CENTRAL CORNEAL ENDOTHELIAL CELL DENSITY AND CENTRALCORNEAL THICKNESS IN OCULAR HYPERTENSION AND PRIMARY

OPEN-ANGLE GLAUCOMA

MICHAEL KOREY, M.D., DAVID GIESER, M.D., MICHAEL A. Kxss, M.D.,STEPHEN R. WALTMAN, M.D., MAE GORDON, PH.D.,

AND BERNARD BECKER, M.D.St. Louis, Missouri

In order to assess the effect of increased intraocular pressure on thecorneal endothelium, we classified 254 patients into four groups: Group1, those with normal intraocular pressures; Group 2, those withuntreated ocular hypertension; Group 3, those with treated ocularhypertension; and Group 4, those with primary open-angle glaucoma.One eye of each patient underwent specular microscopy and pachymet­ry. The eyes in the four groups did not differ significantly as to centralcorneal endothelial cell density or central corneal thickness. Thesemeasurements were not related to sex, race, or intraocular pressure (P> .12 in all cases). Increasing age was associated with a decrease incentral corneal endothelial cell density (P = .0(01), but was notassociated with a change in central corneal thickness (P = .22). Therewas no significant relationship between the use of topical ocularhypotensive medications and central corneal endothelial cell density (P= .38) or central corneal thickness (P = .07) in patients with ocularhypertension or primary open-angle glaucoma. Neither uncomplicatedperipheral iridectomy nor trabeculectomy produced significant changeswhen preoperative measurements were compared to measurementsmade 12 weeks postoperatively (P > .30 in all cases). Two eyes with flatanterior chambers following trabeculectomy had substantial decreasesin central corneal endothelial cell density.

There is evidence suggesting that in­creased intraocular pressure and medicaland surgical treatment of glaucoma canaffect the structure and function of thecorneal endothelium. Corneal edema oc-

Accepted for publication Aug. 13, 1982.From the Department of Ophthalmology, Wash­

ington University School of Medicine, St. Louis,Missouri. This study was supported in part by grantEY 00004 from the National Eye Institute and by anunrestricted grant from Research to Prevent Blind­ness, Inc., New York, New York.

Reprint requests to Michael A. Kass, M.D., De­partment of Ophthalmology, Washington UniversitySchool of Medicine, 660 S. Euclid Ave., St. Louis,MO 63110.

curs in some eyes with glaucoma, particu­larly in eyes that experience a rapid andmarked increase in intraocular pressure.Decreased central corneal endothelialcell density occurs in eyes with acuteangle-closure glaucoma.P glaucomato­cyclitic crisis," and glaucoma capsulare."Experimentally induced ocular hyperten­sion produces morphologic changes inrabbit! and monkey'' corneal endotheliumthat may be associated with decreasedcorneal clarity and increased central cor­neal thlckness.F' Topical or intracameraladministration of ocular hypotensivemedications and their preservatives candamage the corneal endothelium in

610 ©AMERICAN JOURNAL OF OPHTHALMOLOGY 94:610-616, 1982

VOL. 94, NO.5 CORNEAL ENDOTHELIUM 611

human and animal eyes.r" Peripheral iri­dectomy and trabeculectomy can producea decrease in central corneal endothelialcell density. 1 Although these reports sug­gest that increased intraocular pressureand antiglaucoma treatment can have ad­verse effects on the corneal endothelium,there are no published studies evaluatingthese matters systematically.

Specular mieroscopy'<P and pachy­metry" allow in vivo examination of cor­neal endothelial morphology and func­tion. Our purpose was to use thesetechniques to determine whether in­creased intraocular pressure, primaryopen-angle glaucoma, and medical andsurgical glaucoma therapy affect centralcorneal endothelial cell density and cen­tral corneal thickness.

SUBJECTS AND METHODSA total of 254 patients participated in

this study after giving informed consents.If both eyes of a patient were eligible forinclusion in the study, we randomlychose one eye. All subjects had at leastthree intraocular pressure measurementswith a minimum follow-up period of sixmonths (range, six to 278 months).

We divided the 254 patients into fourgroups (Table 1). Group I-We classified103 patients as having normal intraocularpressures on the basis of normal visualfields, normal optic disks, and intraocularpressure readings of::5 20 mm Hg. Group2-We classified 62 patients as havingocular hypertension on the basis of threeor more intraocular pressure measure­ments of 21 mm Hg or higher, normal

TABLE 1

CLINICAL FINDINGS

Group 1· Group 2· Group 3· Group 4·Clinical Findings (No. = 103) (No. = 62) (No. = 35) (No. = 54)

Age (yrs)Mean ± S.D. 54.5 ± 13.4 57.3 ± 10.3 58.9 ± 12.8 64.5 ± 10.6Range 22 to 78 29 to 74 27 to 79 30 to 80

SexMale 44 (42.7%) 35 (56.4%) 16 (45.7%) 23 (42.6%)Female 59 (57.3%) 27 (43.6%) 19 (54.3%) 31 (57.4%)

RaceWhite 95 (92.7%) 55 (88.7%) 23 (65.7%) 24 (44.4%)Black 8 (7.8%) 7 (11.3%) 12 (34.3%) 30 (55.6%)

Duration of disease (mos)Mean ± S.D. 129.2 ± 71.1 121.4 ± 71.2 64.6 ± 40.6Range 6 to 272 21 to 278 6 to 158

Intraocular pressure (mm Hg)At time of pachymetry

Mean ± S.D. 17.0 ± 2.7 23.8 ± 3.5 22.6 ± 3.9 20.3 ± 4.7Range 11 to 20 14 to 31 15 to 32 12 to 32

During follow-upMean ± S.D. 16.0 ± 2.0 22.4 ± 2.3 24.1 ± 3.6 22.6 ± 3.3Range 11 to 20 18 to 28 18 to 32 15 to 31

Central corneal endothelial celldensity (cells/mm')Mean ± S.D. 2,117.4 ± 373.2 2,164.4 ± 330.6 2,120.5 ± 291.8 2,079.0 ± 270.6Range 1,345 to 3,508 1,591 to 3,152 1,506 to 2,270 1,424 to 2,769

Central corneal thickness (mrn)Mean ± S.D. 0.52 ± 0.02 0.52 ± 0.01 0.52 ± 0.04 0.51 ± 0.02Range 0.44 to 0.61 0.47 to 0.56 0.45 to 0.67 0.46 to 0.61

"Croup 1, normal; Group 2, untreated ocular hypertension; Group 3, treated ocular hypertension; Group 4,primary open-angle glaucoma.

612 AMERICAN JOURNAL OF OPHTHALMOLOGY NOVEMBER, 1982

OCULAR HYPOTENSIVE MEDICATIONS

TABLE 2

*Some patients received more than one medica­tion.

optic disks, and normal visual fields. Themean intraocular pressure for a givenpatient included all values recorded dur­ing follow-up for that individual. Thesepatients had not used ocular hypotensivemedication at any time. Group 3-­Another 35 patients with ocular hyper­tension (classified by the same criteria asGroup 2) had received topical ocular hy­potensive medication. Table 2 shows thedrugs used and the duration of therapy.The mean intraocular pressure record fora subject who had received ocular hypo­tensive therapy included both treatedand untreated measurements in order toreflect the intraocular pressures to whichthe cornea had been exposed over a peri­od of time. Group 4-We classified 54patients as having primary open-angleglaucoma on the basis of mean untreatedintraocular pressures of 21 mm Hg orhigher, open angles on gonioscopy, andtypical glaucomatous optic disk cuppingand visual field loss. Table 2 showsthe topical ocular hypotensive agentsused and the duration of therapy. Themean intraocular pressure for a givenpatient included both treated and un­treated measurements during the follow­up period.

We also studied 23 patients scheduled

for glaucoma surgery prospectively. Sev­enteen patients (18 eyes) had uncon­trolled open-angle glaucoma (14 primaryopen-angle glaucoma, two pigmentaryglaucoma, and one posttraumatic anglerecession) necessitating filtering surgery.Excisional trabeculectomies were per­formed anterior to the scleral spur. Sixpatients (eight eyes) scheduled for pe­ripheral iridectomy were also studiedprospectively. Seven eyes underwentsurgical iridectomy for chronic angle­closure glaucoma and one eye had a pro­phylactic iridectomy.

We excluded all eyes with a history orfindings of previous corneal disease, ocu­lar inflammation, ocular trauma, or ocularsurgery. (The exclusion criteria of previ­ous ocular surgery or ocular trauma didnot apply to eyes undergoing trabeculec­tomy.) Eyes undergoing iridectomy wereexcluded from the study if there wasevidence of previous attacks of acuteangle-closure glaucoma, including glau­comflecken, sector iris atrophy, periph­eral anterior synechiae, and posteriorsynechiae.

Central corneal thickness was mea­sured in a masked fashion with a pachy­meter mounted on a slit lamp as de­scribed by Mishima and Hedbys." Threemeasurements of central corneal thick­ness were obtained, recorded to the near­est 0.01 rnm, and averaged.

Specular microscopy was performed bythe method of Waltman and associates."Five photographs were taken of the cen­tral corneal endothelium before any pro­cedure that might affect corneal clarity.The photographs were analyzed in amasked fashion with a rho-theta plottingdevice interfaced to a minicomputer viaan analog-to-digital converter. Twenty to50 cells were traced on each photograph.The total area of the cells traced wascalculated to yield a mean cell area whichwas converted to a mean endothelial celldensity per square millimeter.

32.8 ± 27.614.9 ± 23.418.6 ± 32.7

2.0 ± 0.042.1 ± 36.410.8 ± 9.410.4 ± 8.919.5 ± 19.014.2 ± 15.5

Duration ofTherapy (mos)(Mean ± S.D.)

No. ofMedication Patients*

Pilocarpine 62Carbachol 22Echothiophate iodide 14Demecarium bromide 2Epinephrine 73Dipivefrin 16Timolol 59Acetazolamide 29Methazolamide 25

VOL. 94, NO. 5 CORNEAL ENDOTHELIUM 613

Central corneal endothelial cell densityand central corneal thickness were mea­sured in all patients. Additionally, pa­tients undergoing glaucoma surgery hadpachymetry and specular microscopyperformed before surgery and again 12weeks postoperatively.

We conducted a test-retest reliabilitystudy to evaluate how much of the varia­bility in central corneal endothelial celldensity between individuals was the re­sult of measurement variability. Each eyeof 22 normal volunteers was photo­graphed in two sessions. The technicianperformed the cell counts in a maskedfashion. Test-retest reliability was esti­mated with an intraclass correlation coef­ficient in which variability resulting fromphotography, technician, and the interac­tion of these sources with the subjectwere included in the error term. Theintraclass correlation was .93, suggestingthat the variability observed between in­dividuals was not affected substantially bymeasurement variability. The mean per­centage difference between the centralcorneal endothelial cell densities record­ed at the first and the second sessions forthe same eye was 5.6% ± 5.4% (range,0.2% to 19.7%). The difference betweenthe two measurements was more than10% in seven of 44 eyes (15%).

RESULTS

Neither mean central corneal endothe­lial cell density nor central corneal thick­ness differed significantly among the fourgroups (Table 1). We evaluated the effectsof age, sex, race, and intraocular pressurein each group and in the combined sam­ple with a linear regression model. Theinfluence of each factor was consideredalone as well as in combination with theother factors. Increasing age was associat­ed with a decrease in central cornealendothelial cell density (r = -.31; P =.0001) but was not associated with a

change in central corneal thickness (P =.22). Sex, race, and intraocular pressures(both mean values and those obtained onthe day of pachymetry and specular mi­croscopy) were not related significantly tocentral corneal endothelial cell density(P > .12 in all cases) or to central cornealthickness (P > .28 in all cases).

To further assess the effect of intraocu­lar pressure, we compared Group 1 toGroup 2. There was no substantial differ­ence between the two groups with re­spect to central corneal endothelial celldensity (P = .10) or central corneal thick­ness (P = .69) after controlling for theeffects of age.

Group 1 and Group 4 did not differwith respect to central corneal endotheli­al cell density (P = .53) or central cornealthickness (P = .19) after controlling forthe effects of age.

We assessed the possible effect of topi­cal ocular hypotensive medication in twoways. (1) We compared Group 2 to Group3. After controlling for the effects of age,there was no significant difference be­tween the two groups with respect tocentral corneal endothelial cell density(P = .38) or central corneal thickness(P = .07). (2) The effect of the duration oftopical ocular hypotensive medicationwas evaluated in Groups 3 and 4. Dura­tion of antiglaucoma therapy was not re­lated to central corneal endothelial celldensity (P = .33) or central corneal thick­ness (P = .18). Because most patients hadreceived several medications, it was notpossible to determine the effect of anindividual drug or an individual preserva­tive.

Neither trabeculectomy nor peripheraliridectomy produced significant changesin central corneal endothelial cell density(P > .29 in both cases) or central cornealthickness (P > .40 in both cases) whenpreoperative measurements were com­pared to those made 12 weeks postopera­tively (Table 3). Three eyes had shallow

614 AMERICAN JOURNAL OF OPHTHALMOLOGY NOVEMBER, 1982

TABLE 3

EFFECT OF SURGERY

Clinical Trabecu-Findings lectomy Iridectomy

No. of patients 17 6No. of eyes 18 8Age (yrs)

Mean ± S.D. 61 ± 12.1 66.6 ± 8.4Range 38 to 81 57 to 76

SexMale 13 (76.5%) 1 (16.7%)Female 4 (23.5%) 5 (83.3%)

RaceWhite 10 (58.8%) 5 (83.3%)Black 7 (41.2%) 1 (16.7%)

Central cornealendothelialcell density(cells/mm')Preoperative

Mean ± S.D. 2,004 ± 465.8 2,066 ± 178.5Range 826 to 2,613 1,722 ± 2,350

PostoperativeMean ± S.D. 1,941 ± 522.4 2,127 ± 216.0Range 833 to 2,626 1,782 to 2,489

Central cornealthickness (mm)Preoperative

Mean ± S.D. 0.53 ± 0.03 0.53 ± 0.02Range 0.47 to 0.59 0.49 to 0.56

PostoperativeMean ± S.D. 0.52 ± 0.03 0.51 ± 0.02Range 0.47 to 0.56 0.49 to 0.53

anterior chambers after trabeculectomyand two developed substantial decreasesin central corneal endothelial cell densi­ty. One eye with a flat anterior chamberthat required drainage of suprachoroidalfluid and re-formation of the chamber onthe fifth postoperative day had a decreaseof 497 cells/mm'', Another eye with ashallow anterior chamber from a leakingconjunctival wound that was resutured onthe third postoperative day had a de­crease of 597 cells/mm'',

DISCUSSION

Most of the eyes with ocular hyperten­sion or primary open-angle glaucoma hadintraocular pressures between 20 and26 mm Hg (range, 15 to 32 mm Hg) and

normal central corneal endothelial celldensities. Bigar" noted normal centralcorneal endothelial cell densities in 12patients with open-angle glaucoma.Thus, modest increases in intraocularpressure, even for prolonged periods,seem to have little effect on the cornealendothelium. (This study provided nodata on the effect of severely increasedintraocular pressure.) Decreased centralcorneal endothelial cell densities havebeen reported in eyes with acute angle­closure glaucorna-v" and glaucomato­cyclitic cnsts." Both of these diseases arecharacterized by sudden and marked in­creases in intraocular pressure and areassociated with anterior segment inflam­mation. Animal studies indicate that in­creases in intraocular pressure to 33 to70 mm Hg are necessary to produce mor­phologic evidence of endothelial dam­age. 5,6

Vannas, Setala, and Ruusuvaara' re­ported a small decrease in central cornealendothelial cell density in the affectedeyes of patients with unilateral glaucomacapsulare. Although they contended thatthe decrease was the result of increasedintraocular pressure, they did not reportthe intraocular pressure differential be­tween the affected and the unaffectedeyes. Furthermore, the difference in den­sity between the two eyes was not relatedto the intraocular pressure or the dura­tion of the disease. Only eight of 27patients (29.6%) had a difference in cen­tral corneal endothelial cell density be­tween the two eyes of more than 5%. Thedecrease in density reported by Vannas,Setala, and Buusuvaara' seems to havelittle clinical significance and supportsour contention that modest increases inintraocular pressure have little effect.

Some authorities consider a history ofglaucoma to be a contraindication to theuse of donor corneas for penetrating kera­toplasty. This study suggests that mostcorneas from patients with ocular hyper-

VOL. 94, NO. 5 CORNEAL ENDOTHELIUM 615

tension or primary open-angle glaucomahave enough endothelial cells to serve asdonor material. Similarly, there is con­siderable debate about whether intraocu­lar lenses should be implanted in eyeswith increased intraocular pressures. Al­though this is a complex decision thatmust be made on an individual basis foreach patient, this study indicated goodendothelial reserve in most eyes withocular hypertension or primary open­angle glaucoma.

Topical or intracameral administrationof ocular hypotensive medications andtheir preservatives can damage cornealendothelium in human and animaleyes.8-10 Waltman and associates" found a6.2% decrease in central corneal endo­thelial cell density in a small group ofeyes with ocular hypertension treatedwith topical epinephrine hydrochloridecompared to untreated fellow eyes. Ourstudy found no association between ocu­lar hypotensive medication and centralcorneal endothelial cell density. Becausemost patients had received several medi­cations, it was not possible to determinethe effect of individual drugs or preserva­tives and, thus, a small effect of a givendrug may not have been detected. It isunlikely that a clinically important effectwas missed.

In the present study, uncomplicatedtrabeculectomy and iridectomy had littleeffect on central corneal endothelial celldensity. It is possible that greater lossesof endothelium occur at the site of sur­gery, as has been reported with cataractextraction" and intraocular lens implanta­tion." However, since the incision inglaucoma surgery is relatively small, it isunlikely that such a discrepancy would beof importance. A flat anterior chamberfollowing trabeculectomy can cause amarked decrease in central corneal endo­thelial cell density. In one patient, thisoccurred even though surgical re-for­mation of the anterior chamber was not

required. Thus, it seems likely that theflat chamber itself, rather than the sur­gical re-formation, was the cause of theendothelial cell loss. The decrease incentral corneal endothelial cell densitysuggested that an anterior chamber thatis flat following filtering surgery shouldbe re-formed early. The endothelial cellloss may be the result of mechanicaleffects of the iris or lens on the endotheli­um, inflammation, altered aqueous hu­mor metabolism, hypotony, or some com­bination of these factors.

In our small series, uncomplicated pe­ripheral iridectomy produced no signifi­cant effect on central corneal endothelialcell density. Other studies have reporteddecreases of 4.8%1 and 1.3%2 as a resultof iridectomy. Acute attacks of angle­closure glaucoma have been reported tocause decreases of 9.7%,1 23.1%,2 and39%.15 The greater loss of endothelialcells after acute angle-closure glaucomasupports the practice of prophylactic iri­dectomy in eyes at substantial risk ofdeveloping this disease (that is, felloweyes of eyes that developed acute angle­closure glaucoma).

The reports on intraocular pressureand central corneal thickness are contra­dictory. Increased intraocular pressurehas been said to have no effect on centralcorneal thickness, 18 to increase it, 19 and todecrease it. 2,20-2'2 Many of these studies aredifficult to interpret because they wereperformed on eyes with secondary f&:-msof glaucoma,10,18,19 inflammation,IO,18,19,2'2 orextreme levels of intraocular pres­sure,2,10,18.20,21 or were performed on eyesrecovering from surgery.P' These find­ings suggest that increased intraocu­lar pressure thins the cornea if the endo­thelium is functioning we1l2,20,21 but thatthis effect is of small magnitude. Ehlers"reported that decreasing intraocularpressure by 24 mm Hg with medicationincreases central corneal thickness byless than 0.02 mm. Olson and Kaufman'"

616 AMERICAN JOURNAL OF OPHTHALMOLOGY NOVEMBER, 1982

noted that intraocular pressure increasedby 18 mm Hg after combined cataractextraction-keratoplasty decreased centralcorneal thickness by 0.034 mm. In ourstudy, the difference in intraocular pres­sure among the groups was too small toproduce detectable changes in centralcorneal thickness. Once again, however,this suggested that the corneal endotheli­um functions well in most eyes with ocu­lar hypertension or primary open-angleglaucoma.

REFERENCES

1. Setala, K.: Corneal endothelial cell densityafter an attack of acute glaucoma. Acta OphthalmoL57:1004, 1979.

2. Olsen, T.: The endothelial cell damage in acuteglaucoma. On the corneal thickness response tointraocular pressure. Acta Ophthalmol. 58:257, 1980.

3. Setala, K., and Vannas, A.: Endothelial cells inthe glaucomato-cyclitic crisis. Adv. OphthaimoL36:218, 1978.

4. Vannas, A., Setala, A., and Ruusuvaara, P.:Endothelial cells in capsular glaucoma. Acta Oph­thalmol. 55:951, 1977.

5. Melamed, S., Ben-Sira, I., and Ben-Shaul, Y.:Corneal endothelial changes under induced intraocu­lar pressure elevations. A scanning and transmissionelectron microscopic study in rabbits. Br. J. Ophthal­mol. 64:164, 1980.

6. Svedbergh, B.: Effects of artificial intraocularpressure elevation on the corneal endothelium in thevervet monkey iCeriopithecus ethiops). Acta Oph­thalmol. 53:839, 1975.

7. Ytteborg, J., and Dohlman, C.: Corneal edemaand intraocular pressure. 1. Animal experiments.Arch. Ophthalmol. 74:375, 1965.

8. Jay, J. L., and MacDonald, M.: Effects ofintra­ocular miotics on cultured bovine corneal endotheli­um. Br. J. Ophthalmol. 62:815, 1978.

9. Waltman, S. R., Yarian, D., Hart, W., [r., andBecker, B.: Corneal endothelial changes with long­term topical epinephrine therapy. Arch. Ophthal­mol. 95:1357, 1977.

10. Gasset, A. R., Ishii, Y., Kaufman, H. E., andMiller, T.: Cytotoxicity of ophthalmic preservatives.Am. J. OphthalmoL 78:98, 1974.

ll. Maurice, D. M.: Cellular membrane activityin the corneal endothelium of the intact eye. Expert­entia 24:1094, 1975.

12. Laing, R. A., Sandstrom, M. M., and Lei­bowitz, H. M.: In vivo photomicrography of thecorneal endothelium. Arch. Ophthalmol. 93:143,1975.

13. Bourne, W. M., and Kaufman, H. E.: Specu­lar microscopy of human corneal endothelium invivo. Am. J. Ophthalmol, 81:319, 1976.

14. Mishima, S., and Hedbys, B. 0.: Measure­ment of corneal thickness with the Haag-Streit pa­chometer. Arch. OphthalmoL 80:710, 1968.

15. Bigar, F.: Specular microscopy of the cornealendothelium. Optical solutions and clinical results.Rev. Ophthalmol. 6:1, 1982.

16. Hoffer, K. J.: Vertical endothelial cell dispari­ty. Am. J. Ophthalmol. 87:344, 1979.

17. Blackwell, W. L., Gravenstein, N., and Kauf­man, H. E.: Comparison of central corneal endothe­lial cell numbers with peripheral areas. Am. J.Ophthalmol, 84:473, 1977.

18. Ytteborg, J., and Dohlman, C. H.: Cornealoedema and intraocular pressure. II. Clinical results.Arch. OphthalmoL 74:477, 1965.

19. DeCevallos, A., Dohlman, C. H., and Rein­hart, W. J.: Corneal thickness in glaucoma. Ann.Ophthalmol. 8:177, 1976.

20. Olson, R. J., and Kaufman, H. E.: Intraocularpressure and corneal thickness after penetrating ker­atoplasty. Am. J. OphthalmoL 86:97, 1978.

21. Ehlers, N.: On corneal thickness and intraocu­lar pressure. II. A clinical study on the thickness ofthe corneal stroma in glaucomatous eyes. Acta Oph­thalmoL 48:1107, 1970.

22. Ehlers, N., and Riise, D.: On corneal thick­ness and intraocular pressure. A clinical study of thethickness of the cornea in eyes with retinal detach­ment. Acta Ophthalmol. 45:809, 1967.