- Email: [email protected]
Threshold for Retinal Damage Associated with the use of High-Power Neodymium-Yag Lasers in the Vitreous
THRESHOLD FOR RETINAL DAMAGE ASSOCIATED WITH THE USE OF HIGH-POWER NEODYMIUM-YAG LASERS IN THE VITREOUS ROBERT F. BONNER, P H . D . , SANFORD M. MEYERS, AND DOUGLAS E. GAASTERLAND, M.D. Bethesda, Maryland
M.D.,
Ultrashort, Q-switched or mode-locked, neodymium-YAG laser puls es focused within 2 mm of the retina caused reproducible retinal damage in four eyes of two monkeys and in four eyes of three rabbits. The distance of the laser focus from the retina for clinically observed threshold retinal damage was characterized for pulse energies up to 9 mj. For the 2- to 6-mJ pulse energies necessary to rupture vitreal membranes in clear media in rabbits, the high-power laser pulses could not be focused within 2 mm of the retina without substantial risk of damaging the underlying retina. These laser pulses did not rupture vitreal membranes in hazy ocular media that prevented precise focus ing. The retinal damage was somewhat greater than expected for retinal absorption of 1.06-μηι laser energy, suggesting that secondary effects such as self-focusing and shock waves emanating from the focus may be important. Short-pulse, Q-switched or modelocked, neodymium-YAG ophthalmologic laser systems are currently used primari ly for opening pupillary membranes, al though some investigators have ruptured selected vitreal membranes.1"6 When fo cused to a sufficiently small spot size (50 to 90 μπι in our system), short highpower neodymium-YAG laser pulses can rupture any targeted tissues in the eye because of nonlinear absorption. In bio logic tissue this occurs for power densi ties above 109 watts/cm2. To cause signifi cant disruption, the laser pulse must also
Accepted for publication May 16, 1983. From the Biomédical Engineering and Instrumen tation Branch, Division of Research Services (Dr. Bonner); and the Clinical Branch, National Eye Institute (Drs. Meyers and Gaasterland), National Institutes of Health, Bethesda, Maryland. Reprint requests to Robert F. Bonner, Ph.D., Biomédical Engineering and Instrumentation Branch, National Institutes of Health, Bldg. 13, Rm. 3W13, Bethesda, MD 20205.
have enough energy to vaporize the tar geted tissue explosively. This explosion creates a shock wave that may further the disruption, but that may also damage the underlying retina. The underlying retina may also be damaged by that part of the primary radiation not absorbed at the target tissue or by secondary spark dis charges. The absorption of the laser pulse at the focus and at the retina, the magni tude of pressure in the shock wave at various distances from the site of absorp tion, and the retinal damage may be complicated functions of pulse power and energy. Therefore, the effects of various high-power laser pulses are not easily predictable. Thus, it is important to es tablish empirically the thresholds for reti nal damage as a function of both pulse characteristics and distance of the focus from the retina. Previous investigators have described histopathologic findings in threshold le sions in monkeys for high-power neo-
AMERICAN JOURNAL OF OPHTHALMOLOGY 96:153-159, 1983
153
154
AMERICAN JOURNAL OF OPHTHALMOLOGY
dymium-YAG laser pulses focused on the retina. 6,7 These findings suggested that the damage results primarily from local ized absorption in the melanin granules of the pigment epithelium with disrup tion of the outer retina by ensuing shock waves. In our study, we examined the potential for clinically observed retinal damage as a function of pulse energy and distance of the focus from the retina in monkeys and rabbits and determined the pulse energy and power density needed to rupture vitreal membranes in rabbits. MATERIAL AND METHODS
One adult pigtail monkey (Macaca nemestrina; two eyes) and one adult rhe sus monkey (M. mulatta; two eyes) with clear media and normal fundi and without previous vitreous or retinal surgery were anesthetized with an intramuscular mix ture of ketamine (7 mg/kg of body weight) and xylazine (1 mg/kg of body weight) or intravenous pentobarbital (15 mg/kg of body weight). Three adult pigmented rabbits (four eyes) with clear media and normal fundi and without previous sur gery were anesthetized with a mixture of ketamine (35 mg/kg of body weight) and xylazine (1 mg/kg of body weight) given intramuscularly. We used 1% tropicamide and 10% phenylephrine to dilate all the eyes. All the animals underwent indirect ophthalmoscopy. Preoperatively, no ocular lesions were present and the media were clear. We used three adult pigmented rabbits (six eyes) with clear vitreous and vitreal membranes and two adult pigmented rabbits (three eyes) with hazy vitreous and vitreal membranes to determine the thresholds needed to rupture vitreal membranes. The vitreal membranes were created by a double penetrating wound. 8,9 We used an uncoated Goldmann contact lens on each animal during laser surgery. Selected eyes underwent fundus photography. The animals under
AUGUST, 1983
went indirect ophthalmoscopy at one day, two days, one week, and four weeks postoperatively. We used a prototype neodymium-YAG pphthalmologic laser system (Coherent) coupled to a slit lamp capable of deliver ing either single Q-switched 20-nsec pulses or a series of ten 25-psec modelocked pulses within a 50-nsec train to compare the relative safety of the two pulse modalities. The delivery system focused the neodymium-YAG pulses (ap proximately gaussian with a 14-degree cone angle between the l e points) along with a coaxial targeting helium-neon laser beam in the central field of a slit lamp. The neodymium-YAG beam, the heliumneon targeting beam, and the slit lamp were all confocal. The minimum spot size of the neodymium-YAG laser beam at the focal plane was 50 to 90 μπι. For each application, the targeting helium-neon beam was focused on the retina; the slit lamp was then moved away from the eye by a given (0 to 5 mm) distance. The accuracy of this displacement was approx imately 0.125 mm. We estimated the distance of the focal point of the laser pulse in front of the retina by correcting the displacement with ray tracings based on the optical equivalent eye for a mon key10 and a rabbit 11 in the presence of a Goldmann contact lens, which negates the power of the anterior cornea. For monkeys and humans, the corrections predicted by these ray tracings were small and would not be substantially af fected by variations among individuals. For rabbit lenses, which have much greater refractive powers, the corrections based on ray tracings were much larger and might result in errors caused by variations from the normal model in spe cific eyes. The threshold for clinically observed retinal damage by a given pulse energy was evaluated by determining the maximum distance of the focal spot in front of the retina at which an immediate-
VOL. 96, NO. 2
RETINAL DAMAGE FROM YAG LASER
ly observable retinal lesion was obtained for most pulses. In this way we deter mined the threshold for clinically ob served retinal lesions for various pulses focused at different distances from the retina in three areas in monkeys (parafovea, fovea, and inferior to the optic disk) and in the central area (below the horizontal raphe) in rabbits. RESULTS
The threshold for a clinically observed retinal lesion was a faint whitening of the
155
retina observable immediately after irra diation. Mild to moderate lesions pro duced more extensive and more intense whitening of the retina. Severe lesions were even more extensive and produced subretinal and retinal hemorrhages that occasionally broke through into the vitre ous (Fig. 1). At subthreshold distances (0.2 to 0.4 mm greater than the threshold distance), no clinical retinal lesions were observed 48 hours and four weeks postoperatively. If no clinical retinal le sions were observed at surgery, none
Fig. 1 (Bonner, Meyers, and Gaasterland). Repre sentative fundus photographs in a rhesus monkey two hours (top left), 48 hours (top right), and four weeks (bottom left) after exposure to neodymiumYAG laser pulses. Subthreshold laser pulses (distance of focus in front of the retina was 120% of the threshold distance) were delivered to areas A (modelocked) and B (Q-switched). Threshold lesions were obtained in areas C (mode-locked) and D (Qswitched). A moderate lesion at E and a severe lesion with hemorrhage at F were created by a suprathreshold pulse (Q-switched and mode-locked respective ly). The retinal lesions enlarged at 48 hours, but no new lesions appeared. Only minimal pigmentary changes remained four weeks later.
156
AMERICAN JOURNAL OF OPHTHALMOLOGY
developed in the postoperative period. Threshold lesions were larger at 48 hours, but only mild pigmentary changes remained at four weeks (Fig. 1). We delivered a minimum of three threshold pulses and six subthreshold pulses to establish each data point in Figure 2. The threshold distance for dam age was reproducible in the areas of the retina examined in both the monkeys and rabbits. Figure 2 includes all damage thresholds obtained. In general, for any given distance, the threshold energy for retinal damage was somewhat higher for Q-switched pulses (20 nsec) than for the ten 25-psec mode-locked pulses deliv ered in a 50-nsec train. For example, at 1.5 mm (distance of the laser focus from the retina) the average energy of the
Focal Point (mm from retina) Fig. 2 (Bonner, Meyers, and Gaasterland). The energy (in millijoules) at the cornea of Q-switched (square, rabbit; circle, rhesus monkey; and triangle, pigtail monkey) and mode-locked (x, rabbit +, rhesus monkey; and w, pigtail monkey) neodymiumYAG laser pulses causes a threshold clinical retinal lesion when focused the corresponding distance in front of the retina. The upper curve represents the best least-squares fit of the Q-switched data to a constant retinal irradiance (2.1 joules/cm2, assuming no absorption or self-focusing at the focus). The lower curve is a theoretical curve of constant retinal irradi ance of 0.8 joules/cm2 which approximates the dam age thresholds for mode-locked pulse trains with energies less than 2 mj.
AUGUST, 1983
Q-switched pulse at threshold was about 4 m j vs 2 m j for that of the mode-locked pulse train. The severity of the retinal damage increased dramatically when the focal spot was moved closer to the retina. Moving the focal spot 0.3 to 0.8 mm toward the retina usually increased the damage from threshold to a severe lesion with hemorrhage. In Figure 2, these se vere lesions would appear at the left of the threshold values at a given energy. For the Q-switched pulses, the thresh old energy for retinal damage showed a dependence of approximately r2 (where r was the distance of the laser focus from the retina), equivalent to a constant peak retinal irradiance of 2.1 joules/cm 2 , as suming that nonlinear effects at the laser focus were negligible. For all modelocked pulses above 0.5 m j and all Qswitched pulses above 4 m j , we observed a spark discharge from the plasma formed at the laser focus, indicating considerable nonlinear absorption of energy at the focus. For pulse energies of 3 m j or less, the mode-locked pulses created more ret inal damage than the equivalent Q-switch pulse. The most severe lesions occurred with the mode-locked pulses focused close to the retina. The energy, measured at the cornea, in a pulse train of mode-locked pulses need ed to puncture vitreal membranes in rab bits was 1.8 to 5.0 m j . Similarly, single 2.4- to 6.5-mJ Q-switched pulses punc tured vitreal membranes. The energy de livered to the intraocular focus was ap proximately 0.76 of these values because of linear absorption of the eye anterior to the focal plane. 7 Thin (clinically estimat ed to be 100 μπι) membranes could be ruptured by pulses with the above ener gies, but thick membranes (> 250 μπι) appeared to be disrupted only with puls es, either Q-switched or mode-locked, of more than 4 mj each. Effective disruption depended on the sharpness of focus of the laser beam on the membranes, particu-
VOL. 96, NO. 2
RETINAL DAMAGE FROM YAG LASER
larly at the lower pulse energies. In the presence of vitreal opacities and haze, which prevented precise focusing and caused scattering of the laser energy, we were unable to disrupt vitrèal mem branes with any of these pulses. DISCUSSION
Clinically observed retinal damage thresholds for the 2- to 6-mJ pulses nec essary to rupture vitreal membranes were roughly comparable for the two laser modes. However, there appeared to be somewhat more risk of retinal damage associated with the mode-locked pulse trains. Almost all the mode-locked dam age thresholds (Fig. 2, clustered about the lower curve) occurred at greater dis tances than those for the Q-switched pulse of the same energy (Fig. 2, upper curve). From our studies with vitreal mem branes in rabbits it appeared that similar energies (2 to 6 mj) were needed to rupture vitreal membranes in both Qswitched and mode-locked laser modes. Our data showed that attempting to rup ture vitreal membranes at a point within 2 mm of the retina incurred a substantial risk of damaging the underlying retina (Fig. 2). When pulses of 4 to 8 m j are used to rupture vitreal membranes 2 to 4 mm from the retina, errors in focusing can easily result in severe retinal damage. It is important to note that damage to the retina may be substantially greater with ophthalmologic neodymium-YAG highpower laser systems with a smaller intra ocular cone angle (less than the 14degree cone angle to the l e points of our system) because of higher equivalent reti nal irradiance for the pulse characteristics (Fig. 2). Previous studies 6,7 have examined clini cal damage thresholds and histopathologic findings when ultrashort high-power pulses were focused on the retinas of monkeys. The reported threshold energy
157
density for retinal damage in monkeys was 2.0 ± 0.5 joules/cm 2 for single 30psec pulses and 10 ± 2 joules/cm 2 for 15nsec pulses (Q-switched) from a neody mium-YAG laser focused to a 25-μπι spot on the retina. 6,7 For damage caused by mode-locked pulses, the reported thresh old energy density varied only slightly from 2.7 to 1.6 joules/cm 2 with the spot size increasing from 25 μπι to 430 μπι. 7 Histologie studies of threshold lesions showed that the retinal damage apparent ly results from the rapid absorption of the 1.06-μπι radiation by the pigment epithe lium and disruption of the outer retina by shock waves propagated from melanin granules. In our study we examined the damage to the retina caused by high-power neodymium-YAG laser pulses focused at various distances in front of the retina with the spot diameter of the beam on the retina ranging from 70 to 600 μπι. For Q-switched pulses focused to a 70-μπι spot on the retina, we observed a thresh old energy density of 8 ± 4 joules/cm 2 in rabbits and monkeys, values similar to those of previous reports. 6 As the focus was moved away from the retina ( > 0.5 mm), the pulse energy for thresh old retinal damage increased to values where the power density at the focus (> 1 gigawatt/cm 2 ) exceeded nonlinear absorption thresholds. For these Qswitched pulses focused in front of the retina, the calculated threshold energy density on the retina (assuming that non linear effects at the focus were negligible) was an approximately constant 2.1 joules/ cm 2 as the energy in the pulse (1 to 9 mj) and spot diameter on the retina (200 to 600 μπι) increased. This calculated value of 2.1 joules/cm 2 for the retinal energy density of the 1.06-μπι primary irradia tion was roughly fivefold less than values obtained when the energy was focused on the retina. 6,7 This increased sensitivity may be the result of "hot spots" in the
158
AMERICAN JOURNAL OF OPHTHALMOLOGY
beam caused by self-focusing of the pri mary radiation or of the larger-spot sizes on the retina in our experiment. 7 Allow ing for either of these effects, the damage thresholds we obtained appeared to be consistent with most of the energy being transmitted through the focus to the un derlying retina. The r2 dependence of the threshold energy on distance observed for Q-switched pulses (Fig. 2, upper curve) was expected for damage caused by a divergent beam from its point of focus. Alternatively, retinal damage asso ciated with the shock wave emanating from the laser focus in front of the retina may have accounted for the increased retinal sensitivity. The damage thresholds for modelocked pulses were less than those for Q-switched pulses and showed a more complicated dependence on distance from the retina. For pulse energies below 2 mj that created threshold lesions, the calculated retinal energy density was ap proximately 0.8 joules/cm 2 , twofold less than the reported values 6,7 for single mode-locked pulses focused to compara ble spot sizes on the retina. For pulse energies above 2 mj, the damage thresh olds became more scattered and in creased more rapidly than r2. Thus, the apparent retinal energy density for threshold damage increased by roughly twofold for 4-mJ pulse trains. This was consistent with 50% of the energy being absorbed by the plasma formed at the focus. We measured the nonlinear ab sorption of the plasma in water for these pulses and obtained variable results in the range of 50% absorption. Thus, the retina was apparently dam aged primarily by the primary radiation diverging from the focus and transmitted to the retina. The retina is roughly three fold more sensitive to mode-locked pulse energy than to Q-switched pulse energy at low pulse energies. At high pulse ener
AUGUST, 1983
gies the retina may be partially shielded from the mode-locked pulses by the con siderable absorption at the focus, but for our system this variable effect never made the mode-locked pulses less damag ing than the Q-switched pulses. We em phasize that our studies of damage thresholds were performed in clear vitre ous. In vitreous with mild haze, scatter ing of the beam increases the spot size at the focus. Thus, one effect of vitreal haze is to decrease the nonlinear absorption at the focus, resulting in greater retinal irradiance and damage behind the focus. Our thresholds for clinically observed lesions should not be interpreted as indi cating that all pulses below the thresholds presented in Figure 2 are safe. A hazy vitreous or smaller intraocular cone angle may substantially decrease the threshold for damage. Additionally, van der Zypen, Bebie, and Fankhauser 12 reported that as pulse energy increases the absorptive zone of neodymium-YAG pulses may grow anteriorly with respect to the helium-neon focal plane. Thus, for opti mal cutting of vitreal membranes the helium-neon aiming beam may be fo cused posteriorly to the membrane. In this case the threshold distance of the helium-neon focus above the retina (given in Fig. 2) would be less than the distance of the membrane above the retina. Further investigation is necessary to determine the long-term effects of this noninvasive surgery, such as regrowth of vitreal membranes and inflammatory re actions. Additionally, our inability to rup ture vitreal membranes in a hazy vitreous showed that the neodymium-YAG laser can be used only in carefully selected vitreal membranes. Although caution is needed in relating animal data to clinical situations, our study indicated that there is a substantial risk of retinal damage when ultrashort neodymium-YAG pulses
VOL. 96, NO. 2
RETINAL DAMAGE FROM YAG LASER
are used to rupture vitreal membranes close to the retina, especially near the fovea. REFERENCES 1. Aron-Rosa, D., Aron, J. J., Griesemann, M., and Thyzel, R. : Use of the neodymium-YAG laser to open the posterior capsule after lens implant surgery. A preliminary report. J. Am. Intraocul. Implant Soc. 6:352, 1980. 2. Aron-Rosa, D., Griesemann, J. C , and Aron, J. J. : Use of a pulsed neodymium-YAG laser (picosec ond) to open the posterior lens capsule in traumatic cataract. A preliminary report. Ophthalmic Surg. 12:496, 1981. 3. Peyman, G. A., Kraft, M., Viherkoski, E., and Ressler, N.: Noninvasive capsulectomy using a pulsed infrared laser. J. Am. Intraocul. Implant Soc. 8:239, 1982. 4. Fankhauser, F., Roussel, P., Steffen, J., van der Zypen, E., and Chrenkova, A.: Clinical studies on the efficiency of high power laser radiation upon some structures of the anterior segment of the eye. Int. Ophthalmol. 3:129, 1981. 5. Fankhauser, F., Lortscher, H., and van der Zypen, E.: Clinical studies on high and low power laser radiation upon some structures of the anterior
159
and posterior segments of the eye. Int. Ophthalmol. 5:15, 1982. 6. Ham, W. T., Mueller, H. A., Goldman, A. I., Newman, B. E., Holland, L. M., and Kuwabara, T.: Ocular hazards from picosecond pulses of ND:YAG laser radiation. Science 185:362, 1974. 7. Goldman, A. I., Ham, W. T., and Mueller, H. A.: Ocular damage thresholds and mechanisms for ultrashort pulses of both visible and infrared laser radiation in rhesus monkey. Exp. Eye Res. 24:45, 1977. 8. Topping, T. M., Abrams, G. W., and Machemer, R.: Experimental double-perforating injury of the posterior segment in rabbit eyes. Arch. Ophthalmol. 97:735, 1979. 9. Meyers, S. M., and Rodrigues, M. M.: Effect of selected intravitreal drugs after severe penetrating injury in rabbits. Curr. Eye Res. 1:471, 1981. 10. Bito, L. Z., DeRousseau, C. J., Kaufman, P. L., and Bito, J. W.: Age-dependent loss of ac commodative amplitude in rhesus monkeys. An ani mal model for presbyopia. Invest. Ophthalmol. Vis. Sei. 23:23, 1982. 11. Hughes, A.: A schematic eye for the rabbit. Vision Res. 12:123, 1972. 12. van der Zypen, E., Bebie, H., and Fankhauser, F.: Morphological studies about the efficiency of laser beams upon structures of the angle of the anterior chamber. Int. Ophthalmol. 1:109, 1979.
M.D.,
Ultrashort, Q-switched or mode-locked, neodymium-YAG laser puls es focused within 2 mm of the retina caused reproducible retinal damage in four eyes of two monkeys and in four eyes of three rabbits. The distance of the laser focus from the retina for clinically observed threshold retinal damage was characterized for pulse energies up to 9 mj. For the 2- to 6-mJ pulse energies necessary to rupture vitreal membranes in clear media in rabbits, the high-power laser pulses could not be focused within 2 mm of the retina without substantial risk of damaging the underlying retina. These laser pulses did not rupture vitreal membranes in hazy ocular media that prevented precise focus ing. The retinal damage was somewhat greater than expected for retinal absorption of 1.06-μηι laser energy, suggesting that secondary effects such as self-focusing and shock waves emanating from the focus may be important. Short-pulse, Q-switched or modelocked, neodymium-YAG ophthalmologic laser systems are currently used primari ly for opening pupillary membranes, al though some investigators have ruptured selected vitreal membranes.1"6 When fo cused to a sufficiently small spot size (50 to 90 μπι in our system), short highpower neodymium-YAG laser pulses can rupture any targeted tissues in the eye because of nonlinear absorption. In bio logic tissue this occurs for power densi ties above 109 watts/cm2. To cause signifi cant disruption, the laser pulse must also
Accepted for publication May 16, 1983. From the Biomédical Engineering and Instrumen tation Branch, Division of Research Services (Dr. Bonner); and the Clinical Branch, National Eye Institute (Drs. Meyers and Gaasterland), National Institutes of Health, Bethesda, Maryland. Reprint requests to Robert F. Bonner, Ph.D., Biomédical Engineering and Instrumentation Branch, National Institutes of Health, Bldg. 13, Rm. 3W13, Bethesda, MD 20205.
have enough energy to vaporize the tar geted tissue explosively. This explosion creates a shock wave that may further the disruption, but that may also damage the underlying retina. The underlying retina may also be damaged by that part of the primary radiation not absorbed at the target tissue or by secondary spark dis charges. The absorption of the laser pulse at the focus and at the retina, the magni tude of pressure in the shock wave at various distances from the site of absorp tion, and the retinal damage may be complicated functions of pulse power and energy. Therefore, the effects of various high-power laser pulses are not easily predictable. Thus, it is important to es tablish empirically the thresholds for reti nal damage as a function of both pulse characteristics and distance of the focus from the retina. Previous investigators have described histopathologic findings in threshold le sions in monkeys for high-power neo-
AMERICAN JOURNAL OF OPHTHALMOLOGY 96:153-159, 1983
153
154
AMERICAN JOURNAL OF OPHTHALMOLOGY
dymium-YAG laser pulses focused on the retina. 6,7 These findings suggested that the damage results primarily from local ized absorption in the melanin granules of the pigment epithelium with disrup tion of the outer retina by ensuing shock waves. In our study, we examined the potential for clinically observed retinal damage as a function of pulse energy and distance of the focus from the retina in monkeys and rabbits and determined the pulse energy and power density needed to rupture vitreal membranes in rabbits. MATERIAL AND METHODS
One adult pigtail monkey (Macaca nemestrina; two eyes) and one adult rhe sus monkey (M. mulatta; two eyes) with clear media and normal fundi and without previous vitreous or retinal surgery were anesthetized with an intramuscular mix ture of ketamine (7 mg/kg of body weight) and xylazine (1 mg/kg of body weight) or intravenous pentobarbital (15 mg/kg of body weight). Three adult pigmented rabbits (four eyes) with clear media and normal fundi and without previous sur gery were anesthetized with a mixture of ketamine (35 mg/kg of body weight) and xylazine (1 mg/kg of body weight) given intramuscularly. We used 1% tropicamide and 10% phenylephrine to dilate all the eyes. All the animals underwent indirect ophthalmoscopy. Preoperatively, no ocular lesions were present and the media were clear. We used three adult pigmented rabbits (six eyes) with clear vitreous and vitreal membranes and two adult pigmented rabbits (three eyes) with hazy vitreous and vitreal membranes to determine the thresholds needed to rupture vitreal membranes. The vitreal membranes were created by a double penetrating wound. 8,9 We used an uncoated Goldmann contact lens on each animal during laser surgery. Selected eyes underwent fundus photography. The animals under
AUGUST, 1983
went indirect ophthalmoscopy at one day, two days, one week, and four weeks postoperatively. We used a prototype neodymium-YAG pphthalmologic laser system (Coherent) coupled to a slit lamp capable of deliver ing either single Q-switched 20-nsec pulses or a series of ten 25-psec modelocked pulses within a 50-nsec train to compare the relative safety of the two pulse modalities. The delivery system focused the neodymium-YAG pulses (ap proximately gaussian with a 14-degree cone angle between the l e points) along with a coaxial targeting helium-neon laser beam in the central field of a slit lamp. The neodymium-YAG beam, the heliumneon targeting beam, and the slit lamp were all confocal. The minimum spot size of the neodymium-YAG laser beam at the focal plane was 50 to 90 μπι. For each application, the targeting helium-neon beam was focused on the retina; the slit lamp was then moved away from the eye by a given (0 to 5 mm) distance. The accuracy of this displacement was approx imately 0.125 mm. We estimated the distance of the focal point of the laser pulse in front of the retina by correcting the displacement with ray tracings based on the optical equivalent eye for a mon key10 and a rabbit 11 in the presence of a Goldmann contact lens, which negates the power of the anterior cornea. For monkeys and humans, the corrections predicted by these ray tracings were small and would not be substantially af fected by variations among individuals. For rabbit lenses, which have much greater refractive powers, the corrections based on ray tracings were much larger and might result in errors caused by variations from the normal model in spe cific eyes. The threshold for clinically observed retinal damage by a given pulse energy was evaluated by determining the maximum distance of the focal spot in front of the retina at which an immediate-
VOL. 96, NO. 2
RETINAL DAMAGE FROM YAG LASER
ly observable retinal lesion was obtained for most pulses. In this way we deter mined the threshold for clinically ob served retinal lesions for various pulses focused at different distances from the retina in three areas in monkeys (parafovea, fovea, and inferior to the optic disk) and in the central area (below the horizontal raphe) in rabbits. RESULTS
The threshold for a clinically observed retinal lesion was a faint whitening of the
155
retina observable immediately after irra diation. Mild to moderate lesions pro duced more extensive and more intense whitening of the retina. Severe lesions were even more extensive and produced subretinal and retinal hemorrhages that occasionally broke through into the vitre ous (Fig. 1). At subthreshold distances (0.2 to 0.4 mm greater than the threshold distance), no clinical retinal lesions were observed 48 hours and four weeks postoperatively. If no clinical retinal le sions were observed at surgery, none
Fig. 1 (Bonner, Meyers, and Gaasterland). Repre sentative fundus photographs in a rhesus monkey two hours (top left), 48 hours (top right), and four weeks (bottom left) after exposure to neodymiumYAG laser pulses. Subthreshold laser pulses (distance of focus in front of the retina was 120% of the threshold distance) were delivered to areas A (modelocked) and B (Q-switched). Threshold lesions were obtained in areas C (mode-locked) and D (Qswitched). A moderate lesion at E and a severe lesion with hemorrhage at F were created by a suprathreshold pulse (Q-switched and mode-locked respective ly). The retinal lesions enlarged at 48 hours, but no new lesions appeared. Only minimal pigmentary changes remained four weeks later.
156
AMERICAN JOURNAL OF OPHTHALMOLOGY
developed in the postoperative period. Threshold lesions were larger at 48 hours, but only mild pigmentary changes remained at four weeks (Fig. 1). We delivered a minimum of three threshold pulses and six subthreshold pulses to establish each data point in Figure 2. The threshold distance for dam age was reproducible in the areas of the retina examined in both the monkeys and rabbits. Figure 2 includes all damage thresholds obtained. In general, for any given distance, the threshold energy for retinal damage was somewhat higher for Q-switched pulses (20 nsec) than for the ten 25-psec mode-locked pulses deliv ered in a 50-nsec train. For example, at 1.5 mm (distance of the laser focus from the retina) the average energy of the
Focal Point (mm from retina) Fig. 2 (Bonner, Meyers, and Gaasterland). The energy (in millijoules) at the cornea of Q-switched (square, rabbit; circle, rhesus monkey; and triangle, pigtail monkey) and mode-locked (x, rabbit +, rhesus monkey; and w, pigtail monkey) neodymiumYAG laser pulses causes a threshold clinical retinal lesion when focused the corresponding distance in front of the retina. The upper curve represents the best least-squares fit of the Q-switched data to a constant retinal irradiance (2.1 joules/cm2, assuming no absorption or self-focusing at the focus). The lower curve is a theoretical curve of constant retinal irradi ance of 0.8 joules/cm2 which approximates the dam age thresholds for mode-locked pulse trains with energies less than 2 mj.
AUGUST, 1983
Q-switched pulse at threshold was about 4 m j vs 2 m j for that of the mode-locked pulse train. The severity of the retinal damage increased dramatically when the focal spot was moved closer to the retina. Moving the focal spot 0.3 to 0.8 mm toward the retina usually increased the damage from threshold to a severe lesion with hemorrhage. In Figure 2, these se vere lesions would appear at the left of the threshold values at a given energy. For the Q-switched pulses, the thresh old energy for retinal damage showed a dependence of approximately r2 (where r was the distance of the laser focus from the retina), equivalent to a constant peak retinal irradiance of 2.1 joules/cm 2 , as suming that nonlinear effects at the laser focus were negligible. For all modelocked pulses above 0.5 m j and all Qswitched pulses above 4 m j , we observed a spark discharge from the plasma formed at the laser focus, indicating considerable nonlinear absorption of energy at the focus. For pulse energies of 3 m j or less, the mode-locked pulses created more ret inal damage than the equivalent Q-switch pulse. The most severe lesions occurred with the mode-locked pulses focused close to the retina. The energy, measured at the cornea, in a pulse train of mode-locked pulses need ed to puncture vitreal membranes in rab bits was 1.8 to 5.0 m j . Similarly, single 2.4- to 6.5-mJ Q-switched pulses punc tured vitreal membranes. The energy de livered to the intraocular focus was ap proximately 0.76 of these values because of linear absorption of the eye anterior to the focal plane. 7 Thin (clinically estimat ed to be 100 μπι) membranes could be ruptured by pulses with the above ener gies, but thick membranes (> 250 μπι) appeared to be disrupted only with puls es, either Q-switched or mode-locked, of more than 4 mj each. Effective disruption depended on the sharpness of focus of the laser beam on the membranes, particu-
VOL. 96, NO. 2
RETINAL DAMAGE FROM YAG LASER
larly at the lower pulse energies. In the presence of vitreal opacities and haze, which prevented precise focusing and caused scattering of the laser energy, we were unable to disrupt vitrèal mem branes with any of these pulses. DISCUSSION
Clinically observed retinal damage thresholds for the 2- to 6-mJ pulses nec essary to rupture vitreal membranes were roughly comparable for the two laser modes. However, there appeared to be somewhat more risk of retinal damage associated with the mode-locked pulse trains. Almost all the mode-locked dam age thresholds (Fig. 2, clustered about the lower curve) occurred at greater dis tances than those for the Q-switched pulse of the same energy (Fig. 2, upper curve). From our studies with vitreal mem branes in rabbits it appeared that similar energies (2 to 6 mj) were needed to rupture vitreal membranes in both Qswitched and mode-locked laser modes. Our data showed that attempting to rup ture vitreal membranes at a point within 2 mm of the retina incurred a substantial risk of damaging the underlying retina (Fig. 2). When pulses of 4 to 8 m j are used to rupture vitreal membranes 2 to 4 mm from the retina, errors in focusing can easily result in severe retinal damage. It is important to note that damage to the retina may be substantially greater with ophthalmologic neodymium-YAG highpower laser systems with a smaller intra ocular cone angle (less than the 14degree cone angle to the l e points of our system) because of higher equivalent reti nal irradiance for the pulse characteristics (Fig. 2). Previous studies 6,7 have examined clini cal damage thresholds and histopathologic findings when ultrashort high-power pulses were focused on the retinas of monkeys. The reported threshold energy
157
density for retinal damage in monkeys was 2.0 ± 0.5 joules/cm 2 for single 30psec pulses and 10 ± 2 joules/cm 2 for 15nsec pulses (Q-switched) from a neody mium-YAG laser focused to a 25-μπι spot on the retina. 6,7 For damage caused by mode-locked pulses, the reported thresh old energy density varied only slightly from 2.7 to 1.6 joules/cm 2 with the spot size increasing from 25 μπι to 430 μπι. 7 Histologie studies of threshold lesions showed that the retinal damage apparent ly results from the rapid absorption of the 1.06-μπι radiation by the pigment epithe lium and disruption of the outer retina by shock waves propagated from melanin granules. In our study we examined the damage to the retina caused by high-power neodymium-YAG laser pulses focused at various distances in front of the retina with the spot diameter of the beam on the retina ranging from 70 to 600 μπι. For Q-switched pulses focused to a 70-μπι spot on the retina, we observed a thresh old energy density of 8 ± 4 joules/cm 2 in rabbits and monkeys, values similar to those of previous reports. 6 As the focus was moved away from the retina ( > 0.5 mm), the pulse energy for thresh old retinal damage increased to values where the power density at the focus (> 1 gigawatt/cm 2 ) exceeded nonlinear absorption thresholds. For these Qswitched pulses focused in front of the retina, the calculated threshold energy density on the retina (assuming that non linear effects at the focus were negligible) was an approximately constant 2.1 joules/ cm 2 as the energy in the pulse (1 to 9 mj) and spot diameter on the retina (200 to 600 μπι) increased. This calculated value of 2.1 joules/cm 2 for the retinal energy density of the 1.06-μπι primary irradia tion was roughly fivefold less than values obtained when the energy was focused on the retina. 6,7 This increased sensitivity may be the result of "hot spots" in the
158
AMERICAN JOURNAL OF OPHTHALMOLOGY
beam caused by self-focusing of the pri mary radiation or of the larger-spot sizes on the retina in our experiment. 7 Allow ing for either of these effects, the damage thresholds we obtained appeared to be consistent with most of the energy being transmitted through the focus to the un derlying retina. The r2 dependence of the threshold energy on distance observed for Q-switched pulses (Fig. 2, upper curve) was expected for damage caused by a divergent beam from its point of focus. Alternatively, retinal damage asso ciated with the shock wave emanating from the laser focus in front of the retina may have accounted for the increased retinal sensitivity. The damage thresholds for modelocked pulses were less than those for Q-switched pulses and showed a more complicated dependence on distance from the retina. For pulse energies below 2 mj that created threshold lesions, the calculated retinal energy density was ap proximately 0.8 joules/cm 2 , twofold less than the reported values 6,7 for single mode-locked pulses focused to compara ble spot sizes on the retina. For pulse energies above 2 mj, the damage thresh olds became more scattered and in creased more rapidly than r2. Thus, the apparent retinal energy density for threshold damage increased by roughly twofold for 4-mJ pulse trains. This was consistent with 50% of the energy being absorbed by the plasma formed at the focus. We measured the nonlinear ab sorption of the plasma in water for these pulses and obtained variable results in the range of 50% absorption. Thus, the retina was apparently dam aged primarily by the primary radiation diverging from the focus and transmitted to the retina. The retina is roughly three fold more sensitive to mode-locked pulse energy than to Q-switched pulse energy at low pulse energies. At high pulse ener
AUGUST, 1983
gies the retina may be partially shielded from the mode-locked pulses by the con siderable absorption at the focus, but for our system this variable effect never made the mode-locked pulses less damag ing than the Q-switched pulses. We em phasize that our studies of damage thresholds were performed in clear vitre ous. In vitreous with mild haze, scatter ing of the beam increases the spot size at the focus. Thus, one effect of vitreal haze is to decrease the nonlinear absorption at the focus, resulting in greater retinal irradiance and damage behind the focus. Our thresholds for clinically observed lesions should not be interpreted as indi cating that all pulses below the thresholds presented in Figure 2 are safe. A hazy vitreous or smaller intraocular cone angle may substantially decrease the threshold for damage. Additionally, van der Zypen, Bebie, and Fankhauser 12 reported that as pulse energy increases the absorptive zone of neodymium-YAG pulses may grow anteriorly with respect to the helium-neon focal plane. Thus, for opti mal cutting of vitreal membranes the helium-neon aiming beam may be fo cused posteriorly to the membrane. In this case the threshold distance of the helium-neon focus above the retina (given in Fig. 2) would be less than the distance of the membrane above the retina. Further investigation is necessary to determine the long-term effects of this noninvasive surgery, such as regrowth of vitreal membranes and inflammatory re actions. Additionally, our inability to rup ture vitreal membranes in a hazy vitreous showed that the neodymium-YAG laser can be used only in carefully selected vitreal membranes. Although caution is needed in relating animal data to clinical situations, our study indicated that there is a substantial risk of retinal damage when ultrashort neodymium-YAG pulses
VOL. 96, NO. 2
RETINAL DAMAGE FROM YAG LASER
are used to rupture vitreal membranes close to the retina, especially near the fovea. REFERENCES 1. Aron-Rosa, D., Aron, J. J., Griesemann, M., and Thyzel, R. : Use of the neodymium-YAG laser to open the posterior capsule after lens implant surgery. A preliminary report. J. Am. Intraocul. Implant Soc. 6:352, 1980. 2. Aron-Rosa, D., Griesemann, J. C , and Aron, J. J. : Use of a pulsed neodymium-YAG laser (picosec ond) to open the posterior lens capsule in traumatic cataract. A preliminary report. Ophthalmic Surg. 12:496, 1981. 3. Peyman, G. A., Kraft, M., Viherkoski, E., and Ressler, N.: Noninvasive capsulectomy using a pulsed infrared laser. J. Am. Intraocul. Implant Soc. 8:239, 1982. 4. Fankhauser, F., Roussel, P., Steffen, J., van der Zypen, E., and Chrenkova, A.: Clinical studies on the efficiency of high power laser radiation upon some structures of the anterior segment of the eye. Int. Ophthalmol. 3:129, 1981. 5. Fankhauser, F., Lortscher, H., and van der Zypen, E.: Clinical studies on high and low power laser radiation upon some structures of the anterior
159
and posterior segments of the eye. Int. Ophthalmol. 5:15, 1982. 6. Ham, W. T., Mueller, H. A., Goldman, A. I., Newman, B. E., Holland, L. M., and Kuwabara, T.: Ocular hazards from picosecond pulses of ND:YAG laser radiation. Science 185:362, 1974. 7. Goldman, A. I., Ham, W. T., and Mueller, H. A.: Ocular damage thresholds and mechanisms for ultrashort pulses of both visible and infrared laser radiation in rhesus monkey. Exp. Eye Res. 24:45, 1977. 8. Topping, T. M., Abrams, G. W., and Machemer, R.: Experimental double-perforating injury of the posterior segment in rabbit eyes. Arch. Ophthalmol. 97:735, 1979. 9. Meyers, S. M., and Rodrigues, M. M.: Effect of selected intravitreal drugs after severe penetrating injury in rabbits. Curr. Eye Res. 1:471, 1981. 10. Bito, L. Z., DeRousseau, C. J., Kaufman, P. L., and Bito, J. W.: Age-dependent loss of ac commodative amplitude in rhesus monkeys. An ani mal model for presbyopia. Invest. Ophthalmol. Vis. Sei. 23:23, 1982. 11. Hughes, A.: A schematic eye for the rabbit. Vision Res. 12:123, 1972. 12. van der Zypen, E., Bebie, H., and Fankhauser, F.: Morphological studies about the efficiency of laser beams upon structures of the angle of the anterior chamber. Int. Ophthalmol. 1:109, 1979.