Conjunctival changes induced by prostaglandin analogues and timolol maleate: a histomorphometric study

1 Post-Graduated in Ophthalmology (Doctorate) from the Universidade de São Paulo USP São Paulo (SP), Brazil; Hospital Universitário Evangélico de Curitiba (PR) Brazil. 2 Post-doctorate degree; Lecturer of the Department of Ophthalmology of the USP São Paulo (SP), Brazil; Department of Ophthalmology of the Universidade de Campinas UNICAMP Campinas (SP) Brazil. 3 Department of Veterinary Medicine of the Universidade Federal do Paraná UFPR Curitiba (PR) Brazil; Michigan State University, Department of Small Animal Clinical Sciences USA. 4 Member in training, Hospital Universitário Evangélico de Curitiba (PR) Brazil. 5 Member in training, Hospital Universitário Evangélico de Curitiba (PR) Brazil. 6 Member in training, Hospital Universitário Evangélico de Curitiba (PR) Brazil. 7 Department of Veterinary Medicine of the UFPR Curitiba (PR) Brazil.

In patients treated with topical medication, the conjunctiva acts like a semi-permeable membrane that, along with the cornea, allows absorption of ocular hypotensive drugs. On the other hand, the conjunctiva responds to chronic treatment with antiglaucoma drugs with inflammation, scar formation, keratinization and neovascularization, which may directly affect its architecture and function (13) . Recently, several studies have evaluated the conjunctival changes induced by latanoprost (14)(15)(16) , but there is insufficient information about the consequences of the use of other PG analogues, such as travoprost and bimatoprost.
Previous studies with patients who received prolonged topical medication showed that both hypotensive drugs and its preservatives (benzalkonium chloride -BAK) may increase the number of inflammatory cells and fibroblasts in the substantia propria of the conjunctiva and reduce the number of goblet cells, inducing ocular surface modifications expressed as dry eye (13,(17)(18) . The length of administration, concentration and amount of medication were related to severity of side effects (17)(18) . Furthermore, there is strong evidence suggesting that these changes may increase the risk of failure of trabeculectomy (13,(18)(19)(20)(21) . However, most pieces of information on this topic were published before the introduction of PG analogues.
The purpose of this study was to compare histomorphometric changes induced by latanoprost, travoprost and bimatoprost with those induced by timolol maleate and BAK in rabbit conjunctiva.

METHODS
Fifty female New Zealand rabbits with similar age and weight were selected and divided into 5 groups of 10 animals. The rabbits were treated with bimatoprost 0.03% (Lumigan TM , Allergan, Irvine-CA, USA) (n=10), travoprost 0.004% (Travatan TM , Forth Worth, TX, USA) (n=10), latanoprost 0.005% (Xalatan TM , Pfizer, New York, USA) (n=10), timolol maleate 0.5% (Timoptic-XE TM , Merck, Sharp & Dome, New Jersey, USA) (n=10) or artificial tears (hypromellose and dextran 70) (Lacribell™ , Latinofarma, São Paulo, Brazil) (n=10). The drop volume was 27.05 µl for bimatoprost, 24.26 µl for travoprost, 23.81 µl for latanoprost, 27.9 µl for timolol maleate and 28.3µl for artificial tears. One daily drop of the selected substance was instilled onto the left cornea of each animal at 8:00 AM for 30 days, while the right eye served as a control. BAK was the preservative in all tested eye drops at the following concentrations: bimatoprost (BAK 0.005%), travoprost (BAK 0.015%), latanoprost (BAK 0.02%), timolol maleate (BAK 0.01%), and artificial tears (BAK 0.006%). Limbic superior conjunctival biopsies (5 mm x 5 mm) were performed at the 8 th and 30 th days in 5 rabbits of each group, except for the BAK group, where biopsies were performed only at the 30 th day. After conjunctival biopsies were performed at the 8 th day, these animals were excluded from the study, and the remaining animals underwent biopsies at the 30 th day. A conscious attempt to perform the conjunctival biopsy consistently at the same peri-limbic location was made. All samples were taken at the same time of the day (4:00 -5:00 PM), maintaining the same interval between time of instillation and time of biopsy for all groups.
The conjunctiva was immediately fixed with 10% formaldehyde for 24 hours, followed by routine paraffin embedding and staining with HE (hematoxylin-eosin) and PAS (periodic acid-Schiff). During the embedding procedure, an effort was made to ensure that the histological samples were placed with the epithelium facing the bottom of the histological cassette in order to avoid cuts that were not tangential to the plane of the tissue.
Morphohistometric quantitative analyses were performed with the software Image Pro-Plus version 4 (IP4) (Media Cybernetics, Silver Spring, MD). Digital images were acquired under 400x magnification and stored using the same software. Subsequently, two segments of 200 µm of length and with 5 µm of thickness of epithelial tissue were randomly selected from each examined field of all slides. These same-sized linear segments allowed measurement of the following parameters: number and diameter of blood vessels, number of goblet cells and conjunctival epithelium thickness. Diameter and thickness were measured using default pre-programmed virtual tools for tissues. Number of blood vessels and goblet cells were manually counted using a 200-µm virtual ruler.
Optical qualitative microscopy evaluation was made in linear conjunctival segments of 200 µm using a virtual ruler on digital images. The intensity of inflammatory infiltrate was analyzed qualitatively (absent, mild, moderate or severe) in all biopsy samples comparing pictures taken from 200 and 400x fields of each slide. The histologist (FMF) responsible for measuring all parameters was masked to the medication group of the rabbit. Repeat masked measurements performed by the same examiner (FMF) resulted in low coefficients of variability.
One-way ANOVA with a significance level of 5% was used to compare continuous variables. If any statistically significant difference was found, the data were further analyzed using post hoc comparisons with Fisher's test (for comparisons of up to three groups) or Tukey-Kramer test (comparisons of more than 3 groups). Differences were deemed statistically significant when p values were lower than 0.05.
The experimental procedures used in the investigation followed the ARVO Statement for the Use of Animals in Ophthalmic and Vision Reasearch, and the study was appro-ved by the Committee Review Boards of the Universidade de São Paulo and the Hospital Universitário Evangélico de Curitiba (CAPPESQ nº 161/05).

RESULTS
A significant increase in goblet cells was detected at the 8 th day in the travoprost group (p=0.0006). After 30 days of treatment, there was a significant increase in the number of goblet cells in the bimatoprost (p=0.0021) and latanoprost (p=0.009) groups. There were no significant changes in the number of goblet cells in the timolol and BAK groups (p>0.05) ( Table 1) (Figures 1 and 2).
There was a significant increase in epithelial thickness (p=0.0035) at the 30 th day in the group treated with timolol maleate (Table 1). Figure 3 illustrates the increased conjunctival epithelial thickness observed in this group, along with a higher number of epithelial layers. A statistically significant rise in the mean number of blood vessels was observed at the 30 th day in the latanoprost (p=0.0205) and timolol groups (p=0.0012) ( Table 1). There was an overall trend, though not statistically significant, toward reduced vascular diameter in the travoprost group at the 8 th day (p=0.0758), which became statistically significant at the 30 th day (p=0.0348) ( Table 1). No significant differences were observed between the BAK and control groups regarding epithelial thicknesses and number of blood vessels.
There were no significant differences between drop volume of all medications (mean of 25.668 µl, standard deviation of 2.051 µl and standard error of 0.837 µl).
In the control group and in the group treated with artificial   tears, the inflammatory infiltrate was focal, of mild intensity, with a preponderance of lymphocytes and rare neutrophils. The eyes treated with antiglaucoma medications, in addition to the same focal inflammatory pattern found in controls, presented a more diffuse inflammatory reaction, consisting of a mixed population of neutrophils and lymphocytes. This pattern was observed in a moderate intensity in the groups that used PG analogues and in severe intensity in the timolol group, which also showed an increase in subepithelial collagen density (Figure 3). There were no significant differences between the samples obtained at the 8 th and 30 th post-treatment days regarding intensity of the inflammatory reaction.
Moreover, these surface changes may increase the risk of failure when patients undergo glaucoma filtering surgery (27)(28) . Broadway et al. (18) observed that patients using different drops (beta-blocker, miotic + beta-blocker, or miotic + betablocker + sympathomimetic) for more than three years showed conjunctival changes, including an increased number of fibroblasts in the substantia propria. These changes were proportional to the amount of medication instilled and to length of drug administration. The reduction of cellular viability and apoptosis of conjunctival epithelial cells following the use of antiglaucoma medications have also been related to exposure time, number of drugs in use and their concentration, but there is little information available about the influence of PG analogues (13,21,(29)(30)(31)(32) .
Our study demonstrated that PG analogues induce less severe changes in the rabbit conjunctival epithelium than timolol maleate. In the group treated with timolol, there was a significant increase in conjunctival epithelial thickness and a marked increase of subepithelial collagen density, a finding that is consistent with the literature (13,18,21,24,(29)(30) . This finding may be secondary to epithelial cell edema, or to the presence of an inflammatory infiltrate, also found in greater intensity in timolol-treated eyes, and may partially explain the occurrence of dry eye symptoms in patients using timolol maleate. Several studies also demonstrated that timolol may induce a drop in goblet cells and a crystallization pattern associated with keratoconjunctivitis sicca (17,21,29) . Timolol has the larger decline of all drugs, but no significant differences were observed.
We observed a rise in goblet cells in the groups treated with PG analogues, a finding that was previously described for latanoprost (22)(23)(24)(25)(26)(27)(28)(29)(30)(31)(32)(33) . However, our study is the first to confirm that the increase in goblet cells may also be observed with other PG analogues, such as travoprost and bimatoprost. Although the consequences of an increased goblet cell population remain unknown, we hypothesize that this finding may explain why the long-term use of PG analogues is associated with less severe dry eye symptoms than those induced by timolol maleate.
Our study also demonstrated a rise in blood vessels in the timolol and latanoprost groups. Interestingly, bimatoprost and travoprost, which were shown to induce conjunctival hyperemia in glaucomatous patients (4)(5) , were not found to increase the number of blood vessels in the rabbit model.
A diffuse inflammatory infiltrate was noted in moderate intensity in the groups receiving PG analogues and in severe intensity in the timolol group, which also showed an increased subepithelial collagen density. The subclinical inflammation associated with the use of latanoprost was reported by others (14,31) , who described inflammatory cells in the conjunctival epithelium and substantia propria, demonstrated by histopathological analysis as well as by immunohistochemical markers (HLA-DR, IL-6, IL-8). Our study suggests that the inflammatory infiltrate observed with the use of latanoprost is also found with other PG analogues. The finding of a dense inflammatory infiltrate in eyes using timolol maleate is in agreement with previous reports (13,(20)(21)24,29,34) .
It was shown that long-term topical treatment of glaucoma induces an increase in collagen deposition (14,35) . Mietz et al. (14) investigated rabbits that used timolol, latanoprost, or the fixed combination of timolol and latanoprost for 18 months. At the end of the follow-up period, the conjunctiva showed an increased density of collagen fibers in the substantia propria of eyes treated with timolol and the fixed combination of timolol and latanoprost, but not in eyes treated with latanoprost alone. In our study, this finding was confirmed for timolol maleate and was detected earlier than in other reports (14,35) . An up-regulation of matrix metalloproteinases may explain why PG analogues did not induce changes in collagen density. Metalloproteinases are essential to control tissue remodeling and may influence the scarring process of filtering blebs after glaucoma surgery (14)(15)(35)(36) . Latanoprosttreated eyes showed an up-regulation of matrix metalloproteinases (MMP-3) and their tissue inhibitors (TIMP-2), leading to reduced extracellular accumulations of amorphous material in the conjunctiva.
Several studies suggested that the surgical outcome of trabeculectomy is less favorable in patients who received previous topical antiglaucoma therapy (13,19,24,28,37) . Broadway et al.s (19) identified preoperative subclinical conjunctival inflammation induced by previous topical medication as a risk factor for failure of trabeculectomy. The reduction of success rates of glaucoma filtering procedures is probably due to the activation of fibroblasts from the conjunctiva, Tenon's, and episclera, leading to the formation of scar tissue surrounding the scleral flap (27,37) . The presence of an inflammatory infiltrate (induced by all antiglaucoma medications, but more severe in the timolol group) may induce fibroblast activation (13,17,29,37) . Furthermore, the increased collagen density detected in the timolol group may already be a consequence of enhanced fibroblast activity.
Clinical, laboratory and experimental studies have suggested that conjunctival morphological changes were strongly associated with BAK, the most frequently used preservative in ophthalmic topical medications (16,(38)(39) . Pisella et al. (40) , using impression citology to compare toxicity of latanoprost and unpreserved or preserved timolol maleate, evaluated inflammatory markers and MUC5ACrelated mucin production in the conjunctiva of chronic users in a case-control ex-vivo study. Only patients treated for at least one year were included in this study and compared to normal subjects. HLA-DR, ICAM-1 and mucin were evaluated in a masked manner by flow citometry. For the in vitro study, a human conjunctiva derived cell-line was treated with 0.02% BAK-containing latanoprost or timolol, unpreserved timolol, or 0.02% BAK alone for 15 minutes, and cell viability and chromatin condensation were evaluated. These in vitro studies demonstrated that BAK-containing latanoprost and timolol exhibited higher proinflammatory and proapoptotic effects on conjunctival cells than unpreserved timolol. Latanoprost was found to induce fewer changes than preserved timolol, which is in agreement with our study.
The only study that investigated the effects of the three available PG analogues on the conjuntiva was reported by Guenoun et al. (32) . The authors performed in-vitro and exvivo experiments using human conjunctiva-derived cell lines to evaluate toxicity and expression of human inflammatory markers induced by PG analogues and BAK. PG analogues did not induce expression of inflammatory markers and even reduced the expression of ICAM-1 and PECAM-1 in the solutions with higher concentrations of BAK. The results of the previously described studies (32,40) suggest a protective effect of PG analogues against preservative (BAK) toxicity due to possible antioxidative properties.
Contrary to the literature available, the present investigation was not able to detect marked conjunctival changes in the BAK group. This discrepancy may be due to short exposure time to the preservative in our study, to its low concentration in the solution used (0.006% compared to other studies that used 0.02%), or to the association of BAK and hypromellose and dextran 70 in artificial tears (41) .
In conclusion, although a moderate, diffuse inflammatory infiltrate was observed in PG-treated eyes, no changes in conjunctival epithelial thickness or subconjunctival collagen density were observed with these medications, suggesting that these drugs induce fewer changes than timolol maleate in the rabbit conjunctiva. Further studies are needed to investigate the influence of PG analogues versus other classes of antiglaucoma medications on the success rates of eyes undergoing trabeculectomy.