The Morphology and Function of Healing Cat Corneal Endothelium

Pefer T. Huang, Leif W. Nelson, and William M. Bourne

We mechanically damaged the entire corneal endothelium of one eye of each of ten cats and then examined both eyes by fluorophotometry and specular microscopy for 5 months. Six weeks after damage, when the corneas had cleared sufficiently to make accurate measurements, the mean endothelial permeability to carboxyfluorescein was increased 11% (P = 0.02) and the mean central corneal thickness was increased 11% (P - 0.05) in the damaged eyes. The mean endothelial pump rate was decreased 29% (P = 0.05), indicating that the increase in permeability was insufficient to explain the increase in thickness. The permeability returned to normal by 3 months and the pump rate by 5 months. Six weeks after damage, the mean endothelial cell size was increased 89% (P < 0.01), the mean coefficient of variation of cell size was increased 200% (P < 0.01), and the mean percentage of hexagonal cells was decreased 34% (P < 0.01). By 5 months, the mean cell size had changed very little, and none of the three morphologic measurements had returned to normal. As in rabbits, the endothelial barrier in cats recovers before the pump after wounding. Unlike in rabbits, functional recovery in cats requires at least several months. Such prolonged functional recovery after endothelial trauma might also be expected in humans who, like cats and unlike rabbits, have little capacity for endothelial mitosis during healing. Invest Ophthalmol Vis Sci 30:1794-1801,1989

Morphologic changes in the corneal endothelium Finally, we used carboxyfluorescein as a fluorescent -\ after wounding have been studied in rats,1 guinea tracer molecule rather than fluorescein because car- pigs,2 rabbits,3"8 cats,4'9"12 dogs,13 monkeys14"16 and, boxyfluorescein is less lipid-soluble and more likely to a limited extent, humans.17"19 Functional distur- to pass across the endothelium by the paracellular bances have been less well characterized, with endo- pathway.24 thelial barrier function during healing of severe wounds reported only in rabbits5 and monkeys.16 Materials and Methods Studies of barrier function during healing in humans We used ten adult cats weighing 3 to 4 kg for this have also been reported, but the wounding of the study. All procedures and measurements were per- endothelium was necessarily minor and not stan- 20 22 formed under anesthesia with xylazine (1 mg/kg) and dardized. " In order to better characterize corneal ketamine (5-8 mg/kg). Balanced salt solution was in- endothelial function during healing, we measured en- stilled frequently into all open eyes during anesthesia dothelial permeability after endothelial wounding in to maintain corneal hydration. We performed all cats, which, like humans, have limited mitotic capa- 4 procedures in accordance with the ARVO Resolution bilities. We damaged the endothelial cells mechani- on the Use of Animals in Research. On initial slit- cally in order to avoid injury to the corneal stroma, as lamp examination, all animals had clear corneas, no *••*• necessarily occurs with transcorneal freezing, and to evidence of trauma, absence of external infections or cells of the trabecular meshwork, iris and lens, as 23 intraocular inflammation, normal iris configurations occurs with intracameral benzalkonium chloride. and normal lenses.

Wounding From the Department of Ophthalmology, Mayo Clinic, Roches- ter, Minnesota. Several days after the baseline examination, the Supported in part by NIH grant EY-02037, Research to Prevent cats were fasted overnight and received 20 mg of gen- Blindness, Inc., New York, New York, and the Mayo Clinic and tamicin intramuscularly 1 hr preoperatively and 0.04 Foundation, Rochester, Minnesota. mg/kg atropine sulfate subcutaneously 30 min pre- Submitted for publication: July 15, 1988; accepted February 16, operatively. Anesthesia was then induced with xyla- 1989. Reprint requests: William M. Bourne, MD, Department of Oph- zine and ketamine hydrochloride intramuscularly. thalmology, Mayo Clinic, 200 First Street, SW, Rochester, MN Supplemental ketamine was used as necessary. One 55905. eye was randomly chosen for endothelial damage.

1794

Downloaded from iovs.arvojournals.org on 09/24/2021 No. 8 HEALING CAT CORNEAL ENDOTHELIUM / Huong er ol 1795

Under the operating microscope a 1 mm square grid disappearance of carboxyfluorescein from the cornea pattern was marked on the epithelial surface of the and the gradient between the stroma and the anterior cornea with a dye marker pen. An incision was then chamber for each of the two consecutive intervals made into the anterior chamber at the temporal (0800 to 1500 and 1500 to 0800) and the results limbus with a razor blade knife. Healon was placed averaged. The concentration of carboxyfluorescein in into the anterior chamber and a sterile Graether col- the stroma at any time was determined from a line fit lar button retractor (Storz Instrument, Co., St. Louis, by the method of least squares to the logarithms of MO) inserted into the anterior chamber. This instru- the stromal measurements versus time. The values of ment was then pressed firmly against the endothe- the stromal measurements were corrected for stromal lium along each line on the grid pattern, contacting thickness by multiplying them by a correction factor the entire endothelial surface from limbus to limbus. derived from measurements of an artificial cornea of At the completion of the procedure the incision site variable thickness with the fluorophotometer used in 26 was closed with one interrupted 10-0 nylon suture, this study, as outlined by McLaren and Brubaker. and 20 mg of gentamicin were injected subconjuncti- For a cat cornea of normal stromal thickness of 0.56 vally. The opposite eye was not disturbed. mm (estimated as 90% of the central corneal thick- •.V- ness of 0.62 mm), the correction fractor was 1.64. Measurements The correction factor was necessary because the measurement window of the fluorophotometer does not Anterior segment fluorophotometry with the fit entirely within the anterior-posterior boundaries Fluorotron Master with anterior segment modifica- of the cornea. Therefore, the fluorescence signal is tion (Coherent Medical Group, Palo Alto, CA), cor- underestimated and must be corrected. neal pachometry with an ultrasonic pachometer (Ac- The cornea-to-anterior chamber mass transfer co- cutome, Inc., Frazer, PA), calibrated for an acoustic efficient (kc.ca) was calculated for each time interval velocity in the corneal stroma of 1590 m/sec,25 to- from the following equation, derived from those of nometry with a pneumatonometer (Alcon, Inc., Fort 27 Jones and Maurice : Worth, TX), and endothelial photography with a wide-field specular microscope (Keeler Instruments, ACC Inc., Broomall, PA) were performed on both eyes of — (r C -C )At each cat before wounding (baseline) and 6 weeks, 3 ca a c months, 4 months and 5 months after wounding. where ACC is the change in concentration of carboxy- Each session comprised 3 days. On day 1, autofluo- fluorescein in the stroma over the time interval, At, rescence of the cornea and anterior chamber was Cc and Ca are the average concentrations of carboxy- measured at 1500. One drop of 5% 5(6)-carboxyfluo- fluorescein in the stroma and anterior chamber dur- rescein (Eastman Kodak, Rochester, NY, no. 134 ing the time interval, t, to t2, calculated on the as- 9448) was instilled, and 3 min later the cornea was sumption that the loss of dye is a first order process: rinsed with balanced salt solution and the corneal boundary function (apparent fluorescein in the ante- C = c,-c2 rior chamber due to fluorescein in the cornea) mea- In (C,/C2) sured. One drop of 5% carboxyfluorescein was then t is time, and rca is the stroma/anterior chamber dis- instilled every 5 min for 15 min and the lids closed for tribution ratio of carboxyfluorescein at equilibrium, an additional 15 min. Then the eye and lids were assumed to be constant and equal to 1.6, the value rinsed with balanced salt solution to remove residual found in rabbits.28 dye.^ Fluorescence of the central cornea and anterior Endothelial permeability to carboxyfluorescein chamber were measured at 0800 and 1500 on day 2 was calculated from the following relationship29: and at 0800 on day 3. At each session, three consecutive fluorophotometric scans were performed per eye Permeability = kc.ca X q X rca and the results averaged. Appropriate corrections for autofluorescence and boundary function were made. where q is the stromal thickness, estimated as 90% of After fluorophotometry on day 3, pachometry, to- the total central corneal thickness as measured with the ultrasonic pachometer. The thickness of the cen- nometry and endothelial photography were per- 30 formed. tral and peripheral cornea in the cat are similar, as is the case in rabbits.31 Therefore, the central stromal thickness, which is 90% of the total central corneal Calculations thickness1632 suffices as an estimate of the mean The corneal endothelial permeability of each eye to thickness of the entire stroma in which the carboxy- carboxyfluorescein was calculated from the rate of fluorescein is distributed.

Downloaded from iovs.arvojournals.org on 09/24/2021 1796 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / Augusr 1989 Vol. 00

The rate of clearance of carboxyfluorescein from Inherent in assumptions 1 and 3 is the following the anterior chamber was determined as the average steady-state relationship: of that calculated for each of the two consecutive measurement intervals, according to the following Pump rate = leak rate = SP X Lp (0 relationship27: where SP is the stromal swelling pressure and Lp is AC V + AC V the hydraulic conductivity of the endothelium. The rate of clearance = C C a a swelling pressure can be derived from the corneal C At a thickness and the hydraulic conductivity is related to where Va and Vc are the geometric volumes of the the endothelial permeability to carboxyfluorescein anterior chamber and corneal stroma, respectively. (assumption 2). ACa is the change in concentration in the anterior Because adequate data for the relationship between chamber over the time interval, At. Volumes were corneal thickness, hydration and swelling pressure for estimated for each measurement session by employ- the cat cornea were not available, we used existing ing the geometric formula for the volume of a spheri- data from two other mammalian species to derive an cal segment: expression that we assume is valid for all mammalian 2 2 corneas. This required three additional assumptions Va = 1/6 7rh(h + 3/4 y ) about mammalian corneas: where h is the axial depth of the anterior chamber as measured with the Haag-Streit Pachometer II cor- 4. Clear corneas of normal thickness have the rected for anterior corneal curvature and y is the di- same hydration. ameter of the chamber, taken as the external horizon- 5. The relationship between stromal hydration tal limbal diameter less two corneal thicknesses. The and swelling pressure is the same in all corneas. total corneal volume was estimated as the amount by 6. The relationship between hydration and nor- which the anterior segment volume exceeded Va. The malized thickness is the same in all corneas. anterior segment volume was estimated by employ- ing the axial chamber depth plus the central corneal Several investigators have found similar hydrations thickness as h and the external limbal diameter as y. (approximately 3.4 mg water/mg dry weight) in cor- Ninety percent of the total corneal volume was taken neas of normal thickness in rabbits, cats, and cows,34 as the stromal volume, Vc, in which the carboxy- and in humans,35 justifying assumption 4. Assump- fluorescein was distributed. tion 5 is justified by Hedbys and Dohlman,36 who Ten percent of the clearance of carboxyfluorescein found similar swelling pressure-hydration relation- was assumed to be due to diffusional losses and 90% ships over a wide range of hydrations in rabbits, cows, due to the flow of aqueous humor through the ante- and humans. To determine if assumption 6 is reason- rior chamber: rate of flow of aqueous humor = 0.9 able for mammalian corneas, we tested it with data X rate of clearance of carboxyfluorescein. from two species for which the relationship between In an attempt to determine whether changes in hydration and corneal thickness is known. Ytteborg corneal thickness could be explained by the changes and Dohlman35 found the following relationship for in endothelial permeability alone, or whether a human corneas: change in the endothelial fluid pump33 may have also occurred, we calculated the endothelial pump rate for CThuman = 0.9 +0.142 H (2) each cornea at each examination, based upon the fol- where CT = central corneal thickness (including epi- lowing assumptions: thelium) in mm and H = hydration in mg water/mg 1. The rate of water leakage through the endothe- dry weight. Hedbys and Mishima37 found the follow- lium into the stroma is proportional to the hydraulic ing relationship for bovine corneas: conductivity of the endothelium and to the pressure CT = 0.126 +0.189 H gradient across the endothelium (the stromal swelling bovine (3) pressure). Given a normal hydration of 3.4 mg water/mg dry 2. The hydraulic conductivity of the endothelium weight for both species, the normal corneal thickness is proportional to its permeability to carboxyfluores- for humans from equation (2) is 0.57 mm and for cein. cows from equation (3) is 0.77 mm. To normalize the 3. In the steady state (constant corneal thickness), corneal thickness, as in assumption 6, we divided the rate of water leakage into the stroma from the both sides of equation 2 by 0.57 mm and both sides of aqueous humor equals the endothelial pump rate equation (3) by 0.77 mm. The resulting equations for (transfer of water across the epithelium and limbus is human and bovine corneas were identical to two dec- neglected). imal places, providing justification for assumption 6.

Downloaded from iovs.arvojournals.org on 09/24/2021 No. 8 HEALING CAT CORNEAL ENDOTHELIUM / Huong er ol 1797

Therefore, we assumed that cat corneas would also Optics, Inc., Arlington, MA). The cell density (cells/ conform to the derived relationship: mm2) for each cornea was calculated by dividing the mean cell size (^m2) into one million. = 0.16 +0.25 (H) (4) Statistical comparisons were made with a two- C^ Ti n tailed student t-test for paired samples and the Spear- and: man test for correlation; P < 0.05 was considered statistically significant. CT •3 (5) Results

where CTj is the individual corneal thickness and CTn is the normal corneal thickness for that mammalian There were no significant differences between right species. and left eyes at the baseline examination for any of The empirical relationship between swelling pres- the measured values (Table 1). After the mechanical sure in mm Hg and stromal hydration in mg water/ endothelial damage, the corneas became thick and mg dry weight was derived by Burns et al38 from the hazy. There was a mild anterior chamber inflamma- experimental data of Ytteborg and Dohlman35: tory reaction that cleared within 48-72 hr. There was epithelial edema, but no epithelial defect, in the dam- SP = 2.193(6-H)3.275 (6) aged eyes. The corneal haze cleared gradually with no vascularization. The haze was completely gone by 6 The relationship between swelling pressure and cor- weeks post-damage so that it was possible to perform neal thickness, assumed valid for all mammalian cor- fluorophotometry. Corneal thickness was not mea- neas, is derived by substituting equation (5) into sured until the 6 week examination except for two equation (6): cats measured 4 days after wounding, when the CT;\ 3.275 thicknesses in the two experimental eyes were 1.26 SP = 205.5 1.66 (7) mm and 1.02 mm. The iris and lens remained nor- CTnj mal in appearance. i- From assumption 2, As shown in Table 1 and Figure 1, the central corneal thickness and the endothelial permeability to LPi = (Pi/Pn) X LPn (8) carboxyfluorescein were significantly increased 6 where P is endothelial permeability and the subscripts weeks after endothelial damage. Both then returned i and n indicate individual measurements and nor- toward normal and were not significantly different mal values, respectively, for that species. A general- from baseline values 3 to 5 months after wounding. ized expression for the mammalian cornea is ob- The calculated endothelial pump rate remained de- tained by substituting equations (7) and (8) into creased by more than 25% for 4 months after wound- equation (1): ing (P = 0.031). The pump rate gradually returned to normal by 5 months post-damage. Pump rate = leak rate As shown in Table 1 and Figure 2, the aqueous 3.275 p humor flow rate remained significantly increased for = 205.5^1.66- X^XLPn (9) 5 months after wounding. The intraocular pressure CTn was significantly decreased 6 weeks postoperatively, For the normal values CTn and Pn in this study, we but returned to normal 3 months post-damage. used the mean values of all 20 eyes at the baseline As shown in Table 1 and Figure 3, 6 weeks after examination. The normal value for pump rate was mechanical injury the mean endothelial cell size had calculated as a factor times Lpn from equation (9) by almost doubled (P = 0.001), and it did not decrease setting CTj/CTn and Pj/Pn equal to one. Individual significantly over the ensuing 3.5 months. There was relative pump rates were expressed as a percentage of no significant correlation between mean cell size and the normal value. corneal thickness at any of the examinations. The Morphologic analysis of the corneal endothelium percentage of hexagonal cells dropped from 78% at was accomplished by digitizing the apices of 100 cells baseline to 51% 6 weeks after damage (P = 0.001). of each cornea from photographic images enlarged This percentage increased to 60% by 5 months after 500 times. The mean and standard deviation of cell wounding, although it remained significantly less size, the coefficient of variation of cell size (standard than the baseline value or that of the control eyes. deviation/mean) and the percentage of cells with five, The coefficient of variation of cell size tripled after six and seven sides were determined from the digi- mechanical damage and remained elevated 5 months tized data with an endothelial analysis system (Bio- after wounding.

Downloaded from iovs.arvojournals.org on 09/24/2021 o

Table 1. Findings in both eyes often cats after unilateral endothelial wounding Time after damage

Baseline 6 Weeks 3 Months 4 Months 5 Months

Measurement Eye Mean ± SD Mean ± SD P* Mean ± SD P* Mean ± SD P* Mean ± SD P*

-z.VESTK

Corneal thickness (mm) Damaged 0.62 ± 0.05 0.69 ±0.12 0.048 0.65 ± 0.07 0.075 0.66 ± 0.07 0.087 0.63 ± 0.04 0.223 Control 0.62 ± 0.05 0.63 ± 0.04 0.520 0.62 ± 0.05 0.591 0.66 ± 0.06 0.106 0.63 ± 0.05 0.236 > m Endothelial permeability Damaged 2.31 ±0.31 2.56 ±0.35 0.021 2.25 ± 0.34 0.570 2.29 ± 0.32 0.863 2.19 ±0.30 0.261 o (Xl(r4cm/min) Control 2.26 ±0.29 2.41 ±0.36 0.280 2.36 ± 0.32 0.541 2.29 ±0.31 0.800 2.36 ± 0.24 0.055 TJ

HAU —1 Endothelial pump rate Damaged 1.01 ±31 , 72 ±50 0.052 75 ±41 0.040 74 ±40 0.031 92 ±36 0.132 (% normal) Control 99 ±31 99 ±32 0.947 98 ±33 0.961 78 ±32 0.075 95 ±28 0.632 O Aqueous humor flow rate Damaged 6.50 ± 1.00 8.02 ± 1.72 0.002 7.46 ± 1.64 0.127 7.59 ± 1.62 0.005 7.44 ± 1.63 0.046 O Oxl/min) Control 6.82 ± 1.49 6.96 ± 2.35 0.781 6.95 ± 1.63 0.801 6.88 ± 1.63 0.894 7.16 ± 1.17 0.508 <. Intraocular pressure Damaged 23.6 ± 3.63 18.5 ±6.17 0.011 20.9 ± 7.05 0.215 21.7 ±7.06 0.555 22.3 ± 6.53 0.583 >^, s> (mm.Hg) Conrol 24.2 ± 4.32 24.1 ±4.65 0.914 22.5 ± 5.74 0.279 22.0 ± 7.66 0.626 23.1 ±6.17 0.524 I— Yk Mean endothelial cell Damaged 390 ± 46 736 ± 159 0.001 676 ± 126 0.001 690 ± 154 0.001 670 ± 158 0.001 m i i size (urn) Control 396 ± 50 387 ± 36 0.509 417 ± 31 0.060 415 ±37 0.167 406 ± 20 0.550 m > Coefficient of variation Damaged 0.14 ±0.01 0.42 ±0.16 0.001 0.42 ±0.18 0.001 0.35 ±0.12 0.001 0.37 ± 0.08 0.001 of cell size Control 0.14 ±0.01 0.15 ±0.02 0.611 0.14 ±0.02 0.68 0.13 ±0.02 0.585 0.15 ±0.01 0.323

Hexagonal cells (%) Damaged 77.7 ± 12.09 51.0 ±6.78 0.001 54.8 ± 6.99 0.001 57.4 ± 10.67 0.005 59.7 ±6.83 0.001 00 Control 80.0 ± 8.46 74.4 ± 6.20 0.117 77.5 ±9.18 0.504 82.8 ± 5.88 0.613 76.6 ± 7.37 0.636

Endothelial cell Damaged 2606 ± 343 1411 ±277 0.000 1530 ±313 0.000 1523 ±375 0.000 1564 ± 353 0.000 density (cells/mm2) Control 2559 ±321 2602 ± 249 0.654 2410 ± 176 0.069 2426 ±215 0.207 2467 ± 123 0.646

* Paired t-test, 2-tailed, versus baseline. Vol. 30

Downloaded from iovs.arvojournals.org on 09/24/2021 No. HEALING CAT CORNEAL ENDOTHELIUM / Huong er ol 1799

Corneal thickness /J

12 3 4 5 0 12 3 4 Months after damage Months after damage Fig. 1. Changes in corneal endothelial function for 5 months Fig. 2. Changes in aqueous humor flow rate and intraocular after endothelial wounding. Mean percent changes are plotted; pressure for 5 months after endothelial wounding. Mean percent vertical lines indicate standard errors of the mean. changes are plotted; vertical lines indicate standard errors of the mean.

None of the mean values in the control eyes were cat, the return of endothelial function follows a simi- significantly different from baseline over the 5 month lar pattern in both species. Both barrier and pump period of measurement. Five months after wounding, function are reestablished sooner in the rabbit, how- the geometric volumes of the corneal stroma (control: ever; this perhaps is because of cellular mitosis in that 137 ± 10 ^1, experimental: 137 ± 10 /A) and anterior species. That the endothelial barrier function was chamber (control: 401 ± 36 /ul, experimental: 403 reestablished before the pump function may indicate ± 37 /xl) had not changed significantly from baseline. that the pump is more delicate and susceptible to damage from trauma or disease than is the barrier. Discussion The recent finding of decreased endothelial pump function in the presence of normal barrier function in We developed a model for endothelial wound patients with moderately advanced Fuchs' dys- healing in the cat. The adult cat was chosen because, 39 17 19 trophy lends support to this possibility. Tsuru et as with humans, " the healing process occurs by 16 al, however, found that the corneal thickness re- hypertrophy and migration of cells4'910 rather than turned to normal before endothelial permeability in mitosis, as seen in the rabbit.3"7 We chose mechanical monkey corneas after wounding. The pump might damage rather than cryodamage because we wished also be decreased after wounding if there is not to damage only endothelial cells and not Descemet's 8 enough lateral membrane to contain the pump sites. membrane or stroma. Freezing the cornea damages The gradual increase in pump function may reflect the stroma, and we reasoned that such damage may alter either rca or the stromal swelling pressure-hydration-thickness relationship. Nevertheless, our re- 240 sults were similar to those found in rabbit corneas after freezing in that endothelial permeability re- 200 turned to normal before corneal thickness, although the return to normality of both values was faster in the rabbit.5 The finding of decreased endothelial pump rate in the presence of increased endothelial permeability 6 weeks after wounding indicates that the increase in permeability was insufficient to account for the increase in corneal thickness; thus, the pump must also have been involved. Because endothelial permeability to carboxyfluorescein returned to normal before

corneal thickness, the calculated pump rate remained 1 2 3 4 ' J 5 depressed until corneal thickness returned to normal Months after damage levels. Fig. 3. Changes in corneal endothelial morphology for 5 months Although cellular mitosis plays a significant role after endothelial wounding. Mean percent changes are plotted; during endothelial healing in the rabbit and not in the vertical lines indicate standard errors of the mean.

Downloaded from iovs.arvojournals.org on 09/24/2021 1800 INVESTIGATIVE OPHTHALMOLOGY b VISUAL SCIENCE / August 1989 Vol. 30

an increase in the area of the lateral membranes as Because human corneal endothelium is more simi- they become more convoluted. ' lar to that of cats than rabbits, delayed functional Given the many assumptions necessary to calcu- recovery in human corneas after endothelial wound- late our results, they should be considered only as ing might also be expected. This may explain clinical estimates. For example, we assumed that the cornea observations by one of us (WMB) of increased cor- to anterior chamber steady-state distribution ratio for neal thickness for several months in corneal trans- carboxyfluorescein, rca, was the same in cats as that plants after episodes of endothelial trauma, despite found in rabbits. We also assumed that rca was con- adequate numbers of endothelial cells in uninflamed stant, not varying with stromal hydration. Neither of eyes. these assumptions may be correct, although we have We noted that the intraocular pressure dropped no data to the contrary. Although the absolute values significantly 6 weeks after endothelial damage and of our estimates may not be correct, they nevertheless that the aqueous humor flow rate increased signifi- are quite adequate for the comparisons between eyes cantly and remained so for 5 months. We have no for which we have used them. valid explanation for these findings at present. We chose to report only relative endothelial pump rates (percentage of normal or baseline values) be- Key words: endothelial permeability, endothelial pump, cat cause of the many assumptions necessary for their cornea, endothelial wound, carboxyfluorescein calculation. By assuming a constant relationship between stromal hydration and swelling pressure (as- References sumption 5 above), we have ignored the possibility 1. Tuft SJ, Williams KA, and Coster DJ: Endothelial repair in the that stromal proteoglycans may be decreased after rat cornea. Invest Ophthalmol Vis Sci 27:1199, 1986. endothelial damage40'41 or have altered water sorptive 2. Ilmonen M, Lehtosalo JI, Virtanen J, Uusitalo H, and Pal- capacity.42 Nevertheless, an indication of the propri- kama A: Initial healing of the posterior corneal surface following perforating trauma in guinea pig: A scanning electron mi- ety of these assumptions can be obtained by calculat- croscope study. Acta Ophthalmol 62:787, 1984. ing the normal pump rate for cats from equation (9) 3. Faure JP, Kim YZ, and Graf B: Formation of giant cells in the 43 using the value reported by Stanley et al for normal corneal endothelium during its regeneration after destruction 4 cat hydraulic conductivity (Lpn) of 9.6 X 10~ mm by freezing. Exp Eye Res 12:6, 1971. hr"1 mm Hg~'.38 The calculation yields a value of 51 4. Van Horn DL, Sandile DD, Leideman S, and Buco PJ: Re- /zm hr"1, which is similar to the endothelial pump generation capacity of the corneal endothelium in rabbit and 33 cat. Invest Ophthalmol 16:597, 1977. rate found experimentally in rabbits. 5. Minkowski JS, Bartels SP, Delori FC, Lee SR, Kenyon KR, In the human,17"19 primate1416 and cat,49"" other and Neufeld AH: Corneal endothelial function and structure investigators have found that repair following endo- following cryo-injury in the rabbit. Invest Ophthalmol Vis Sci thelial damage occurs by a process of migration and 25:1416, 1984. 6. Olsen EG and Davanger M: The healing of rabbit corneal hypertrophy. We found a similar process of healing. endothelium. Acta Ophthalmol 62:796, 1984. The endothelial cells almost doubled in size and re- 7. Matsuda M, Sawa M, Edelhauser HF, Bartels SP, Neufeld AH, mained so. The progress toward stability after and Kenyon KR: Cellular migration and morphology in cor- wounding is reflected by the gradual increase in bex- neal endothelial wound repair. Invest Ophthalmol Vis Sci agonality and decrease in coefficient of variation of 26:443, 1985. 8. Yee RW, Geroski DH, Matsuda M, Champeau EJ, Meyer LA, cell size over a 5 month period. Similar results were and Edelhauser HF: Correlation of corneal endothelial pump found in human corneal endothelial cells after pene- site density, barrier function, and morphology in wound re- 44 trating keratoplasty. pair. Invest Ophthalmol Vis Sci 26:1191, 1985. Landshman et al found in cats that if the corneal 9. Ogita Y, Higuchi S, Kari K, and Honda N: Wound healing of endothelial cell density was reduced to 45% of its the endothelium of the living cat cornea: A specular microscopic study. Jpn J Ophthalmol 25:326, 1981. preoperative value (or the mean cell size increased to 10. Honda H, Ogita Y, Higuchi S, and Kari K: Cell movements in 2.2 times its preoperative value), then the corneal a living mammalian tissue: Long-term observation of individ- thickness was increased." We found no correlation ual cells in wounded corneal endothelia of cats. J Morphol between mean cell size and corneal thickness at any 174:25, 1982. time. Five months after wounding, however, none of 11. Landshman N, Ben-Hanan I, Assia E, Ben-Chaim O, and Bel- kin M: Relationship between morphology and functional abil- the corneas had mean cell sizes at least 2.2 times ity of regenerated corneal endothelium. Invest Ophthalmol Vis baseline. Six weeks after wounding, the mean cell size Sci 29:1100, 1988. was at least 2.2 times baseline in three corneas; the 12. Ling T, Vannas A, and Holden BA: Long-term changes in mean thickness of these three corneas was less than corneal endothelial morphology following wounding in the that of the other seven, and in the one with the largest cat. Invest Ophthalmol Vis Sci 29:1407, 1988. 13. Befanis PJ, Peiffer Jr RL, and Brown D: Endothelial repair of increase in cell size (2.9 times baseline), the thickness the canine cornea. Am J Vet Res 42:590, 1981. was the same as the baseline value. Thus, our results 14. Van Horn DL and Hyndiuk RA: Endothelial wound repair in do not support the conclusions of Landshman et al." primate cornea. Exp Eye Res 21:113, 1975.

Downloaded from iovs.arvojournals.org on 09/24/2021 No. 8 HEALING CAT CORNEAL ENDOTHEUUM / Huong er ol 1801

15. Matsubara T and Tanishima T: Wound healing in the mon- fluorophotometry nomogram. Invest Ophthalmol Vis Sci v key: A morphometric study. Jpn J Ophthalmol 26:264, 1982. 18:288, 1979. 16. Tsuru T, Araie M, Matsubara M, and Tanishima T: Endothe- 30. Ling T: Osmotically induced central and peripheral corneal lial wound-healing of monkey cornea: Fluorophotometric and swelling in the cat. Am J Optom Physiol Opt 64:674, 1987. specular microscopic studies. Jpn J Ophthalmol 28:105, 1984. 31. Chan T, Payor S, and Holden BA: Corneal thickness profiles in 17. Doughman DJ, Van Horn DL, Rodman WP, Byrnes P, and rabbits using an ultrasonic pachometer. Invest Ophthalmol Vis Lindstrom RL: Human corneal endothelial layer repair during Sci 24:1408, 1983. organ culture. Arch Ophthalmol 94:1791, 1976. 32. Maurice DM: Cornea and sclera. In The Eye, Vol. lb, Davson 18. Treffers WF: Human corneal endothelial wound repair: In H, editor. London, Academic Press, 1984, pp. 1-158. vitro and in vivo. Ophthalmology 89:605, 1982. 33. Maurice DM: The location of the fluid pump in the cornea. J 19. Olsen EG and Davenger M: The healing of human corneal Physiol 221:43, 1972. endothelium: An in vitro study. Acta Ophthalmol 62:885, 34. Duane TD: The steady state of corneal hydration. Am J 1984. Ophthalmol 32(Part II):203, 1949. 20. Ota Y: Endothelial permeability to fluorescein in corneal grafts 35. Ytteborg J and Dohlman C: Corneal edema and intraocular and bullous keratopathy. Jpn J Ophthalmol 19:286, 1975. pressure: II. Clinical results. Arch Ophthalmol 74:477, 1965. 21. Sawa M, Araie M, and Tanishima T: Permeability of the cor- 36. Hedbys BO and Dohlman CH: A new method for the determi- neal endothelium to fluorescein: A follow-up of keratoplasty nation of the swelling pressure of the corneal stroma in vitro. cases. Jpn J Ophthalmol 26:326, 1982. Exp Eye Res 2:122, 1963. 22. Sawa M, Araie M, and Tanishima T: A fluorophotometric 37. Hedbys BO and Mishima S: Flow of water in the corneal study of the barrier functions in the anterior segment of the eye stroma. Exp Eye Res 1:262, 1962. after intracapsular cataract extraction. Jpn J Ophthalmol 38. Burns R, Bourne WM, and Brubaker RF: Endothelial function 27:404, 1983. in patients with cornea guttata. Invest Ophthalmol Vis Sci 20:75, 1981. 23. Maurice D and Perlman M: Permanent destruction of the 39. Bourne WM and Wilson SE: Effect of dexamethasone on cor- corneal endothelium in rabbits. Invest Ophthalmol 16:646, neal endothelial function in Fuchs' dystrophy. Invest Oph- 1977. thalmol Vis Sci 29:357, 1988. 24. Grimes PA, Stone RA, Laties AM, and Li W: Carboxyfluores- 40. Anseth A: Studies on corneal polysaccharides: V. Changes in cein: A probe of the blood ocular barriers with lower mem- corneal glycosaminoglycans in transient stromal edema. Exp brane permeability than fluorescein. Arch Ophthalmol Eye Res 8:297, 1969. 100:635, 1982. 41. Kangas TA, Edelhauser HF, Twining SS, and O'Brien J: Loss 25. Ling T, Ho A, and Holden BA: Method of evaluating ultra- of stromal proteoglycans during corneal edema. ARVO Ab- sonic pachometers. Am J Optom Physiol Opt 63:462, 1986. stracts. Invest Ophthalmol Vis Sci 29(Suppl):215, 1988. 26. McLaren JW and Brubaker RF: A scanning ocular spectro- 42. Castoro JA, Bettelheim AA, and Bettelheim FA: Water gra- fluorophotometer. Invest Ophthalmol Vis Sci 29:1285, 1988. dients across bovine cornea. Invest Ophthalmol Vis Sci 29:963, 27. Jones RF and Maurice DM: New methods of measuring the 1988. rate of aqueous flow in man with fluorescein. Exp Eye Res 43. Stanley JA, Mishima S, and KJyce SD: In vivo determination 5:208, 1966. of endothelial permeability to water. Invest Ophthalmol 5:371, 28. Araie M: Carboxyfluorescein: A dye for evaluating the corneal 1966. endothelial barrier function in vivo. Exp Eye Res 42:141, 1986. 44. Matsuda M and Bourne WM: Long-term morphologic 29. Coakes RL and Brubaker RF: Method of measuring aqueous changes in the endothelium of transplanted corneas. Arch humor flow and corneal endothelial permeability using a Ophthalmol 103:1343, 1985.

Downloaded from iovs.arvojournals.org on 09/24/2021