Investigative Ophthalmology & Visual Science, Vol. 32, No. 3, March 1991 Copyright © Association for Research in Vision and Ophthalmology Retinal Ganglion Cell Loss Is Size Dependent in Experimental Glaucoma Yoseph Glovinsky,* Harry A. Quigley,f and Gregory R. Dunkelbergerf Thirty-two areas located in the temporal midperipheral retina were evaluated in whole-mount prepara- tions from four monkeys with monocular experimental glaucoma. Diameter frequency distributions of remaining ganglion cells in the glaucomatous eye were compared with corresponding areas in the normal fellow eye. Large cells were significantly more vulnerable at each stage of cell damage as determined by linear-regression analysis. The magnitude of size-dependent loss was moderate at an early stage (20% loss), peaked at 50% total cell loss, and decreased in advanced damage (70% loss). In glaucomatous eyes, the lower retina had significantly more large cell loss than the corresponding areas of the upper retina. In optic nerve zones that matched the retinal areas studied, large axons selectively were damaged first. Psychophysical testing aimed at functions subserved by larger ganglion cells is recommended for detection and follow-up of early glaucoma; however, assessment of functions unique to small cells is more appropriate for detecting change in advanced glaucoma. Invest Ophthalmol Vis Sci 32:484-491, 1991 Current psychophysical tests do not detect glau- tage of ideal cellular preservation. Eyes with mild, comatous damage until a substantial minority of reti- moderate, and late damage were evaluated. In addi- nal ganglion cells have died.1'2 To develop more sen- tion, we correlated the damage patterns in the retinas sitive tests, a comprehensive understanding of the and optic nerves of the glaucomatous eyes. patterns of glaucomatous ganglion cell damage is es- sential. Retinal ganglion cells of different sizes have Materials and Methods distinct physiologic functions. Small cells that project to the parvocellular layers of the lateral geniculate This investigation adhered to the principles of the body belong to the "P pathway" or the "color sys- ARVO Resolution on Use of Animals in Research tem," while large cells that project to the magnocel- and was approved and monitored by the Institutional lular layers, belong to the "M pathway" or the "lumi- Animal Care and Use Committee of the Johns Hop- 3 kins University School of Medicine. nance system." Large optic nerve fibers selectively 7 are lost in chronic experimental and human glau- Monocular glaucoma was induced in four cyno- coma.4'5 Furthermore, in human glaucomatous eyes, molgus monkeys (Macaca fascicularis) by argon laser large ganglion cells die faster in studies of the mid- trabecular treatment. In another monkey, one optic peripheral retina.2 We studied the number and size of nerve was surgically transected 6 mm posterior to the remaining retinal ganglion cells in monkey eyes with globe. Both glaucomatous and transected animals experimental glaucoma, comparing them with nor- were monitored serially by applanation tonometry, slit mal fellow eyes. Chronic experimental glaucoma in lamp and fundus examination, and color stereopho- monkeys is similar to chronic human glaucoma both tography of the disc. After at least several months of clinically and histopathologically6 and has the advan- pressure elevation (intraocular pressure [IOP] range, 24-55 mm Hg; disease duration range, 6-24 months), various levels of optic nerve damage occurred, and From the tGlaucoma Service, Wilmer Ophthalmological Insti- disc pallor developed in the transected eye. tute, Johns Hopkins University School of Medicine, Baltimore, The monkeys were killed by exsanguination under Maryland, and the *Goldshleger Eye Institute, Sackler School of Medicine, Tel-Aviv University, Israel. sodium pentobarbital intravenous anesthesia. The Supported in part by PHS Research Grants EY 02120 (H.A.Q.), monkey with the transected eye was killed 6 months EY 01765 (Core Facility Grant, Wilmer Institute), a Senior Inves- after the operation. The eyes were then rapidly enu- tigator Award from Research to Prevent Blindness, Inc., New cleated and immersed in cold 2% paraformaldehyde York, New York, and by a Kreiger Foundation Grant. in 0.1 M phosphate buffer. The retinas were sepa- Submitted for publication: September 10, 1990; accepted Oc- tober 17, 1990. rated from the remainder of the globes and further Reprint requests: Harry A. Quigley, Maumenee Bl 10, Wilmer, fixed for at least 2 hr. The vitreous humor was re- Johns Hopkins Hospital, 600 N. Wolfe, Baltimore, MD 21205. moved mechanically as completely as possible. Five Downloaded from iovs.arvojournals.org on 09/30/2021 No. 3 RETINAL GANGLION CELL LOSS IN GLAUCOMA / Glovinsky er ol 485 to, seven relaxing incisions were made radially to 29 31 29 allow a flat mount of the retina, ganglion cell side up. <$) 2 Each retina was stained in 0.05% cresyl-violet. A 26 32 *; 29 31 cross-section of the retrolaminar optic nerve was i> 29 fixed in the same paraformaldehyde, postfixed in 1% 31 31 A 30 29 A * 8 osmium tetroxide, and embedded in epoxy resin. 32 32 32 32 33 30 29 One-micron sections of the nerve of the four glau- A I" 33 comatous and one transected eye were stained with 31 33 36 33 • 31 1 toluidine blue. The Zeiss IBAS image (Carl Zeiss, 34 32 6A 35 - • • 30 ' Inc., Thornwood, NY) analysis system was used to A i) measure the number and diameter distribution of 32 32 30 <§) 32 33 !• 32 30 fibers in 16 zones of the optic nerve.5 As expected, 30 31 33 34 31 there were no remaining fibers in the transected <$) nerve. The quantitative data from the glaucomatous 30 34 32 31 30 29 eyes were compared with a data base of ten normal cynomolgus nerves identically prepared and counted. 29 29 31 32 This seemed appropriate as our previous research has shown that substantial variation in fiber number Fig. 1. Humphrey field projection of the eight retinal areas (•) among monkeys can be partially compensated by studied in every monkey, plotted as a left eye. 1 mm on the retina using a larger sample of controls. equals 4° of visual angle. Distances from the fovea were 13°, 15°, 17°, and 21°. Ganglion cell diameter was determined by man- ually encircling at least 100 cells in each retinal area, using a camera lucida attached to a microscope at pared with their normal fellow eyes. Cell loss was 1000 X magnification and a planimeter connected to then correlated to the cell diameter by linear-regres- a personal computer. The diameter was calculated sion analysis.9 A significant correlation with positive from the circumference, assuming a circular shape. slope would indicate that large cells were more vul- Ganglion cell size distribution was determined in nerable. The slope of the regression line indicated the eight retinal areas of glaucomatous eyes and com- magnitude of the size-selective damage. A standard- pared statistically to the same areas in their normal ized regression coefficient10 was calculated by multi- fellow eyes. These eight areas corresponded to eight plying the regression coefficient by the ratio of the locations that are tested in the 30-2 program of the standard deviation of the independent variable to the Allergan-Humphrey Field Analyzer (San Leandro, standard deviation of the dependent variable. This CA) (Fig. 1). The cells were considered to be ganglion resulted in a dimensionless coefficient (beta) that rep- cells if they were round or oval and had large nuclei resents the slope of the regression line when both with some metachromatically stained cytoplasm. variables are expressed as standardized scores. Retinal capillary endothelial cells, pericytes, and as- trocytes that are found in the ganglion cell layer were Results easily excluded from counting. However, to estimate Normal Eyes the possibility that we included amacrine cells in our ganglion cell data, six areas of the retina from the The average ganglion cell densities in the different transected eye were similarly assessed. Since all gan- retinal locations were inversely correlated to their glion cells had atrophied, only amacrine cells would distance from the fovea (Fig. 2), ranging from 3705 be expected to remain and possibly to fulfill our ±556 cells/mm2 at 13° to 1476 ± 472 cells/mm2 at criteria. 21°. The variance among monkeys was substantial. In the normal eyes of monkeys 2 and 4, only the The 95% confidence interval width for pooled density upper retina was technically suitable for cell size mea- data of four upper retinal areas in the four monkeys surements. Since upper and lower size distributions was 26% of the mean value. In two normal eyes, we were found to be indistinguishable, we used the upper were able to compare the upper and the lower retina. retina as the control for both the upper and lower As previously reported," we found 14% greater den- areas in these monkeys. In all monkeys data could sity of cells in the four lower retinal areas compared not be obtained from the retina in the central 10° with their matching upper ones. The cell diameter (within 3 mm of the foveal center) due to the dense distribution peaked at about 14 ^m and had a posi- overlapping of cells. tive skewness toward the larger diameters. The size The percent cell loss was calculated for each diame- distribution of the upper and the lower areas had ter grouping (bin) in the glaucomatous eyes com- similar patterns (Fig. 3). Downloaded from iovs.arvojournals.org on 09/30/2021 486 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / March 1991 Vol. 32 4500 After studying their morphology, we tried to avoid counting them in the eight eyes that were evaluated in 4000- this study. The cell diameter distribution in the tran- 3500- sected eye overlapped that of the four normal eyes only at the 12-/nm range (Fig.