Topical glaucoma medications – Clinical implications for the ocular surface

Glaucoma is a leading cause of irreversible blindness. The use of topical eye drops to reduce intraocular pressure remains the mainstay treatment. These eye drops frequently contain preservatives designed to ensure sterility of the compound. A growing number of clinical and experimental studies report the detrimental effects of not only these preservatives but also the active pharmaceutical compounds on the ocular surface, with resultant tear film instability and dry eye disease. Herein, we critically appraise the published literature exploring the effects of preservatives and pharmaceutical compounds on the ocular surface.


Introduction
The tear film is classically divided into three individual layers: inner mucin, intermediate aqueous and outer lipid.However, the Tear Film and Ocular Surface Society Dry Eye Workshop II (TFOS DEWS II) tear film subcommittee recently advocated a two-layer model with an inner mucoaqueous layer and a superficial lipid layer [1].The superficial lipid layer consists of meibum, the secretory product of the meibomian glands (MGs), which are modified sebaceous glands dispersed along the tarsal plates of the upper and lower eyelids [2].Meibum helps stabilize the tear film, prevents evaporation and protects against external harmful agents [3].Dry eye disease (DED) is the end result of a multifactorial pathway leading to disruption of tear film homeostasis [4].The main underlying aetiologies include aqueous-deficient and evaporative dry eye, entities that often coexist and overlap along a spectrum, where the evaporative form is the most frequent.Meibomian gland dysfunction (MGD) is the most common cause of evaporative dry eye.
Glaucoma is a chronic, progressive optic neuropathy frequently associated with increased intraocular pressure (IOP) and is the leading cause of irreversible blindness globally [5].The aqueous humour in the eye is produced in the ciliary body and mainly drains through the trabecular meshwork.The resulting pressure gradient determines the IOP, which has an average physiological range of 10-20 mmHg [6].The use of topical IOP-lowering eye drops remains the primary treatment for glaucoma [7,8].The objective of this review is to critically assess the clinical effects of topical medications on the ocular surface.

Methods
A PubMed and EMBASE search was conducted on April 1st of 2022.The search included any combination of the three keyword groups: (i) 'dry eye disease', 'meibomian gland dysfunction', or 'meibography', and (ii) 'topical medications' or 'eye drops' and (iii) 'glaucoma'.The PubMed search yielded 5027 results, while 4015 articles were identified in EMBASE.The inclusion criteria included relevance, full-text access and English language, and the exclusion criteria included lack of relevance, non-English language and abstract only.Letters to the editor, conference abstracts, review articles and case reports were excluded for clinical evaluation, although review articles have been used for background information.The remaining articles were evaluated for relevance, first based on the title and second on the abstract.Finally, the full text was appraised for inclusion based on relevance: studies including topical glaucoma medications and specification regarding the use of preservatives and quantitative results and outcome measures pertaining to the ocular surface (including symptoms, Schirmer test, tear film break up time, ocular surface staining, tear osmolarity and inflammatory markers).Both animal and in vitro studies describing the effect of topical glaucoma medications or preservatives on the ocular surface, tear film or meibomian glands were generally not included, although some are used for supplementary information.A critical appraisal of the included articles' reference sections was performed as well.In total, 253 research articles were included in the present review.

Effect of preserved topical glaucoma medications on the ocular surface
Most available topical IOP-lowering medications contain preservatives, among which benzalkonium chloride (BAC) is one of the most common [8][9][10][11].BAC is a quaternary ammonium exhibiting detergent-like qualities, affecting the cell membrane and metabolic processes [12].BAC induces apoptosis and necrosis and reduces the amount of occludin in the corneal epithelium [13,14].Furthermore, BAC has been shown to result in upregulation in vitro of the proinflammatory cytokines interleukin-1 (IL), IL-10, IL-12, C-reactive protein and tumor necrosis factor [12].A recent murine and in vitro study examined the effects of BAC on the cornea and cultivated primary mouse corneo-limbal epithelial cells [15].The mice received 5 μl BAC topically either once or twice daily for seven days and were randomly assigned BAC concentrations of 0% (phosphate buffered saline controls), 0.05%, 0.1% or 0.2%.Analysis of the murine corneas revealed a dose-dependent BAC toxicity ranging from punctate fluorescein staining to severe epithelial defects.Furthermore, cell contraction, vacuolation and necrosis were noted in cultured corneo-limbal epithelial cells upon exposure to BAC.Upon analysis of 27 common ocular hypotensives, the pH was 4.0-7.4(mean = 6.25), and the free radical concentration was 0-4.54 mmol/L (mean = 0.66 mmol/L) [16].
Several studies have revealed a higher prevalence of dry eye symptoms among patients using IOP-lowering eye drops when compared to healthy, age matched controls [17][18][19].Significant alterations in the Schirmer test [17,[20][21][22][23], tear film break-up time (TBUT) [17,[20][21][22]24], conjunctival staining [17,[20][21][22], corneal and conjunctival sensitivity [25,26], corneal clarity [27], ocular redness [28,29], lipid layer thickness [30], and tear osmolarity [23] have been found in patients under glaucomatous treatment.Moreover, the use of topical antiglaucoma medications is the most common cause of drug induced cicatrizing conjunctivitis [31].A recent case series including 23 patients with this diagnosis reported BAC as the most common instigating agent [31].One case, however, was caused by stabilized oxychloro complex preserved antiglaucoma medication, while another subject reacted to various PF topical medications, including apraclonidine, bimatoprost, dorzolamide with timolol and timolol in isolation.Compared to healthy controls, proteomic analysis revealed upregulation of S100-A8, S100-A9, mammaglobin B and 14-3-3 ζ/δ in patients treated with preserved topical medications [32].Alterations of clinical parameters among patients treated with preserved glaucoma medications are listed in Table 1.A short term randomized study reported significantly decreased tear film stability 3 h and three days following instillation of preserved carteolol, with no such findings among patients receiving the unpreserved equivalent [33].Positive correlations between the number of topical IOP-lowering medications administered and the severity of signs and symptoms have been reported [34][35][36][37].Further, a positive correlation between the number of BAC-containing eye drops, Ocular Surface Disease Index (OSDI), TBUT and osmolarity was described [38].Fukuchi et al. reported superficial punctate keratitis in 51% of 749 eyes treated with IOP-lowering eye drops [39], whereas Rossi et al. found superficial punctate keratitis in 31.7% and ocular surface disease in 41.6% of 233 patients, these data correlated to the daily number of instillations, age, lower IOP values, total BAC exposure, and the Glaucoma Symptom Scale questionnaire [40,41].No difference in corneal staining was observed in patients treated for five days with either latanoprost/0.02%BAC, travoprost/0.015%BAC or bimatoprost/0.005%BAC [42].The use of preserved IOP-lowering medications was associated with an increased incidence of ocular signs of dry eye such as conjunctival redness, superficial punctate keratitis and blepharitis [35].Additionally, both signs and symptoms of dry eye decreased upon changing from preserved eye drops to a preservative-free (PF) alternative [35,43].
Furthermore, an increase of TUNEL + cells was reported in human conjunctival cells among patients treated with BAC-preserved ocular hypotensives compared to healthy, age-and gender-matched controls [44].
A decreased number of sub-basal nerves and impaired corneal sensitivity through esthesiometry were found in patients treated with IOP-lowering eye drops preserved with BAC [45].However, Kaminski et al. found no changes in mean corneal thickness, corneal endothelial cell count or corneal sensibility following 90 days of treatment with 2% dorzolamide in 20 patients [46].In addition, latanoprost did not appear to disrupt the corneal epithelial barrier function as measured by fluorophotometric fluorescin uptake in 10 patients treated for 1 month nor in 14 patients treated for 6 months [47].On the contrary, both BAC preserved and PF timolol caused a significant increase in fluorophotometric corneal fluorescin uptake in 20 healthy volunteers 30 minutes after instillation, indicating corneal epithelial barrier function disruption [48].This effect was more pronounced in preservative-containing (PC) timolol which also decreased non-invasive break up time (NiBUT), contrary to the PF formulation.Similar findings were reported in patients treated with BAC preserved timolol when compared to untreated glaucomatous patients and healthy controls [49], as well as in patients treated with BAC preserved timolol compared to healthy controls and patients with cataract treated with pyrenoxine [50].
When glaucomatous subjects treated with various PC PGAs were compared to healthy controls, the glaucoma group demonstrated lower keratocyte density and decreased central corneal thickness (515.2 vs 549.6 μm) measured with IVCM [51].However, a prospective randomized study including 369 glaucoma patients treated with either latanoprost, timolol or latanoprost/timolol fixed combination (FC) for 12 months found no changes in corneal endothelial cell density or central corneal thickness [52].A prospective study including 66 patients treated with latanoprost and 42 patients treated with latanoprost and timolol (preservative status not described) reported a significant increase in central corneal thickness and corneal hysteresis but no change    in corneal resistance factor [53].In a two-year follow-up study latanoprost was demonstrated to decrease the conjunctival thickness from 201.45 μm to 167.81 μm [54].No significant difference was found in patients treated with carteolol with measures of 198.76 μm at baseline and 201.23 μm after two years.However, the preservative status in the included medications is not described.The conjunctival thickness in 49 glaucomatous eyes treated with various topical medications mostly preserved with BAC was 182.76 μm, significantly less than the 235.02 μm measured in the 51 healthy control eyes [55].In a study employing IVCM and impression cytology, the authors reported a transient increase of goblet cell density in patients treated with BAC-preserved latanoprost after one month, returning to baseline at the 6-month follow-up [56].However, in patients treated with PF-PGA, the effect of increased goblet cell density remained, also at the 6-month measurements.Conversely, levobunolol has been found to cause a significant decrease in goblet cell density, which is exacerbated by the addition of BAC [57].Cennamo et al. compared changes to the conjunctiva of 20 human specimens treated with BAC preserved ocular hypotensives and 20 healthy, ageand sex-matched controls through a ferning test and analysis of impression cytology with a light microscope as well as a scanning electron microscope [58].The mean grades of the ferning test, impression cytology from light microscopy and scanning electron microscopy were 2.52, 2.52 and 2.55, respectively, for the treatment group and were significantly lower in the control group (1.22, 1.25 and 1.15, respectively).The treatment duration was correlated with a reduced number of microvilli identified with scanning electron microscopy.Impression cytology from patients treated with timolol, latanoprost, dorzolamide, timolol/latanoprost FC and timolol/dorzolamide FC revealed higher cytology scores with a significant degree of metaplasia when compared to healthy controls [59].These changes were more profound in patients treated with fixed combinations when compared to monotherapy.Conversely, no correlation between cytological grading, age, sex, type of medication, number of formulations or duration of treatment was found in conjunctival impression cytology samples from patients treated with betaxolol (24 eyes), levobunolol (20 eyes), timolol (32 eyes), pilocarpine (22 eyes), betablocker and pilocarpine (52 eyes), betablocker and dipivefrin (34 eyes) or maximum therapy (32 eyes) when compared to 51 healthy control eyes [60].However, the degree of conjunctival metaplasia was significantly increased in eyes exposed to topical therapy.A later study compared the corneal epithelial thickness and number of microvilli in 16 patients treated with PC topical hypotensives and 6 healthy, age matched controls [61].Following an initial decrease in corneal epithelial thickness, the authors report a gradual increase in this thickness as the number of microvilli decreased.Impression cytology was taken from 20 eyes of 12 patients prior to and after 1, 3 and 6 months of treatment with BAC preserved latanoprost [62].There was no change in non-goblet epithelial cell density, but a significant reduction in cell size was observed.Goblet cell density increased during the first month and gradually returned to baseline with a longer treatment duration.Conjunctival impression cytology and mucus staining was performed on human specimens in a prospective study involving healthy controls, patients under ongoing treatment with BAC-preserved timolol and patients initiated with the same medication [63].Patients under ongoing treatment demonstrated a significantly lower goblet cell density compared to healthy controls as well as a deteriorated mucus layer.In the cohort initiated on topical medications impression cytology analysis disclosed a gradual decrease of goblet cells and progressive degradation of the mucus layer.Moreover, both groups receiving BAC preserved timolol had worse Schirmer test, tear film stability and ocular surface staining.A later, randomized, prospective study comparing PF-timolol gel to BAC-preserved timolol drops and healthy controls reported a significant decrease in goblet cell density among patients receiving PC formulations [64].However, no significant differences were found in patients in the PF cohort when compared to both baseline and controls.Patients receiving BAC-preserved latanoprost presented with increased levels of the inflammatory marker matrix metallopeptidase 9 (MMP-9) as well as signs and symptoms of DED when compared to patients treated with PF tafluprost [65].When comparing biomarkers in eyes treated with BAC preserved latanoprost or bimatoprost both medications caused an upregulation of MMP-2, MMP-9 and tissue inhibitors of metalloproteinase (TIMP)-1 when compared to non-glaucomatous controls with cataract [66].The levels of MMP-1 were 1404, 1304, and 1079 pg/ml in bimatoprost, controls and latanoprost, respectively.Moreover, the eyes treated with latanoprost had an increased expression of cytokines related to tissue remodelling whereas the eyes treated with bimatoprost had an increased expression of cytokines associated with allergic reactions.Similarly, upregulation of MMP-2, MMP-3, MMP-9, TIMP-1, TIMP-2 and TIMP-3 were noted upon comparing 25 glaucoma patients treated with various BAC-preserved topical medications to 7 healthy controls [67].Moreover, expression of MMP-1 and MMP-3 correlated with duration of treatment with pilocarpine.However, Pradhan et al. found no difference in the expression of MMP-2 or MMP-9 when comparing concentrations in the aqueous humour of glaucomatous patients treated with PGAs and non-glaucomatous controls [68].A study examining the effects of BAC preserved latanoprost in patients with glaucoma compared to healthy controls including a murine study arm reported increased levels of MMP-9 and decreased levels of TIMP-1 in both humans and mice exposed to treatment [69].These findings were corroborated in a later study, where the authors also reported a higher expression of MMP-9 with an increasing number of medications containing BAC [70].An extracellular matrix metalloproteinase inducer and human leukocyte antigen (HLA)-DR were reportedly increased in the conjunctiva of 18 glaucoma patients treated with BAC preserved ocular hypotensives for at least six months when compared to eight healthy controls through impression cytology and flow cytometry [71].These results were more pronounced in patients receiving higher doses of BAC.The same study had an in vitro arm which indicated that cyclosporin A could inhibit the inflammatory effects of BAC; these findings were corroborated clinically [72].Furthermore, rebamipide demonstrated a protective effect against ocular surface damage induced by BAC preserved latanoprost and timolol [73].In an ex vivo and in vitro study Pisella et al. compared the effects of BAC-preserved latanoprost and timolol to those of unpreserved timolol and healthy controls through impression cytology [74].They reported an increased expression of HLA-DR and intercellular adhesion molecule-1 as well as a decreased MUC-5AC in the BAC receiving cohorts, more so in the PC-timolol group.No difference was described between healthy controls and patients treated with unpreserved timolol.In the in vitro arm of the study, a human conjunctiva-derived cell line was exposed to either latanoprost or timolol preserved with 0.02% BAC or 0.02% BAC alone.All three preparations caused apoptosis, although interestingly more so in the isolated BAC group.The authors hypothesize that the active pharmaceuticals may have a protective effect against the deleterious effect of BAC on conjunctival cells.Another study evaluating the inflammatory effects of BAC reported upregulation of the chemokine CX3CL1 and HLA-DR among patients receiving BAC preserved glaucoma medications when compared to patients on PF formulations, possibly through the TNF-α pathway [75].The two chemokine receptors CCR5 and CCR4 were used as markers of the T helper 1 and T helper 2 cells, respectively [76].The authors reported upregulation of both in patient groups treated with topical antiglaucomatous agents compared to healthy controls.However, overexpression of the HLA-DR class II antigen only reached statistical significance among patients on a multidrug regimen compared to controls.Comparison of cytokine levels in tears of 21 patients treated with PC ocular hypotensives and untreated healthy controls revealed increased concentrations of IL-1β, IL-6, TNF-α, T helper 1 associated interferon-γ and IL-2 as well as T helper 2 associated IL-4, IL-5 and IL-10 in the tears of treated eyes [77].Upregulation of HLA-DR was described in patients on monotherapy with BAC-preserved latanoprost, betaxolol or timolol when compared to healthy controls [78], in patients treated with BAC-preserved latanoprost compared to controls instilling BAC-preserved artificial tears [79], and in patients treated with PC latanoprost, bimatoprost and travoprost [80].Moreover, latanoprost preserved with BAC might increase the number of eosinophils, decrease the number of lymphocytes and cause alterations in both the ferning test and impression cytology in glaucomatous patients with allergic conjunctivitis [81].In a randomized, double-blind clinical trial 20 healthy volunteers were treated with 0.01% BAC in one eye and placebo in the fellow eye three times daily for 12 weeks [82].Langerhans cell density in the central and peripheral corneal epithelium was measured through confocal laser-scanning microscopy at baseline and after 1, 6, 12 and 16 weeks.In the BAC group, an increased Langerhans cell density was seen at 6 and 12 weeks in the central and peripheral cornea, respectively.The Langerhans cell density also increased in the peripheral cornea of the placebo group at week 12, although not to the same extent as in the BAC group.All values returned to normal after four weeks wash-out.A reversal of upregulated fibroblasts and inflammatory cells in human conjunctiva following topical antiglaucoma therapy was also reported following preoperative cessation of medical therapy and topical steroid treatment for 30 days [83].
Conversely, the number of collagen fibres in human eyes treated with latanoprost was significantly decreased in those treated with timolol but not when compared to controls [84].Moreover, the amount of amorphous material was increased in both treatment groups compared to controls (but less in latanoprost when compared to timolol).MMP-1, MMP-3, TIMP-2, and TIMP-3 were upregulated in the epithelial and stromal cells of latanoprost-treated eyes, whereas timolol-treated eyes demonstrated infiltration of inflammatory cells and macrophages.A reduction of pro-inflammatory lipid mediators by as much as 50% was reported in patients treated with topical IOP lowering medications following trabeculectomy [85].Recently, a Polish research team examined the effect of PGAs and BAC on parameters of oxidative stress in the tear film [86].Compared to healthy controls, evidence of increased oxidative stress was found among patients receiving PGAs, a finding that was more profound in patients receiving PGAs containing BAC.The same research team also examined the effect of PF timolol as well as BAC-preserved brimonidine and timolol on the biomarkers of oxidative stress in the tear film [87].In this study, the authors reported significant differences between patients treated with preserved medications vs those treated with PF timolol and healthy controls.In a later study the authors reported increased activity of superoxide dismutase and catalase as well as decreased content of sulfhydryl groups in patients treated with BAC preserved dorzolamide and brinzolamide when compared to healthy controls [88].Additionally, the levels of oxidative stress were lower in patients treated with PF dorzolamide and controls compared to those treated with preserved formulations.Among patients requiring maximum topical glaucoma therapy to lower IOP, dual maximum therapy (two fixed-drug combinations) has been shown to be non-inferior to triple maximum therapy (one fixeddrug combination and two single preparations) concerning IOP control; however, signs and symptoms of DED are significantly better with dual therapy [89,90].
Being a lipophilic molecule, BAC has the potential to penetrate and accumulate in deeper ocular structures.Mass spectrometric studies have revealed the presence of BAC in the trabecula and lens in both animal and human specimens after prolonged exposure [91,92].In a combined animal and human experiment Baudouin et al. examined the histological and inflammatory changes in the conjunctiva and trabeculae of patients and rats treated with various ocular hypotensives [93].The authors reported a dose-dependent infiltration of cells expressing inflammatory and/or fibroblastic markers in both the conjunctiva and trabeculae of human specimens.In rats, isolated BAC and BAC preserved timolol demonstrated similar changes, although, not in the PF or control cohort.A histological examination of conjunctival biopsies demonstrated a time-and dose-dependent effect on epithelial hyperplasia, keratinization and infiltration of lymphocytes, macrophages, fibroblasts, and mast cells in patients treated with preserved antiglaucomatous medications when compared to controls [94].These findings do not agree with those of another group comparing the histological changes in the conjunctiva of glaucoma patients treated with maximal medical therapy and healthy controls, in which no significant differences in any cell line were found [95].

Effect of preservative-free topical glaucoma medications on the ocular surface
Although the cumulative scientific evidence indicates worsening signs and symptoms with the use of BAC, it appears that several PF medications may also have deleterious effects on the ocular surface.A fluorophotometric study found a significant but transient decrease in tear production due to treatment with PF timolol in 24 patients [155].Increased expression of HLA-DR, IL-6, IL-8 and IL-10 have been described as a result of unpreserved timolol, although not to the same extent as in PC formulations [156].A recent study compared the expression of inflammatory cytokines in the tears of glaucoma patients treated with PC and PF ocular hypotensives to patients with DED and healthy controls [157].The glaucoma cohort had significantly higher IL-6, TNF-α and vascular endothelial growth factor (VEGF) as well as lower IL-4 when compared to the DED group.When comparing glaucoma patients to controls, only IL-6 reached statistical significance.Between DED and controls, IL-1β, IL-6 and IL-10 were increased while VEGF were decreased in the DED group.Subgroup analysis found significantly increased levels of the pro-inflammatory and pro-apoptotic IL-1β in patients applying preserved medications compared to the PF cohort.
Several studies have revealed detrimental changes in the clinical parameters of glaucoma patients treated with PF compounds when compared to controls [17,21,22,158,159], whereas others have not [45].Clinical alterations of the ocular surface among patients treated with unpreserved glaucoma medications are listed in Table 2.A prospective, randomized cross-over study demonstrated a superior IOP-lowering effect of PF bimatoprost when compared to PF latanoprost, although PF latanoprost was associated with slightly lower hyperaemia scores [160].
Multiple clinical studies have reported significant worsening of signs and/or symptoms among patients receiving BAC-containing eye drops compared to patients using PF medications [17,35,45,159,[161][162][163][164][165][166].However, Aptel et al. found no differences in efficacy or subjective tolerability when comparing BAC-preserved latanoprost to PF latanoprost in a 12 week follow up study with cross over between medications at 6 weeks [167].Similarly, when comparing BAC-preserved bimatoprost/timolol to the PF equivalent in a randomized 12-week follow-up study, Goldberg et al. reported a comparable self-reported safety profile between the two and noninferiority of the PF formulation regarding IOP-lowering capability [168].Four recent studies demonstrated an improvement of subjective and objective findings by the change from PC-PGA to PF-PGA [169][170][171][172].A reduction in anterior chamber flare from 6.75 photon counts per millisecond (ph/ms) at baseline to 5.78, 5.41 and 5.50 ph/ms at 1, 2 and 3 months respectively (all p < 0.05) was reported in 22 patients changed from BAC preserved to PF latanoprost [173].Similar findings with improved signs and symptoms of DED were found in a prospective study performed by Iester et al. comparing the change from PC to PF betablockers [174].The reported changes in clinical signs at baseline vs. three-months were: eyelid erythema (0.46-0.13), conjunctival hyperemia (0.97-0.33), follicular hyperplasia (0.36-0.08),TBUT (9.82-11.5)and Schirmer test (13.46-15.41).Upon evaluating the effects of changing from BAC preserved latanoprost to PF tafluprost 158 patients experiencing ocular signs and/or symptoms of DED were included in a multicentre study that comprised 12 centres based in Finland, Sweden, and Germany [175].No changes in IOP were recorded.Compared to baseline, the number of patients reporting various symptoms at 12 weeks were: irritation/burning/stinging (56.3%-28.4%),foreign body sensation (49.4%-27.1%),tearing (55.1%-27.1%),itching (46.8%-26.5%)and dry eye sensation (64.6%-39.4%).Percentage of patients with abnormal clinical parameters: TBUT (94.9%-71.6%),corneal fluorescin staining (81.6%-40.6%),conjunctival fluorescin staining (84.2%-43.2%),blepharitis (60.1%-40.6%),conjunctival hyperemia (84.2%-60.0%)and Schirmer test (71.5%-59.4%).All changes in subjective and clinical parameters reached statistical significance.Moreover, a reduction in the number of patients with increased values of HLA-DR positive epithelial cells were reported at six weeks; this reduction was not significant at the final control.The proportion of patients expressing an abnormal level of MUC-5AC-positive goblet cells improved at both 6-and 12-week follow-ups.In a multicentre study by Pisella et al., 4107 patients were included, among whom 956 underwent treatment modification and were seen at a second visit [35].Seven hundred and eighty patients were initially on PC medications and switched to a PF formulation.Patient reported symptoms such as discomfort upon instillation and at least one symptom in-between instillations went from 57.6% and 82.7% to 11.7% and 35.8%, respectively.Clinical signs were as follows: palpebral sign (35.7%-14.5%),conjunctival sign (68.9%-21.9%)and superficial punctate keratitis (25.4%-5.3%).A second multicentre study included 9658 glaucomatous patients, 6083 of whom were seen at a second visit after modification of the treatment regime [176].Among patients changing from PC to PF formulations a reduction in pain or discomfort from 52.4% to 7.8% was reported.Clinical signs of DED were changed as follows: anterior blepharitis (26.38%-8.71%),posterior blepharitis (9.13%-2.80%),eczema (11.00%-2.93%),hyperemia (64.36%-15.99%),follicles (21.13%-5.28%),fluorescin staining (13.65%-2.48%)and superficial punctate keratitis (28.99%-5.66%).Thirdly, 721 patients were included in a prospective multicentre study, 64.8% of whom the antiglaucomatous topical treatment was changed to PF latanoprost [177].Subsequent to this alteration, a reduction in signs and symptoms of dry eye as well as the use of artificial tears was described.A prospective, multicentre clinical audit evaluating the effect of changing from PC to PF formulations reported findings contrary to the aforementioned trials [178].The authors noted a significant reduction in symptom score and the number of patients with decreased TBUT when comparing baseline values to follow-up measurements.However, this reduction was observed in both patients switched to PF formulations and in those kept on PC medications with no significant between group differences.Nevertheless, patients who converted to unpreserved eyedrops experienced a decreased use of artificial tears.Finally, another recent multicentre study involving 577 patients reported significant improvement of IOP as well as signs and symptoms of DED among patients demonstrating insufficient response to betablocker or PGA monotherapy following transition to PF tafluprost/timolol fixed combination [179].The number of participants treated with PC medications prior to transition was not described.

Impact of dry eye disease on the intraocular pressure
Multiple studies have shown reduced IOP upon treatment of DED in glaucoma patients, occasionally to an extent that allows for a decrease in the number of medications required for IOP control [202][203][204][205]. IOP reduction was seen following addressing DED through punctal plugs [202,206], change to PF equivalents [203,205], PF lubricants, eyelid hygiene and antibiotics [203][204][205], and non-steroidal anti-inflammatories or cyclosporine A [203].Some authors have hypothesized that the association between an inflamed ocular surface and uncontrolled IOP may be due to penetration of inflammatory mediators from the surface to the deeper ocular structures or possibly due to direct detrimental effects of drugs or preservatives on the trabecular meshwork, sclera and episclera [203,205,207].Such hypotheses are supported by a study by Baudouin et al. which described loss of trabecular cells and signs of inflammation in surgically excised trabecular meshwork of patients treated with PC IOP-lowering eye drops and induction of apoptosis and inflammation of trabecular cells exposed to BAC in vitro [207].Furthermore, in that study the authors demonstrated that      topically administered BAC can reach the trabecular meshwork in rabbits using mass spectrometry imaging; additionally, increased IOP and apoptosis of trabecular cells was noted in rats given subconjunctival injections of BAC [207].Moreover, increased corneal permeability following BAC exposure was demonstrated in DED patients treated with PC and PF artificial tears, in which case the PF cohort demonstrated decreased permeability whereas patients receiving BAC had increased corneal permeability [208].Finally, the treatment of ocular surface disease among glaucoma patients appears to improve the measurement quality of optical coherence tomography and measured thickness of the retinal nerve fibre layer [209], decrease visual field testing times and enhance results [210], as well as bringing the number of conjunctival goblet cells detected through IVCM closer to normal [211].

Critical appraisal of preservatives
Alternative preservatives exist, such as purite, polyquaternium-1 and SofZia.Purite is reportedly well tolerated, with mild adverse effects [212,213].Animal and in vitro studies indicate a lower cytotoxicity of polyquaternium-1 when compared to BAC [214][215][216][217][218][219][220].An impression cytology study found polyquaternium-preserved travoprost to be better tolerated and safer than BAC-preserved travoprost [221].A later randomized study found upregulation of IL-1β, IL-6 and IL-8 in patients treated with BAC preserved medications when compared to patients treated with PF of polyquaternium-1 preserved medications [222].Some clinical studies have noted increased tolerability and decreased adverse effects with polyquaternium-1 when changed from BAC [223][224][225], while others have found a comparable safety profile between the two preservatives [125,[226][227][228][229]. It is worth noting that among the studies reporting a favourable adverse effect profile after changing preservative, the active pharmaceutical had been changed as well.Thus, it is unclear whether the positive effect stems from the change of preservative, antihypertensive or both.This uncertainty was further exemplified by a study comparing bimatoprost/BAC, latanoprost/BAC, travoprost/polyquaternium-1 and PF tafluprost, with tolerability measured using the OSDI symptom score [158].Travoprost/polyquaternium-1-treated patients had significantly better scores, whereas latanoprost/BAC -treated patients had the worst scores.It is noteworthy that even though tafluprost was unpreserved, it did not display a superior side-effect profile.In a study by Kim et al., polyquaternium-1 did not show a significantly favourable profile concerning objective signs of DED and MMP-9 expression compared to BAC [70].A recent retrospective study analysed the effects of latanoprost/BAC, travoprost/polyquaternium-1, bimatoprost/BAC, brimonidine/purite and brimonidine/BAC on corneal hysteresis and corneal resistance factor as compared to healthy, age-matched controls [230].Corneal hysteresis and resistance factor in controls were 10.26 and 10.60, respectively.Significant differences were reported among patients using bimatoprost/BAC (8.50 and 8.77) and brimonidine/BAC (8.77 and 8.93) with insignificant findings in the other groups.
In vivo and in vitro studies have indicated a better safety profile for SofZia as compared to BAC [215,[231][232][233][234]. Some clinical studies have reported improved symptoms and/or clinical parameters with SofZia when compared to BAC [110,[235][236][237][238], while others have not [144,239,240].Among the clinical studies reporting improvement after changing from BAC to SofZia, the active pharmaceutical component was changed from latanoprost to travoprost.Hence, determining the precise clinical effect produced by changing preservative is complicated due to the manipulation of two factors concomitantly.

Design of included studies
Table 5 summarizes the included studies' different study designs.The majority of the included clinical studies fall within Levels 1 and 2 of evidence as delineated by the modified American Academy of Ophthalmology Preferred Practices guidelines [241] (Table 4).Furthermore, several in vitro and animal studies have been performed, elucidating aspects not possible or considered not ethical to evaluate clinically.

Discussion
Upon reviewing the literature, examples from both the active pharmaceuticals and the preservatives in topical glaucoma medications appear to exert deleterious effects on the ocular surface.Interventional studies have shown a decrease in symptoms and improvement in clinical signs upon conversion from PC to PF PGA and betablockers [159,161,242].However, the underlying pathophysiological mechanisms responsible for the detrimental effects of BAC and ocular hypotensives on the ocular surface are generally still in need of further elucidation.Inflammatory pathways such as induction of apoptosis, cytokines and oxidative stress have been shown to be involved [12][13][14]86,87,157,207,260].Moreover, through its detergent properties, BAC may compromise the lipid layer, thus predisposing treated eyes for excessive evaporation,

Clinical studies
Level 1 Evidence obtained from at least one properly conducted, well-designed, randomized controlled trial or evidence from well-designed studies applying rigorous statistical approaches Level 2 Evidence obtained from one of the following: a well-designed controlled trial without randomization, a well-designed cohort or case-control analytic study, preferably from one or more centres, or a well-designed study accessible to more rigorous statistical analysis Level 3 Evidence obtained from one of the following categories: descriptive studies, case reports, reports of expert committees, and expert opinion F. Fineide et al.
causing dehydration and subsequent inflammation, initiating the vicious cycle of DED through several mechanisms, as illustrated in Fig. 1.The resulting hyperosmolarity and tear film instability may further instigate damage to the ocular epithelium with succeeding inflammation, exacerbating tear film instability, evaporation and hyperosmolarity [23,[267][268][269].Purite, polyquaternium-1 and SofZia appear to have a less deleterious effect on the lipid layer of the tear film [247].It has been noted that topical medications may induce subclinical inflammation of the conjunctiva [243,246], increased inflammatory cells and mast cells [245,246], as well as increased immunoglobulin E (IgE) [244].Whether these alterations result from active pharmaceutical compounds, preservatives, or both, remains unknown.More than 110 million people worldwide are expected to have glaucoma by 2040 [270].With current treatment regimens, the prevalence of ocular surface disease among patients receiving topical glaucoma medications ranges from 38.5% to 75% [9].This illustrates the current need for prospective longitudinal studies comparing the effects of different pharmaceuticals and preservatives on the ocular surface and MGs.Several studies have shown that a change from preserved to unpreserved formulations may decrease the signs and symptoms of DED in glaucoma patients.Moreover, treating DED in afflicted glaucoma patients could improve IOP and reduce the number of formulations needed to reach treatment goals.Therefore, it is vital that clinicians recognize DED in this population and provide appropriate treatment.A decrease in the DED prevalence among glaucoma patients may increase compliance and efficacy of glaucoma medications, thereby improving the prognosis and help to preserve sight.Miriam Kolko: Miriam Kolko serves on the advisory boards of Thea Laboratories, Santen and Abbvie.She delivers talks for the same industries.Finally, she is a consultant for Thea Laboratories.
Tor Paaske Utheim is co-founder and co-owner of The Norwegian dry eye clinic and the Clinic of eye health, Oslo, Norway, which delivers talks for and/or receives financial support from the following: ABIGO, Alcon, Allergan, AMWO, Bausc'h&Lomb, Bayer, European school for advanced studies in ophthalmology, InnZ Medical, Medilens Nordic, Medistim, Novartis, Santen, Specsavers, Shire Pharmaceuticals and Thea Laboratories.He has served on the global scientific advisory board for Novartis and Alcon as well as the European advisory board for Shire Pharmaceuticals.Utheim is the Norwegian Global Ambassador for Tear Film and Ocular Surface Society (TFOS), a Board Member of the International Ocular Surface Society, an International Member of the Japanese Lid and Meibomian gland working group (LIME), a Consultant at the Norwegian Association for the Blind and Partially Sighted, the President of the Oslo Society of ophthalmology, and the Editor-in-Chief of Oftalmolog, an eye journal distributed to all eye doctors in the Nordic region since 1980.Besides publishing articles of presumed interest to our readers, Oftalmolog publishes advertisements from pharmaceutical companies, companies selling ophthalmological equipment, and associations organizing conferences and events in ophthalmology.For more information, visit: oftalmolog.com.

Fig. 1 .
Fig. 1.The vicious circle of dry eye disease can be propagated by multiple interconnected mechanisms that collectively promote dry eye disease and may be hard to break.Benzalkonium chloride could potentially initiate the vicious circle by causing tear film instability, inflammation, damage to the ocular surface and goblet cell loss.Illustration by Sara Nøland.

Table 1
Alterations of clinical parameters among patients treated with preserved glaucoma medications.

Table 2
Alterations of clinical parameters among patients treated with unpreserved glaucoma medications.