK effects at these settings were readily apparent histologically, they were nearly invisible to a surgeon using an operating microscope; the surgeon saw only very faint whitening of the retina along the line of scanning with the probe. In contrast, the cuts in detached retina and epiretinal membranes were readily observable, since these tissues are more opaque, and the gap was easily seen as they split apart during dissection.

To measure the erosion rate of the tungsten wire in a PEAK probe, we took micrographs of the electrode after progressively larger numbers of discharges(Figure 7). At pulse energy of 150µJ, the erosion rate was on the order of 0.065 µm³per pulse. During "maximal" surgery, which we estimate might involve 10 minutes of cutting time and 3 × 10⁴ pulses, the total eroded volume would be on the order of 2 × 10³ µm³ (equivalent to 0.04 µg or 0.2 nmol).

For comparison, we measured the amount of metal released by a standard wet-field diathermy probe in typical operational conditions. As can be seen in Figure 8, about 3 × 10⁵ µm³ of iron was removed during 100 seconds of application. The total duration of coagulation treatment in typical vitreoretinal surgery is less than 20 seconds (with pulses of 1-3 seconds in duration), which would correspond to 6 × 10⁴ µm³ (0.5 µg, 8 nmol) of iron released into vitreous during a coagulation-intensive operation.

To directly evaluate possible toxic effects of the PEAK probe to retina, we applied 3 × 10⁴ pulses of 180 µJ in 5 rabbit eyes. The energy was delivered into vitreous toward the posterior pole about 1 cm from the retina, and the eyes were not perfused to maximize possible toxic effects. Preoperatively and at 7, 11, and 30 days after the operation, ERG was performed. The results from all 5 animals were essentially identical and showed no influence of the procedure on the ERG (Figure 9).

Five additional animals were exposed to the same procedure and were killed 4 and 14 days after the treatment. Light microscopy of histologic sections showed no retinal changes in the eye exposed to the PEAK compared with the control eye.

The PEAK makes cuts through bovine and rabbit retina that are remarkably sharp, with no evidence of cellular damage at distances beyond the first layer of cells adjacent to the edge. Such cutting resembles cracking rather than ablation of tissue, since very little material seems to be ejected from the application site. The cuts are so clean, in fact, that in attached retina they are nearly invisible to a surgeon using an operating microscope. For this reason it will be important to perform further studies to define the proper and safe pulse energies for different applications on the retina. Visibility of cuts is less of a problem with detached retina and epiretinal membranes, since the cuts are readily observable in these more opaque tissues, and because these tissues tend to spread apart during dissection (leaving a gap that is easily observed on a dark background). Longer-term studies are needed to assess whether there is any injury to cells at the margins of the PEAK cuts that was not visible in tissue enucleated immediately after surgery. Our present experiments, which were designed to show only the precision of cutting, did not look beyond the immediate effects of the surgery.

Since histologic examination of the cut retina shows no signs of coagulation effects, the faint whitening of the retinal surface that we observed during cutting of the attached retina may be a result of axoplasmic flow abnormalities from damage to nerve fiber layer in the cut area. It should be emphasized that, since energy deposition is localized at the end of the probe, the instrument cuts most efficiently if it is held close to the tissue (within 100 µm). Cutting width and depth are highly reproducible with a fixed position of the probe. However, reproducibility of cutting in vivo, especially on attached retina, will be influenced by hand control and the fact that natural tremor of a human hand itself is on the order of 100 µm (or more). Pushing the probe into attached retina would eliminate this problem, but it must be avoided to prevent mechanical damage. On the other hand, on detached retina and epiretinal membranes, higher energy can be applied and the probe can safely touch the tissue to ensure a close contact. As a result, clinical cutting of detached retina and epiretinal membranes is more reproducible than cutting of attached retina.

Advantage of the PEAK as compared with scissors or blades is mainly in the tractionless nature of its cutting. Erbium:YAG^{5 - 6} and argon fluoride excimer lasers^{7 - 8} have also been applied for tractionless vitreoretinal cutting, but these devices did not gain widespread acceptance in practice mostly because of the high cost, large size, and somewhat rigid fibers. The PEAK is a much smaller and less expensive device, with a light handpiece and very flexible cable. In addition, PEAK probes can be bent at any angle without losing cutting efficiency.

The availability of a precise microsurgical tool capable of producing microscopic cuts of controlled width and depth without traction should facilitate many aspects of vitreoretinal surgery. The PEAK will be useful in the performance of the circumferential and radial retinotomies and in the dissection of complex epiretinal and fibrovascular membranes. Its value will be particularly evident in eyes where excessive traction or pressure could cause collateral damage, such as those with atrophic underlying retina susceptible to tractional tears, and with membranes adjacent to blood vessels or overlying detached retina. In addition, this tool may be applied to posterior and anterior capsulotomy and other ocular surgical procedures.

The generation of plasma in liquids could, in theory, be toxic to tissue on the basis of the formation of free radicals, the generation of UV light, and the erosion of a metal electrode. To calculate the maximal exposure that is likely to occur, typical applications of PEAK require discharge energies between 50 and 150 µJ, at a pulse repetition rate of about 30 Hz, and the total number of pulses does not exceed 5000. An exceedingly large exposure, which we shall use for the sake of discussion, would be 10 minutes of total application time at 100 µJ and 50 Hz (3 × 10⁴ pulses). With this exposure, the total deposited energy is 3 J, and the average cumulative energy density in the posterior pole of a human eye would be about 0.75 J/mL.

Free radicals ( · H and · OH) can be produced directly by plasma-mediated discharge, but their lifetime in water is only about 70 nanoseconds, during which they can only diffuse about 20 nm³. Thus, free radicals will have no effect beyond the zone of mechanical damage (which may vary between a few and a few hundred micrometers). The recombination of 2 OH radicals can produce hydrogen peroxide (H₂O₂), which is a strong and long-living oxidizer. However, it has been shown³ that the threshold energy density to produce detectable H₂O₂toxicity is about 500 J/mL, which is 3 orders of magnitude greater than would occur with the PEAK.

Our measurements of light emission by plasma indicate that UV toxicity is also unlikely to occur during PEAK surgery, since the total dose of photic irradiation during a PEAK-intensive operation (involving 3 × 10⁴ pulses) would be 6 orders of magnitude lower than the established danger levels for retinal exposure to UV light (5 J/cm²).⁹ One possible source of PEAK toxicity involves deposition of eroded metal from the microelectrode. We compared the amount of metal etched from the PEAK probe in typical operational conditions with that released by a standard wet-field diathermy probe. Diathermy electrodes are made of steel and erode during electrosurgery, producing iron ions in various oxidation states. Iron metal and iron compounds have important toxic actions on the eye. These have been studied for more than a century and have been the subject of an extensive body of literature.^{10 - 11} According to our measurements, 0.5 µg (or 8 nmol) of steel is eroded during 20 seconds of application of diathermy at typical vitreoretinal settings, which is about 25% of the threshold amount of iron shown to be toxic if injected into the vitreous.¹¹ However, eyes are perfused during retinal surgery, which should reduce the toxic load by a considerable margin (a factor of 2 to 10 on the basis of typical perfusion rates and duration). To our knowledge, no observations of retinal (or other) toxic effects from diathermy probes have been reported so far, even though some operations may have involved even longer exposures.

The central electrode in the PEAK probe is made of tungsten. No toxic effect from tungsten metal has been found in experiments with intravitreal implants in rabbit eyes.^{10 ,12} During typical surgery involving 3 × 10⁴ pulses, 0.04 µg(or 0.2 nmol) of tungsten will be eroded. This amount is only 2.5% of the number of moles of iron left in the eye after application of diathermy for 20 seconds. Since 8 nmol of rather toxic metallic iron released by diathermy does not seem to have any adverse effects on ocular tissues during surgery, the much smaller amount (0.2 nmol) of nontoxic tungsten is probably safe. These estimations have also been supported by the lack of adverse effects of PEAK treatments on the ERG and on retinal histologic findings in our experiments.

We conclude that the PEAK is capable of precise cutting through retinal tissue, and that there are no demonstrable retinal toxic effects from the use of this device. The precision and tractionless nature of PEAK cutting offers potential advantages over mechanical tools, as well as over laser-based cutting instrumentation. We believe this new device could be useful in a variety of vitreoretinal surgical applications.

Submitted for publication May 31, 2001; final revision received November 7, 2001; accepted December 21, 2001.

This study was supported by grant 1 R01 EY12888-01A1 from the National Eye Institute, Bethesda, Md, and by Carl Zeiss Inc, Jena, Germany.

We thank Rupa Dalal for her help with sample preparations.

Corresponding author and reprints: Daniel V. Palanker, PhD, W. W. Hansen Experimental Physics Laboratory, Stanford University, Stanford, CA 94305-4085 (e-mail: palanker@stanford.edu).