Effects of Glaucoma Medications on Corneal Endothelium, Keratocytes, and Subbasal Nerves Among Participants in the Ocular Hypertension Treatment Study Keith H

JOBNAME: corn 25#9 2006 PAGE: 1 OUTPUT: Thursday November 16 08:14:31 2006 lww/corn/129949/ICO200376

CLINICAL SCIENCE

Effects of Glaucoma Medications on Corneal Endothelium, Keratocytes, and Subbasal Nerves Among Participants in the Ocular Hypertension Treatment Study Keith H. Baratz,* Cherie B. Nau,* Eric J. Winter,* Jay W. McLaren,* David O. Hodge,† David C. Herman,* and William M. Bourne*

Key Words: nerves, keratocytes, endothelium, glaucoma, medica- Purpose: To compare subbasal corneal nerve and keratocyte density tion, toxicity and endothelial characteristics of ocular hypertensive patients treated with medications or observation. (Cornea 2006;25:1046–1052) Methods: Participants in the Ocular Hypertensive Treatment Study (OHTS) randomized at Mayo Clinic to medication or observation overnmental drug approval processes and extensive were evaluated with specular microscopy annually for 6 years. worldwide clinical experience among millions of patients Confocal microscopy was performed 78 to 108 months after G with glaucoma have established the clinical use of various enrollment. Subbasal nerve density was calculated by manual tracing intraocular pressure–lowering medications. Clinicians have and digital image analysis. Keratocyte density was determined by been concerned about the effects of some of these topical manual counting methods. Data were compared using a t test and medications and their preservatives on ocular surface function a rank sum test. and integrity, and some have associated the use of these drugs Results: After 6 years, corneal endothelial cell density, percent with local allergic reactions, chronic inflammation of the 1,2 3–5 hexagonal cells, and coefficient of variation of cell area for the conjunctiva, tear film abnormalities, or corneal epitheli- 6,7 observation (n = 21) and medication groups (n = 26) were similar opathy. Symptoms from these adverse effects may be 8 (2415 6 300 vs. 2331 6 239 cells/mm2; 63% 6 11% vs. 65% 6 relatively common. Toxicity to deeper corneal layers is not 9 10%; and 0.32 6 0.07 vs. 0.30 6 0.06, respectively). Of 38 normally considered to be a risk, but several reports, in- 10 11,12 participants undergoing confocal examination, the medication group cluding animal and human studies, suggest concern, at (n = 19) had fewer nerves (3.8 6 2.1 vs. 5.9 6 2.0 nerves/frame; P = least in specific clinical settings such as preexisting corneal 0.02) and a lower nerve density (5643 6 2861 vs. 9314 6 3743 abnormalities. mm/mm2; P = 0.007) than the observation patients (n = 10). An The Ocular Hypertension Treatment Study (OHTS) additional 9 patients in the observation group, who began medication provides an ideal opportunity to study the effects of chronic before confocal scanning, had intermediate nerve densities. Full- topical glaucoma medication on the human cornea. The OHTS thickness keratocyte density was similar, with 22,257 6 2419 and is a prospective, randomized, multicenter study to assess the 23,430 6 3285 cell/mm3 in the observation and medication groups, risk of progression to primary open-angle glaucoma in ocular respectively. hypertensive participants treated with either medications or observation.13 In the current ancillary study to the OHTS, we Conclusions: Chronic administration of glaucoma medications examined corneas of patients with specular microscopy an- causes a decrease in the number and density of corneal subbasal nerve nually and with confocal microscopy once, 6 to 9 years after fiber bundles but does not affect keratocyte density or corneal enrollment. Densities of keratocytes, subbasal nerves, and en- endothelial characteristics. dothelium were compared between patients who were treated with ocular hypotensive medications and those who were not treated. Received for publication January 3, 2006; accepted April 25, 2006. From the *Department of Ophthalmology, Mayo Clinic College of Medicine, Rochester, MN; and the †Department of Health Sciences Research, Mayo MATERIALS AND METHODS Clinic College of Medicine, Rochester, MN. Supported in part by NIH Grants EY02037 and EY10406, an unrestricted Patients grant from Research to Prevent Blindness, New York, NY, and the Mayo Foundation, Rochester, MN. The protocols of both the OHTS and the ancillary Presented in part at the World Cornea Congress, April 15, 2005, Washington, DC. specular microscopy and confocal microscopy studies were The authors state that they have no proprietary interest in the products named approved by the Mayo Clinic Institutional Review Board and in this article. the OHTS Data Safety and Monitoring Committee. Study Reprints: Keith H. Baratz, MD, Mayo Clinic, Department of Ophthalmology, 200 First Street SW, Rochester, MN 55905 (e-mail: baratz.keith@mayo. participants were recruited for the OHTS at Mayo Clinic edu). between April 1994 and January 1996 and randomized to Copyright Ó 2006 by Lippincott Williams & Wilkins either medical treatment or observation. Patients were eligible

1046 Cornea Volume 25, Number 9, October 2006

Cornea Volume 25, Number 9, October 2006 Effects of Glaucoma Medications on the Cornea

if they had normal-appearing optic discs, normal visual ﬁelds, and intraocular pressures between 24 and 32 mm Hg in 1 eye and between 21 and 32 mm Hg in the contralateral eye.13 Fifty- ﬁve patients at Mayo Clinic gave written informed consent to participate in the OHTS and this ancillary study. These studies conformed to the tenets of the Declaration of Helsinki. Participants in the medication group were treated bilaterally with topical medications commercially available in the United States to achieve a target IOP of 24 mm Hg or less and a minimum 20% reduction in IOP from the average of the qualifying IOP and IOP at the baseline randomization visit. The minimum target pressure was 18 mm Hg. Topical medication was changed or new medications were added at the discretion of the OHTS investigator (D.C.H.) until both of these goals were met or the participant was receiving maximum- tolerated topical medical therapy.

Endothelial Morphometry FIGURE 1. One frame from a confocal microscopy scan Images of the central corneal endothelial were recorded showing subbasal nerve fibers and a nerve branch (arrow). by contact specular microscopy (SP-580; Keeler Instruments, Broomall, PA) at the time of enrollment and annually for 6 years. The apices of 100 contiguous endothelial cells were was masked to the name and randomization status of the digitized manually (Bambi 6.1; Bio-Optics, Arlington, MA) patients. Nerve fiber density measurements included data from by 1 investigator (E.J.W.) and endothelial cell density, coef- all analyzable scans of all eyes. ficient of variation, and percent hexagonal cells were cal- 14 culated. These variables were compared between groups and Keratocyte Density to the same variables from a control group, consisting of age- Keratocyte density was determined using a manual matched normal subjects without a history of anterior segment method of counting images of cell nuclei in a predefined area eye disease or surgery, diabetes, or contact lens wear, using the of confocal images.18 Measurements were obtained from 1 same methods previously. representative scan from each eye. Two nonblurred frames were selected from each of 5 layers of the stroma: the anterior Confocal Microscopy 10%, 11% to 33%, 34% to 66%, 67% to 90%, and posterior Corneas of each patient were examined with a confocal 10% of stromal depth, based on the frames that contained the microscope (Tandem Scanning Confocal Microscope, Reston, anteriormost keratocytes and the endothelium. One observer VA) as described by Patel et al.15 Each cornea was scanned 3 (C.B.N.) identified and marked bright objects that resembled to 6 times through its entire depth. The video camera of the keratocyte nuclei inside a predefined rectangular field drawn confocal microscope was set in an automatic-gain mode to within each frame. Objects that touched the lower or left optimize image brightness when assessing cell density and in boundaries of the field were included, but those that touched a manual-gain mode to estimate brightness of backscattered the upper or right boundaries were excluded. Cell density, light. At least 1 scan per cornea was recorded in each mode. expressed as cells per cubic millimeter, was the number of objects in the rectangular area divided by the product of the 2 Subbasal Nerve Density area of the field (0.109 mm ) and the optical depth of field Subbasal nerve fiber bundles form a plexus just anterior (11.9 mm). The mean cell density of each stroma was the mean to the Bowman membrane16 and are typically visible through density of the 10 frames weighted by the distance between several frames of a confocal examination (Fig. 1). In all frames frames. The observer was masked to the name of the patient, that contained subbasal nerves, 1 investigator (K.H.B.) using the randomization status, and the depth of the frame in the a custom program17 traced the paths of all main nerves and stroma. Keratocyte density was determined from 1 represen- their branches greater than 50 mm through multiple frames if tative scan of each eye. necessary. The program calculated the total number and length of main nerves and their branches as well as the mean angles of Backscattered Light From the Stroma main nerves in this layer. Nerve number was the number of Because increased backscattered light can increase nerves and branches in a frame of a confocal scan, expressed image intensity and decrease the ability to discern nerves as nerves per frame. Nerve density was expressed as the total and cells in a confocal scan, we attempted to quantify this length of all main nerves and their branches divided by the area variable. The brightness of backscattered light from the stroma of stroma represented by the image (0.191 mm2 for all scans), was determined from the average intensity of each frame in expressed as micrometers per square millimeter. The nerve scans recorded with the video camera in manual-gain mode. number and density and mean nerve angle for each cornea was Average intensities were calculated by using a custom program the mean of the values of the individual scans. The observer that calculated the mean brightness of the central 0.109 mm2 of

q 2006 Lippincott Williams & Wilkins 1047

Baratz et al Cornea Volume 25, Number 9, October 2006 each frame. Mean brightnesses were adjusted for variations in 66 months (range, 8–218 months), and 4 participants in the sensitivity of the confocal microscope (brightness of the light observation group used these drops for a mean of 57 months source and sensitivity of the camera) using the mean bright- (range, 34–74 months). No other participants had used ness of images of a fluorescent glass standard scanned imme- intraocular pressure–lowering medications before enrollment diately before scanning each patient.19 in the OHTS, and no participants had used glaucoma medications other than b-adrenergic antagonists. Medications used after randomization included apraclonidine 0.5%, betaxolol Statistical Analysis 0.25%, brimonidine 0.15% and 0.2%, brinzolamide 1%, Mean nerve number, density, angle, and keratocyte dorzolamide 2%, latanoprost 0.005%, pilocarpine 0.5% and density were calculated for each eye by averaging the values 1%, timolol 0.5%, dorzolamide 2%/timolol 0.5% com- obtained from individual scans. These values and endothelial bination, and unoprostone 0.15%. Both eyes of each partici- morphometric variables for both eyes were averaged for each pant were treated with the same medications generally, but this patient. Descriptive statistics were calculated for each variable was not universal. and each randomization group. Variables were compared between the medication and observation groups by using t tests or a rank sum test, depending on whether data were normally Endothelial Morphometry distributed. Endothelial variables were compared at baseline In addition to the 1 observation participant who moved and at 72 months. Statistical tests were adjusted for 2 com- to another study center, 6-year endothelial data were not parisons by using the Bonferroni technique. The rate of available for 2 patients, 1 in each group, who did not reconsent endothelial cell loss for each participant expressed as a percent k 3 1 year after 5 years. Data were not recorded at 6 years for 1 per year was assumed to be equal to 100 3 e , where k is observation group participant and at 2 and 3 years for 1 the slope of the line fitted by linear regression to the natural medication group participant. Four other patients in the logs of all mean densities versus time between baseline and observation group began medication during the study because 6 years. The percent loss per year was calculated for each subject of conversion to glaucoma. Subsequent endothelial data from and was compared between groups by using a rank sum test. these patients were excluded from analysis. Thus, the available sample for endothelial morphometry at 6 years was 26 in the medication group and 21 in the observation group. RESULTS At baseline or after 6 years of follow-up, the endothelial cell density, percent hexagonal cells, and coefficient of Patients variation did not differ significantly between the observation Of the 55 participants enrolled in the OHTS at our and medication groups (Table 1; Fig. 2). Also, no significant center, 27 were randomized to the medication group and differences in these parameters occurred at any other time 28 were randomized to the observation group. Follow-up data interval. This sample size provides adequate statistical power were not available for 1 participant in the observation group (a = 0.025, b = 0.20) to detect a difference at 72 months of 250 who moved and transferred care to another study center. There cells/mm2 for endothelial cell density, 0.06 for coefficient of were no significant differences in the baseline demographics variation, and 10% for percent hexagonal cells. Except for between the randomization groups. Mean age of the obser- coefficient of variation, the groups also did not differ vation group at enrollment was 59 6 9 years compared with significantly from a historical group of 55 age-matched 57 6 11 years for the medication group (P = 0.44). The controls (Table 1). After excluding the 4 observation proportion of male patients was 44% for the observation participants who started medication within the first 6 years group and 56% for the medication group (P = 0.41). Before of follow-up, the annual rate of corneal endothelial cell loss enrollment in the OHTS, 7 participants in the medication was similar at 0.85% in the observation group and 0.68% in group used b-adrenergic antagonist eye drops for a mean of the medication group (P = 0.70).

TABLE 1. Endothelial Morphometric Characteristics at Baseline and at 72-Month Follow-Up Endothelial Cell Density % Hexagonal Coefﬁcient of Variation of N (cells/mm2 6 SD) Cells (6SD) Cell Area (6SD)

Group Baseline 72 mo Baseline 72 mo Baseline 72 mo Baseline 72 mo

Observation 28 21 2516 6 283 2415 6 300 60 6 9636 11 0.32 6 0.07 0.32 6 0.07 Medication 27 26 2465 6 285 2331 6 239 62 6 6656 10 0.32 6 0.06 0.30 6 0.06 P — — 1.0* 0.6* 1.0† 1.0* 1.0* 0.4† Minimum detectable difference — — — 250 — 10 — 0.06 Age-matched controls‡ 55 — 2534 6 306 — 62 6 7 — 0.30 6 0.06 —

Statistical comparison using *t test or †rank sum test adjusted for 2 comparisons with the Bonferroni adjustment. ‡P value comparing control group with the 55 study participants = 0.44 (endothelial cell density; t test), 0.36 (percent hexagonal cells; t test), and 0.05 (coefﬁcient of variation, rank sum test).

1048 q 2006 Lippincott Williams & Wilkins

Cornea Volume 25, Number 9, October 2006 Effects of Glaucoma Medications on the Cornea

scan for nerves, but 1 participant in the treatment group had no analyzable scans for keratocyte density analysis.

Subbasal Nerve Density The number and density of central, subbasal, and corneal nerves were lower in the medication group than in the observation group (P = 0.02 for nerve number; P = 0.007 for nerve density; Table 2). The mean nerve angle was not statistically different between the 2 groups (P = 0.10). The number and density of nerves in the observation-to-medication group were intermediate to the other groups.

Keratocyte Density The keratocyte density averaged over the full thickness of the cornea did not differ signiﬁcantly between the FIGURE 2. Endothelial cell density versus time for the observation and medication groups (Table 2). Likewise, the observation and medication groups. The number of partic- density of cells in the anterior 10% of the stroma, where the ipants at each time gate is represented by the numbers above keratocytes tend to be most dense, was also similar. The full- (observation group) and below (medication group) the SD thickness stromal keratocyte densities in the study groups were bars. similar to the value (20,522 6 2981 cells/mm2) of a normal group of subjects studied using identical methods.18

Confocal Microscopy Backscattered Light From the Stroma Thirty-eight eligible patients, 19 in each randomization The light intensity for confocal images was similar group, consented to confocal microscopy and were evaluated among the groups (Fig. 3). This similarity was consistent at all 92 6 6 months (range, 78–108 months) from OHTS depths of the cornea. enrollment. The remaining 16 patients declined to participate in this ancillary study or had bilateral cataract surgery. In the instance of unilateral cataract surgery, the contralateral eye DISCUSSION remained in the study. Before confocal microscopy, 9 of the 19 The stromal keratocyte and endothelial cell densities of subjects in the observation group were started on medication corneas of study participants on chronic, topical glaucoma (observation-to-medication group), because they progressed to medications are not different from those of control patients. glaucoma during the first 6 years of the study (n = 4) or elected These results indicate that these medications may be used to start therapy when the OHTS observation group was opened without damage to these cells in most situations. In light of the to treatment at the end of 6 years. extensive clinical experience with these medications, our A total of 312 confocal scans were obtained. Fifty-five findings are not surprising. Previous studies of various scans (18%) were deemed not analyzable for nerve analysis, glaucoma drops in humans also have not found a deleterious usually because of patient movement. The proportion of eyes effect on endothelial cell density,9,20–22 corneal deswelling with at least 1 unanalyzable scan was 44% in the medication rates,12,21 or fluorescein permeability.23 group, 40% in the observation group, and 53% in the In contrast, some animal studies have suggested that the observation-to-medication group (P = 0.72). The median ultrastructure of the endothelium changes after chronic use of number of analyzable scans was 4 per eye (range, 0–6) in each glaucoma medications.10 Also, Konowal et al.11 reported of the 3 study groups. All participants had at least 1 analyzable irreversible corneal decompensation after initiating topical

TABLE 2. Subbasal Nerve and Keratocyte Data 72 to 108 Months After OHTS Enrollment Subbasal Nerves Keratocytes

Nerve Nerve Nerve Anterior 10% Full-Thickness Number Density Angle Keratocyte Density Keratocyte Density Group n (Nerves/frame 6 SD) (mm/mm2 6 SD) (Degrees 6 SD) N (cells/mm3 6 SD) (cells/mm3 6 SD)

Observation 10 5.9 6 2.0 9314 6 3743 99 6 12 10 28,688 6 10,781 22,257 6 2,419 Observation to medication 9 4.9 6 1.4 7913 6 2268 104 6 7 9 32,518 6 13,487 21,996 6 2,868 Medication 19 3.8 6 2.1 5643 6 2861 107 6 13 18 31,522 6 9,621 23,430 6 3,285 P* 0.02 0.007 0.10 0.48 0.33

*t test comparing observation and medication groups only.

q 2006 Lippincott Williams & Wilkins 1049

Baratz et al Cornea Volume 25, Number 9, October 2006

in animal models and in vitro.37,44 Many of these adverse effects have been linked to, but not exclusively attributed to, the preservative benzalkonium chloride.1,8,33,36,37,39,40 Second, are the decreased numbers of subbasal nerves associated with other anatomic changes? Investigators studying the neurotrophic cornea have described decreased epithelial cell mitotic activity,45 thinning of the epithelium, and a decrease in the number of epithelial cell layers.46 Data from the same Mayo OHTS cohort showed a decrease in mean corneal thickness in the medication group, partially attributed to a thinner epithelium.47 Epithelial thinning was probably the result of the medication rather than an abnormality in the observation group, whose epithelial thickness was not different from that of normal subjects18 examined with identical techniques. We propose as a hypothesis that the epithelial FIGURE 3. Mean reflected light intensity of confocal micros- thinning results from the decreased nerve density. Such an copy scan versus corneal depth. effect, in conjunction with the epitheliopathy described by others, would be consistent with a subclinical neurotrophic keratitis in a proportion of these chronically treated ocular dorzolamide therapy in 9 human eyes with preexisting hypertensive patients. endothelial compromise. Other findings attributed to dorzo- Third, how extensive must the loss of nerves be to cause lamide include an increase in endothelial fluorescein perme- the clinical neurotrophic keratitis seen in the few reports of ability12 and the potential for a delayed deswelling rate in eyes subjected to b-adrenergic antagonist therapy? Large a subset of patients.21 Our study found no adverse trend in the limbal incisions that interrupt up to one half of the limbal density or morphology of endothelial cells in patients on nerve branches during intracapsular cataract extraction cause glaucoma medication. The rate of endothelial cell loss in the few clinical problems. Similarly, patients with previously 2 randomization groups was similar, and the 0.68% decrease healthy ocular surfaces who undergo penetrating keratoplasty per year among medicated patients was similar to the 0.6% can recreate a relatively healthy epithelial and tear film layer found during 10 years in normal corneas.24 despite profound sensory loss. It should be noted, however, In contrast to the stability of keratocytes and endothe- that keratoplasty does not affect limbal and stem cell tissue. lium, we found a significant decrease in the density of subbasal Finally, what caused nerve loss in patients on medication? nerve fiber bundles visible with confocal microscopy in the Previous studies described b-adrenergic antagonists as a cause treated group. This effect on the subbasal nerves is somewhat of decreased corneal sensation.28–31 Our sample size is too unexpected. Several questions must be answered before we small to perform any subgroup analysis, which prevents us understand the implications of our findings. from implicating specific medications or preservatives. First, are functional changes linked to the decrease in Despite our prospective, randomized design, this an- nerve fiber bundles? Neurotrophic keratitis, which may result cillary study has limitations. In addition to the study sample from nerve loss in diseases such as herpes zoster ophthalmi- confined to 1 institution, the observation group evaluated with cus, is associated with decreased corneal sensation, punctate confocal microscopy is small, because of conversion of par- keratopathy, tear film abnormalities, and, in the most severe ticipants to medication, bilateral cataract surgery, and failure to cases, persistent epithelial defects and ulceration.25–27 To our consent to the ancillary study. Also, confocal microscopy at knowledge, no large, rigorous studies have established the beginning of the study, rather than only at the end, could a consistent link between clinical neurotrophic disease and have identified potential baseline differences between the glaucoma therapy. However, several cases of corneal in- groups of participants. Most importantly, the simultaneous sensitivity, sometimes in conjunction with profound ocular and sequential use of medications in the study group presents surface abnormalities, were documented in patients who used 1 uncontrollable variable. topical b-adrenergic antagonists.28–31 Commercial prepara- Keratocyte densities and variability in the data derived tions of prostaglandin analogs may also transiently decrease from confocal images with the current technique have been corneal sensation.32 Other case series and case reports have validated in our laboratory by comparison with densities implicated b-adrenergic antagonists, prostaglandin analogs, determined by histologic methods.18 We have not similarly and preservatives as causing symptomatic ocular surface validated our nerve fiber bundle densities, nor have we changes such as punctate corneal epitheliopathy,2,5,7,8,33 medi- evaluated the variability of an observer in estimating nerve cally resistant herpetic keratitis,6,34 and ocular discomfort.8 densities from confocal images frames that are counted only Other researchers have reported less symptomatic findings once in this study. We do attempt to compensate for this such as tear film instability,3,4,35 disruption of the epithelial limitation by evaluating multiple confocal scans of each eye. barrier function,33,36 and chronic inflammatory infiltration37,38 or However, in our experience, the quality of the confocal scan, expression of inflammatory markers39 in the conjunctiva. These which may be degraded by patient movement or background effects, as well as impaired epithelial wound healing40–42 and image intensity, can affect the ability to identify subbasal ultrastructural surface abnormalities,1,42,43 have been confirmed nerves and produce falsely low estimates of nerve density.

1050 q 2006 Lippincott Williams & Wilkins

Cornea Volume 25, Number 9, October 2006 Effects of Glaucoma Medications on the Cornea

Eighteen percent of the scans were graded as unreadable 12. Egan CA, Hodge DO, McLaren JW, et al. Effect of dorzolamide on because of patient movement and 2% were deemed to be high- corneal endothelial function in normal human eyes. Invest Ophthalmol Vis Sci. 1998;39:23–29. quality scans with no visible nerves. However, the proportion 13. Gordon MO, Kass MA, for the Ocular Hypertension Study Group. The of patients with at least 1 confocal scan that was not analyzable Ocular Hypertension Treatment Study. Design and baseline description of was similar among the groups (P = 0.72). Backscattered light the participants. Arch Ophthalmol. 1999;117:573–583. and image intensity also appeared similar among the study 14. Ohno K, Nelson LR, McLaren JW,et al. Comparison of recording systems groups, so a difference in image quality cannot explain the and analysis methods in specular microscopy. Cornea. 1999;18:416–423. 15. Patel SV, McLaren JW, Camp JJ, et al. Automated quantification of difference measured in subbasal nerves. keratocyte density by using confocal microscopy in vivo. Invest Regional variation in subbasal nerve density does Ophthalmol Vis Sci. 1999;40:32–36. occur.48 The orientation of the nerve fiber bundles may also 16. Mu¨ller LJ, Marfurt CF, Kruse F, et al. Corneal nerves: structure, contents be altered as a result of surgery or disease.49 Our measure- and function. Exp Eye Res. 2003;76:521–542. 17. Calvillo MP, McLaren JW, Hodge DO, et al. Corneal reinnervation after ments, which are generally within 1 mm of the center of the LASIK: Prospective 3-year longitudinal study. Invest Ophthalmol Vis Sci. cornea, may not be representative of the entire cornea, and 2004;45:3991–3996. alterations in the distribution or orientation of nerve fibers 18. Patel SV, McLaren JW, Hodge DO, et al. Normal human keratocyte could affect our nerve density data. density and corneal thickness measurement by using confocal microscopy Further clarification of the relationship between glau- in vivo. Invest Ophthalmol Vis Sci. 2001;42:333–339. 19. Patel SV, McLaren JW, Hodge DO, et al. Confocal microscopy in vivo in coma medication and decreased corneal nerve density should corneas of long-term contact lens wearers. Invest Ophthalmol Vis Sci. focus on identifying the specific causative agents and the 2002;43:995–1003. interactions between nerve density, epithelial thickness, and 20. Lass JH, Eriksson GL, Osterling L, et al. Comparison of the corneal corneal sensation. This endeavor would require larger cohorts effects of latanoprost, fixed combination latanoprost-timolol, and timolol. A double-masked, randomized, one-year study. Ophthalmology. 2001; of patients and the use of devices that can reliably detect small 108:264–271. differences in corneal sensitivity and structure. Histologic 21. Giasson CJ, Nguyen TQ, Boisjoly HM, et al. Dorzolamide and corneal confirmation of nerve density measurements derived from recovery from edema in patients with glaucoma or ocular hypertension. confocal microscopy is also appropriate. Am J Ophthalmol. 2000;129:144–150. 22. Kaminski S, Hommer A, Koyuncu D, et al. Influence of dorzolamide on corneal thickness, endothelial cell count and corneal sensibility. Acta Ophthalmol Scand. 1998;76:78–79. ACKNOWLEDGMENTS 23. Beneyto P, Perez TM. Effect of continued treatment with timolol maleate The authors thank Drs. Mae O. Gordon and Michael on corneal endothelium: a fluorophotometric study. Cornea. 1998;17: A. Kass for review of the manuscript. 600–603. 24. Bourne WM, Nelson LR, Hodge DO. Central corneal endothelial cell changes over a ten-year period. Invest Ophthalmol Vis Sci. 1997;38:779– REFERENCES 782. 1. Noecker RJ, Herrygers LA, Anwaruddin R. Corneal and conjunctival 25. Davies MS. Corneal anaesthesia after alcohol injection of the trigeminal changes caused by commonly used glaucoma medications. Cornea. 2004; sensory root. Br J Ophthalmol. 1970;54:577–586. 23:490–496. 26. de Haas EBH. Dessication of cornea and conjunctiva after sensory 2. Yalvac IS, Gedikoglu G, Karagoz Y, et al. Effects of antiglaucoma drugs denervation. Significance of desiccation for pathogenesis of neuro- on ocular surface. Acta Ophthalmol Scand. 1995;73:246–248. paralytic keratitis. Arch Ophthalmol. 1962;67:79–92. 3. Shimazaki J, Hanada K, Yagi Y, et al. Changes in ocular surface caused by 27. Pannabecker CL. Keratitits neuroparalytica. Corneal lesions following antiglaucomatous eyedrops: prospective, randomised study for the operations for trigemimal neuralgia. Arch Ophthalmol. 1944;32:456–463. comparison of 0.5% timolol v. 0.12% unoprostone. Br J Ophthalmol. 28. Van Buskirk EM. Corneal anesthesia after timolol maleate therapy. Am J 2000;84:1250–1254. Ophthalmol. 1979;88:739–743. 4. Thygesen J, Aaen K, Theodorsen F, et al. Short-term effect of latanoprost 29. Calissendorff B. Corneal anesthesia after topical treatment with timolol and timolol eye drops on tear fluid and the ocular surface in patients with maleate. A case report. Acta Ophthalmol (Copenh). 1981;59:347–349. primary open-angle glaucoma and ocular hypertension. Acta Ophthalmol 30. Vogel R, Clineschmidt CM, Hoeh H, et al. The effect of timolol, betaxolol Scand. 2000;78:37–44. and placebo on corneal sensitivity in healthy volunteers. J Ocul 5. Herreras JM, Pastor JC, Asensio VM. Ocular surface alteration after long- Pharmacol. 1990;6:85–90. term treatment with an antiglaucomatous drug. Ophthalmology. 1992;99: 31. Weissman SS, Asbell PA. Effects of topical timolol (0.5%) and betaxolol 1082–1088. (0.5%) on cornea sensitivity. Br J Ophthalmol. 1990;74:409–412. 6. Sudesh S, Cohen EJ, Rapuano CJ, et al. Corneal toxicity associated with 32. Kozobolis VP, Detorakis ET, Maskaleris G, et al. Corneal sensitivity latanoprost. Arch Ophthalmol. 2004;117:539–540. changes following the instillation of latanoprost, bimatoprost, and 7. Inoue K, Okugawa K, Kato S, et al. Ocular factors relevant to anti- travoprost eyedrops. Am J Ophthalmol. 2005;139:742–743. glaucomatous eyedrop-related keratoepitheliopathy. J Glaucoma. 2003; 33. de Jong C, Stolwijk T, Kuppens E, et al. Topical timolol with and without 12:480–485. benzalkonium chloride: epithelial permeability and autofluorescence of 8. Pisella PJ, Pouliquen P, Baudouin C. Prevalence of ocular symptoms and the cornea in glaucoma. Graefes Arch Clin Exp Ophthalmol. 1994;232: signs with preserved and preservative-free glaucoma medication. Br J 221–224. Ophthalmol. 2002;86:418–423. 34. Wand M, Gilbert CM, Liesegang TJ. Latanoprost and herpes simplex 9. Lass JH, Khosrof SA, Laurence JK, et al. A double-masked, randomized, keratitis. Am J Ophthalmol. 1999;127:602–604. 1-year study comparing the corneal effects of dorzolamide, timolol, and 35. Ishibashi T, Yokoi N, Kinoshita S. Comparison of the effects of topical betaxolol. Dorzolamide Corneal Effects Study Group. Arch Ophthalmol. levobunolol and timolol solution on the human ocular surface. Cornea. 1998;116:1003–1010. 2003;22:709–715. 10. Matsuda M, Ishii Y, Awata T, et al. The effect of long-term topical 36. Ishibashi T, Norihiko Y, Kinoshita S. Comparison of the short-term effects administration of commercial beta blockers on the rat corneal on the human corneal surface of topical timolol maleate with and without endothelium. Int Ophthalmol. 1989;13:67–73. benzalkonium chloride. J Glaucoma. 2003;12:486–490. 11. Konowal A, Morrison JC, Brown SV, et al. Irreversible corneal 37. Baudouin C, Pisella PJ, Fillacier K, et al. Ocular surface inflammatory decompensation in patients treated with topical dorzolamide. Am J changes induced by topical antiglaucoma drugs: human and animal Ophthalmol. 1999;127:403–406. studies. Ophthalmology. 1999;106:556–563.

q 2006 Lippincott Williams & Wilkins 1051

Baratz et al Cornea Volume 25, Number 9, October 2006

38. Pisella P-J, Debbasch C, Hamard P, et al. Conjunctival proinflammatory 44. Kaufman HE, Varnell ED, Thompson HW. Latanoprost increases the and proapoptotic effects of latanoprost and preserved and unpreserved severity and recurrence of herpetic keratitis in the rabbit. Am J timolol: an ex vivo and in vitro study. Invest Ophthalmol Vis Sci. 2004;45: Ophthalmol. 1999;127:531–536. 1360–1368. 45. Sigelman S, Friedenwald JS. Mitotic and wound-healing activities of the 39. Baudouin C, Garcher C, Haouat N, et al. Expression of inflammatory corneal epithelium. Effect of sensory denervation. Arch Ophthalmol. membrane markers by conjunctival cells in chronically treated patients 1954;52:46–51. with glaucoma. Ophthalmology. 1994;101:454–460. 46. Alper MG. The anesthetic eye: an investigation of changes in the anterior 40. O’Brien WJ, DeCarlo JD, Stern M, et al. Effects of Timoptic on corneal ocular segment of the monkey caused by interrupting the trigeminal nerve reepithelialization. Arch Ophthalmol. 1982;100:1331–1333. at various levels along its course. Trans Am Ophthalmol Soc. 1975;73: 41. Nork TM, Holly FJ, Hayes J, et al. Timolol inhibits corneal epithelial 323–365. wound healing in rabbits and monkeys. Arch Ophthalmol. 1984;102: 47. Herman DC, Hodge DO, Bourne WM. Changes in corneal thickness in 1224–1228. patients with treated and untreated ocular hypertension. Cornea. 25:639–643. 42. Haruta Y, Ohashi Y, Matsuda S. Corneal epithelial deficiency induced 48. Patel DV, McGhee CN. Mapping of the normal human corneal sub-basal by the use of beta-blocker eye drops. Eur J Ophthalmol. 1997;7:334– nerve plexus by in vivo laser scanning confocal microscopy. Invest 339. Ophthalmol Vis Sci. 2005;46:4485–4488. 43. Segawa K, Nagai T, Tanaka N, et al. Effects of three beta-blocker eye 49. Simo Mannion L, Tromans C, O’Donnell C. An evaluation of corneal drops on the rabbit cornea: an electron microscope study. Clin Ther. 1986; nerve morphology and function in moderate keratoconus. Cont Lens 8:263–268. Anterior Eye. 2005;8:185–192.

1052 q 2006 Lippincott Williams & Wilkins