Association between the use of prostaglandin analogues and ocular surface disease: a systematic review

Abstract

Patients with glaucoma often experience chronic ocular surface diseases, potentially underestimated in frequency and severity. To provide updated estimates of ocular surface diseases linked to prostaglandin analogue antiglaucoma eye medication, a systematic review was conducted. Twenty-seven publications were selected from databases like PubMed, Scopus, and Web of Science, following a search strategy targeting glaucoma and prostaglandins while excluding certain medications ‘(Glaucoma AND prostaglandins OR ‘prostaglandin analogues’)(‘eye drops’ OR ‘artificial tears’ OR ‘ocular surface’ OR ‘dry eye’ OR ‘dry eye syndrome’ OR ‘ocular surface disease’ OR ‘tear film’) NOT (‘beta blockers’ OR ‘alpha adrenergic agonists’ OR ‘carbonic anhydrase inhibitors’ OR ‘rho-quinase’)’. The review revealed a correlation between prostaglandin analogue use and ocular surface damage, assessing parameters such as tear break-up time, Schirmer test value, ocular surface staining, hyperaemia score, and meibomian gland characteristics. Some studies explored switching patients to alternative glaucoma medications, noting varied effects on ocular surface parameters. Comparisons suggested better tolerance and outcomes with preservative-free options over prostaglandins. Additionally, the impact of treatment duration and diquafosol on ocular health, including meibomian gland loss, was examined across different formulations. Although a link between prostaglandin analogues (with or without preservatives) and ocular surface damage was established, inconsistencies in methodologies and assessment across studies were noted. This comprehensive review, spanning a decade of glaucoma research, underscores the need for re-evaluation of treatment strategies in ophthalmology. It stresses the significance of informed decision-making for enhanced glaucoma care, taking into account the observed effects of various medications on eye health.

Introduction

Glaucoma, a leading cause of irreversible blindness worldwide, is not only a chronic and progressive optic neuropathy that damages the optic nerve head and retinal nerve fibre layer but also significantly impacts the ocular surface. Addressing the intertwined challenges of glaucoma and ocular surface diseases is crucial, as these conditions can exacerbate vision loss, affecting both peripheral and central vision, and complicate treatment adherence [1]. If left untreated, it can result in irreversible blindness [1, 2]. Glaucoma is estimated to affect over 76 million individuals, with an anticipated increase of up to 112 million cases by 2040 [3]. Therefore, this disease is a significant contributor to irreversible blindness worldwide [4].

Medication for glaucoma reduces intraocular pressure by either reducing aqueous humour production or increasing its outflow from the eye. Prostaglandin analogues increase intraocular fluid drainage and are the most commonly used first-line medication due to their efficacy in lowering intraocular pressure, once-a-day administration, and relatively mild adverse effect profile, which includes changes in iris colour, eyelash growth, and hyperaemia [5–8].

Ocular surface pathology and glaucoma often coexist, leading to moderate-to-severe signs and symptoms. This association frequently complicates the treatment of each condition and poses a challenge for medication adherence in both patients and ophthalmologists [9]. Ocular surface pathology in patients with glaucoma is believed to be due to the cytotoxicity of active substances and preservatives, for example, benzalkonium chloride (BAK), in the medication or direct damage to the corneal and conjunctival epithelium [10]. In these patients, findings may include medication allergies, corneal ulceration, punctate keratitis, tear film instability, cicatrising conjunctivitis, and dysfunction of the meibomian gland (MG). In severe cases, limbal stem cell deficiency may develop [10].

The elevated occurrence of dry eye, alongside the degradation of diverse ocular parameters and MGs in patients with glaucoma, is linked to the application of topical hypotensive treatment [11–14]. There is, therefore, evidence that patients with glaucoma are at risk of foreseeable chronic conditions affecting the ocular surface, the frequency and severity of which are quite understated. The medical requirement for the topical treatment of a condition that conceivably leads to blindness often makes the ocular surface a secondary concern. We, therefore, aimed to conduct a systematic review to ascertain the relationship between treatment with prostaglandin analogues and alterations in the ocular surface caused by their use. Additionally, we carefully examined the risk of bias and certainty of publication in all included studies.

Materials and methods

This systematic review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [15], which are designed to evaluate the impact of health interventions, regardless of the study design. PRISMA provides a checklist to ensure comprehensive reporting of systematic reviews [15]. The Cochrane database search yielded no results.

Search strategy

We conducted searches in PubMed (226 articles), Web of Science (80 articles), and Scopus (89 articles) databases from January 2023 to December 2023 and included all studies published over the last 10 years. The terms included in the search strategy were ‘(Glaucoma AND prostaglandins OR ‘prostaglandin analogues’)(‘eye drops’ OR ‘artificial tears’ OR ‘ocular surface’ OR ‘dry eye’ OR ‘ocular surface’ OR ‘dry eye syndrome’ OR ‘ocular surface disease’ OR ‘tear film’) NOT (‘beta blockers’ OR ‘alpha adrenergic agonists’ OR ‘carbonic anhydrase inhibitors’ OR ‘rho-quinase’)’.

Inclusion and exclusion criteria

Studies that assessed parameters related to the ocular surface in patients treated with prostaglandin analogues were included in the analysis. We included the following studies: (1) those with humans; (2) case reports; (3) case series; (4) cohort, cross-sectional, and case–control studies; and (5) randomised clinical trials.

Articles were excluded if they lacked data reporting ocular surface parameters, whether objective or subjective; the study patients did not receive prostaglandin analogue monotherapy; the article was a letter, conference abstract, study protocol, literary review, systematic review, or meta-analysis; the trial was performed in vitro or in animals; and the text was not in English or Spanish. If a thorough examination of the full text resulted in the determination that the article did not analyse the impact of prostaglandin analogues on the ocular surface, it was excluded.

Quality of articles, levels of evidence, and data extraction

Two authors, RMC and MCSG, independently conducted article grading and data extraction. To assess article quality, two reviewers with sufficient reliability (RMC and MCSG) worked independently in a blinded manner to construct a summary table (Table 1) using the National Heart, Lung, and Blood Institute Study Quality Assessment Tools. This instrument evaluates the quality of the selected articles in a systematic review and guarantees adequate representation of the elements in case–control, cohort, and randomised controlled trials, as well as in cross-sectional studies.

Table 1.

Quality appraisal of articles.

Author and year	Yes/Total
Rossi et al. [32]	8/14
Lopes et al. [33]	9/14
Walimbe et al. [34]	6/14
Moussa et al. [35]	8/14
Rahmatnejad et al. [36]	8/14
Wong et al. [37]	9/14
Diagourtas et al. [38]	5/14
Kumar et al. [20]	9/14
Zaleska-Żmijewska et al. [16]	7/14
Misiuk-Hojlo et al. [17]	9/14
Lopes et al. [18]	9/14
Kumar et al. [19]	8/14
El Ameen et al. [22]	5/14
Ha et al. [23]	11/14
Lee et al. (2019) [24]	6/14
Sedlak et al. [25]	6/14
Kumar et al. [21]	8/14
Ruangvaravate et al. [26]	8/14
Guo et al. [27]	10/14
Erb et al. [28]	5/14
Hagras et al. [29]	10/14
Seong et al. [30]	5/14
Muz et al. [31]	10/14
Wu et al. [41]	7/14
Kim et al. [39]	6/14
Kim et al. [40]	6/14

We examined the following aspects of the studies: appropriateness of the methodological design; effectiveness of information gathering during recruitment; adequacy and representativeness of participant descriptions; strength and reliability of the research, including the management and mitigation of bias; thoroughness of the data analysis, including qualitative analysis where applicable; control of confounding variables; and clear exploration of the implications arising from the findings. The included articles’ quality was categorised into three equally weighted outcome levels: low (yes = 0–4), moderate (yes = 5–9), and high (yes = 10–14) for observational and cross-sectional studies and low (yes = 0–4), moderate (yes = 5–9), and high (yes = 10–14) for case–control and controlled intervention studies. According to this classification, we observed that all observational and cross-sectional studies [16–38] were of moderate-to-high quality, and the three controlled studies were of moderate quality [39–41].

Regarding observational and cross-sectional studies, we found that four studies [23, 27, 29, 31] had high quality and the remaining [16–22, 24–26, 28, 30, 32–38] had moderate quality. Table 1 illustrates the agreement ratings derived from the National Heart, Lung, and Blood Institute study quality assessment tools.

Results

The systematic review’s selection process is outlined in the Fig. 1 flowchart. A total of 395 articles were initially identified. Following duplicate removal, 287 articles underwent title and abstract scrutiny by two authors, resulting in the exclusion of 255 articles. The full texts of 32 articles were examined, and five were excluded based on relevant criteria. Ultimately, 27 articles were incorporated into this review.

Fig. 1.

Flowchart of the study selection process.

Characteristics of the studies

A total of 3769 patients using prostaglandin analogues were evaluated for ocular surface findings or symptoms (Table 2). The review suggested that there is a relationship between potential damage to the ocular surface and use of the relevant medication. The mean follow-up duration in these studies was 19.74 weeks. All the included studies evaluated patients with ocular hypertension or glaucoma who were already being treated or naïve patients.

Table 2.

Study characteristics.

Author (Year)	Study design	CoI (Yes/No)	Inclusion criteria	N (patients)	Follow-up	N (eyes)	Medication(s)	Time using PGA
Rossi et al. [32]	Observational study	No	OAG or OH; ≥18 years	44	6 months	NM	PC (BAK) LT switched to PC (PQ) TV	NM
Lopes et al. [33]	Multicentre, open-label, single-arm study	No	OAG and OH	173	12 weeks	NM	PC-LT switched to PC (PQ) TV	PC-LT ≥ 4 weeks
Walimbe et al. [34]	Prospective, open-label, single-arm, multicentre study	No	POAG or OHT; ≥18 years; TBUT < 6 s	25	28 and 56 days	40	PC-LT switched to PF-LT	>12 months
Moussa et al. [35]	Single-centre, prospective, open-label study	No	Newly diagnosed POAG; ≥18 years	32	1, 3, 6 months	NM	PC-BM vs PC-LT vs PQ-TV vs PF-TF	Naïve
Rahmatnejad et al. [36]	Prospective, single-centre, open-label, nonrandomised cohort study	No	GC; ≥21 years	55	1, 2 months	110	PC (BAK) LT vs PC (SofZia) TV	Naïve (group 1) or PC (BAK) LT users ≥1 month (group 2)
Wong et al. [37]	Prospective, single-centre study	No	POAG or OH; NEI score >1	51	1, 3 months	NM	PC-LT switched to PC-TF	PC-LT users > 3 months
Diagourtas et al. [38]	Single-centre, prospective, randomised, comparative study	No	OAG or OH; elevated IOP; signs or symptoms of OSD	60	16 weeks	NM	PC-LT (Xalatan®) vs PC-LT (Lataz®) vs PC-LT (Xalaprost®)	Naïve
Kumar et al. [20]	Prospective study	No	POAG; ≥18 years	46	1, 14, 30, 42 and 84 days	23	PC-LT vs PF-LT	NM
Lee et al. [42]	Cross-sectional study	No	NTG	85	One visit	85	PGA (PC-TF, PC-TF, PC-BM) vs CG	PGA > 1 year
Zaleska-Żmijewska et al. [16]	Prospective, unblinded, and single-centre	No	POAG	90	One study visit	180	PF-TF vs PC-LT vs CG	>1 year
Misiuk-Hojlo et al. [17]	Prospective, longitudinal, open-label, multi-centre study	No	POAG or OH; ≥18 years; Stable VF; CCT 500–580 micrometres	140	15, 45, 90 days	NM	PC-LT switched to PF-LT	≥3 months
Lopes et al. [19]	Prospective observational study	No	≥18 years; DED	11	6 weeks	22	PC-BM, PC-LT, PC-TV switched to PF-TF	≥6 months
Kumar et al. [19]	Prospective observational study	No	POAG; ≥40 years	110	12 months	NM	PC (BAK) TV vs PC (PQ) TV vs CG	≥6 months
El Ameen et al. [22]	Observational, cross-sectional, clinical study	No	OAG or OH; ≥18 years; PGA: PC-LT, PC-TV, PC-BM, PF-LT	82	6 months	161	PF-LT vs PC-LT, PC-TV, PC-BM	≥6 months
Ha et al. [23]	Retrospective	No	OAG	80	12 months	NM	PC-PGA (PC-LT, PC-TF) vs PF-PGA (PF-LT, PF-TF) vs CG	1 year
Lee et al. (2019)	Cross-sectional	No	GC; Naïve; PC-TF; or PF-TF users	152	6, 24 months	152	PF-TF for 6 months vs PC-TF for 6 months vs PF-TF for 24 months vs PC-TF for 24 months vs CG	6 or 24 months
Sedlak et al. [25]	Cross-sectional	No	18–70 years	108	One visit	216	PF-LT vs PC-LT vs PF-TF vs PC-BM vs CG	6–12 months
Kumar et al. [21]	Research study	No	POAG; >40 years; Mild-moderate glaucoma; IOP controlled for 3 months	110	One visit	NA	PC-TV vs PF-TV vs CG	≤6 months
Ruangvaravat et al. [26]	Prospective, randomised, investigator-masked, single-blinded, open-label study	No	POAG; PGA user except TF; 18–80 years; OSD; TBUT < 10 s; CFS grade >1; No other types of drops used <3 months	30	6, 12, 24 weeks	60	PC-LT, PC-BM, and PC-TV switched to PC-TF and PF-TF	≥6 months
Guo et al. [27]	Prospective study	No	≥18 years; Newly diagnosed NTG; No MGD-related signs	46	1, 3, 6, 9, 12 months	84	PC-PGA vs PF-PGA vs PC-PGA + 3% DQ	≤2 weeks
Erb et al. [28]	Multicentre, transverse, epidemiological survey	No	GC or OH	1 872	One visit	NA	PF-LT vs PC-PGA, PF-PGA	Naïve or PF LT ≥ 3 months
Hagras et al. [29]	Prospective, randomised, crossover study	No	POAG; >40 years; IOP 21–32 mmHg	30	5 months	30	PF-TF vs PF-LT	NM
Seong et al. [30]	Prospective, open-label, observational study	No	POAG; 19–80 years; IOP stable; Stable VF; CCT 500–580 μm	27	45, 90 days	54	PC-PGA switched to PF-LT	≥6 months
Muz et al. [31]	Prospective study	No	POAG or OH	44	1, 3, 6, 12 months	44	PC (PQ) TV vs PC (BAK) LT	Naïve
Wu et al. [41]	Prospective, randomised, controlled study	No	GC or OH; ≥20 years	84	1, 4 months	150	PC-LT vs PF-BM	Naïve
Kim et al. [39]	Parallel-grouped, investigator-blind, active-control, randomised, multicentre clinical study	No	POAG or OH; ≥19 years; IOP of 15–40 mmHg after washout	51	1, 3 months	NM	PC-LT vs PF-LT	NM
Kim et al. [40]	Multi-centre, randomised, investigator-blind, active controlled, parallel-group, clinical trial	No	OAG or OHT; ≥19 years; IOP of 15–35 mm Hgs	131	4, 8, 12 weeks	NM	PF-TF vs PC-LT vs PF-LT	NM

BAK benzalkonium chloride, BM bimatoprost, CCT corneal central thickness, CFS corneal fluorescein staining, CG control group, DED dry eye disease, DQ diquafosol, GC glaucoma, IOP intraocular pressure, LT latanoprost, MGD meibomian gland dysfunction, NA not applicable, NM not mentioned, NTG normo tense glaucoma, OAG open angle glaucoma, OH ocular hypertension, OSD ocular surface disease, PC preservative-container, PF preservative-free, PGA prostaglandin analogue, POAG primary open angle glaucoma, PQ polyquad, TBUT tear break-up time, TF tafluprost, TV travoprost, VF visual field, NEI National Eye Institute.

The most frequently evaluated assessments in the included studies were tests for tear break-up time (TBUT), ocular surface staining, Schirmer test, and evaluation of hyperaemia scores [16–18, 20, 22–24, 26, 28–42].

Additionally, some studies focused on MG characteristics in these patients [23, 24, 27, 42]. Patient symptoms were mostly assessed through the employment of the Ocular Surface Disease Index (OSDI) questionnaire; 17 of 27 (61%) studies [16, 18, 19, 21, 23, 24, 30–32, 34–36, 38–41] included this test.

Eight of the reviewed articles, eight studies [17, 18, 26, 30, 32–34, 37] evaluated changes in the ocular surface of patients using antiglaucoma medications (whether they were prostaglandin analogues or not and with or without preservatives) by switching them to other active ingredients (prostaglandins with or without preservatives). Three studies switched patients from preserved active ingredients to preservative-free latanoprost [17, 30, 34], among which two switched patients from preservative-containing latanoprost to preservative-free latanoprost [17, 34]. Exploratory ocular tests were conducted in all three articles. As for TBUT, improvements were observed in studies by Walimbe et al. [34] and Misiuk-Hojlo et al. [17]; however, in the other study, no significant results were found [30]. The hyperaemia scores improved in the three articles, whereas no statistically significant difference was found in the study by Walimbe et al. [34] The staining values showed significant differences in the three articles. Misiuk-Hojlo et al. [17] also assessed the blepharitis score based on patient symptoms, which decreased when transitioning to preservative-free latanoprost. Six studies [19, 31–33, 35, 36] compared different preservatives: BAK, the most used preservative, and polyquad [19, 31–33, 35] or SofZia [36]. Also, one study [38] focused on the same active but different BAK concentrations to establish a relationship between the preservative and ocular surface damage.

Eleven articles [16, 20–23, 25, 27, 29, 39–41] evaluated the use of different prostaglandins and their outcomes and reported similar results. Best outcomes were observed in preservative-free latanoprost users.

Among all studies included, seven [16, 19, 21, 23–25, 42] assessed various ocular surface parameters and compared the results between prostaglandin analogue users and controls. When making this comparison, we observed that different values are affected by the medication use, such as the TBUT, Schirmer test result, and ocular surface staining values in the studies by Zaleska-Żmijewska et al. [16], Lee et al. [42], Lee et al. [24], and Ha et al. [23], However, one study [23] reported no significant difference between PC PGA (preservative container prostaglandin analogues), PF PGA (preservative free prostaglandin analogues), and control group. However, preservative-free medications resulted better tolerated by patients. Sedlak et al. [25] evaluated different parameters for corneal toxicity in patients using different active substances, with or without preservatives, and compared them to the control group.

One study [24] compared the same active ingredient, tafluprost, with and without preservatives, in patients who had been using the medication for 6 or 24 months, providing information about how the duration of treatment can affect the ocular surface.

Kim et al. [39] compared the use of latanoprost with and without preservatives and found better scores for preservative-free status only when assessing hyperaemia scores.

Guo et al. [26] observed the rate of MG loss after 12 months in patients using preservatives, preservative-free medication, and preservatives and diquafosol 3% (DQ) and reported that DQ is a good aid in maintaining MG status. The results of the assessments of all articles are presented in Tables 3 and 4.

Table 3.

Ocular surface assessments and outcomes.

Author (Year)	Evaluation on ocular surface performed	Outcomes
Rossi et al. [32]	TBUT	The median value changed from 8 s at baseline to 10 s at the 6-month visit (P = 0.0001).
	Hyperaemia score	No significant difference was noted between BAK and PQ.
	OSS	Punctate keratitis was observed in 70.5% of patients at baseline, and it decreased to 29.5% after 6 months (P < 0.001).
Lopes et al. [33]	Hyperaemia score	Hyperaemia score decreased from 0.94 to 0.74.
Walimbe et al. [34]	TBUT	The TBUT significantly improved from 3.67 ± 1.60 s at baseline to 6.06 ± 3.39 s after 56 days of PF-LT treatment (P = 0.0001).
	Hyperaemia score	The mean value was 0.48 ± 0.75 at baseline for PC-LT and decreased to 0.35 ± 0.48 with PF; this decrease was not statistically significant.
	Inferior corneal staining	The values for corneal staining significantly decreased from 0.85 ± 0.69 to 0.53 ± 0.60 after 56 days with PF-LT (P = 0.0033).
Moussa et al. [35]	Hyperaemia score	Signs were found in all four (PC-BM, PC-LT, PQ-TV, and PF-TF) groups, although no significant difference was found between them.
	OSS (Oxford scale)	Signs were found in all four (PC-BM, PC-LT, PQ-TV, and PF-TF) groups, although no significant difference was found between them.
Rahmatnejad et al. [36]	Schirmer test	Better results were found for PC (SofZia) TV, although there was no significant difference between groups.
	TBUT	During follow-up, patients already treated showed worse results (P = 0.002), although no significant difference was observed at the 2-month follow-up.
	Hyperaemia score	No significant difference was observed between eyes (BAK-preserved LT and SofZia-preserved TV) or between groups 1 and 2.
	OSS	No significant difference was found within groups for conjunctival or corneal staining.
Wong et al. [37]	TBUT	Significant increase was observed after 1 (P = 0.003) and 3 (P = 0.017) months.
	Hyperaemia score	Substantial reductions in hyperaemia score for both bulbar conjunctiva (P = 0.028) and palpebral conjunctiva (P = 0.001) were noted.
	OSS	All individual areas had a significant decrease compared to baseline (P < 0.05).
Diagourtas et al. [38]	TBUT	The overall mean value decreased between baseline and final visits. No statistically significant difference was found among formulations.
Kumar et al. [20]	TBUT	The TBUT decreased during follow-up for both groups. Lower values were observed for the PC group (P < 0.043).
	Hyperaemia score	After 2 weeks of treatment, significant values (P = 0.025) were found between both (PC-LT vs PF-LT) groups. No significant difference was found after the 12-week follow-up.
Lee et al. [42]	Schirmer test	Significant difference was found between PGA users and those in the control group (P < 0.001). Better results were obtained in the control group.
	TBUT	Significant difference was found between PGA users and those in the control group (P < 0.001). Better results were obtained in the control group.
	KEP	Significant difference was found between PGA users and those in the control group (P = 0.015). Better results were obtained in the control group.
	MG parameters	All parameters showed significant difference (P < 0.001). Better results were obtained in the control group.
Zaleska-Żmijewska et al. [16]	Schirmer test	Schirmer test showed a significant difference between groups: preservative-containing medication (8.27 mm/5 min) and preservative-free medication (12.07 mm/5 min), and control (11.63 mm/5 min) (P < 0.05).
	TBUT	The TBUT test results correlate with MMP-9 measures. Fifty percent of participants using BAK-containing medication and 10% of those receiving preservative-free treatment and those in the control group achieved a TBUT of <5 s (P < 0.05).
	OSS (Oxford scale)	Increased corneal and conjunctival staining in the BAK-containing group compared with untreated or preservative-free users
Misiuk-Hojlo et al. [17]	TBUT	The TBUT significantly improved compared with that at baseline in 23.4% of patients at day 45 (D45) and in 30.7% of patients at day 90 (D90) (P = 0.0023 and P < 0.0001, respectively).
	Hyperaemia (McMonnies scale)	Conjunctival hyperaemia decreased progressively, with 1.6% of patients having it at D90 (P < 0.0001).
	OSS (Oxford scale)	Conjunctival fluorescein staining showed marked improvement, with 3.2% of patients showing it at D90 (P < 0.0001). Corneal fluorescein staining significantly improved, with no staining in 90.3% of patients at D90 (P = 0.0001).
	Blepharitis	Blepharitis prevalence decreased significantly, with 0.8% of patients showing it at D90 (P < 0.0001).
Lopes et al. [18]	Schirmer test	Schirmer test mean score of 5.09 ± 2.75 mm decreased to 4.36 ± 2.40 mm (P = 0.198, not statistically significant).
	BUT (DEWS)	TF-BUT mean score of 6.68 ± 2.07 s decreased to 5.5 ± 2.15 s (P = 0.113, not statistically significant).
	GL test (Bijsterveld’s scale)	GL test (Bijsterveld index) mean score of 6.27 ± 2.72 decreased to 3.04 ± 1.25 (P < 0.001, statistically significant improvement).
Kumar et al. [19]	Not performed
El Ameen et al. [22]	BUT (K5M)	The text does not provide specific numerical values or P values for break-up time (BUT). Therefore, it is not possible to determine the exact changes in BUT or whether there were statistically significant differences among the treatment groups based on the information provided.
	Conjunctival hyperaemia: limbal+bulbar (K5M)	Bulbar+limbal hyperaemia was significantly different when results from PC users when compared with those from PF-LT users.
	TMH (K5M)	The text mentions that there were no relevant between-group differences in the mean TMH. This suggests that the different treatments did not significantly affect tear meniscus height.
Ha et al. [23]	Schirmer test	No significant difference was observed between groups (PC-PGA, PF-PGA, and control)
	TBUT	Significant differences were found when comparing PC-PGA vs PF-PGA (P = 0.012) and PC-PGA vs control (P = 0.001).
	OSS (Oxford scale)	All groups showed significant differences between them: PC-PGA vs PF-PGA (P = 0.021); PC-PGA vs control (P = 0.001), and PF-PGA vs control (P = 0.001).
	Meibomian gland dropout (K5M)	No significant difference in MG dropout between groups.
	Meibum score	All groups showed significant differences between them: PC-PGA vs PF-PGA (P = 0.007); PC-PGA vs control (P = 0.001); and PF-PGA vs control (P = 0.031).
	Meibo-score	All groups showed significant differences between them: PC-PGA vs PF-PGA (P = 0.035); PC-PGA vs control (P = 0.001); and PF-PGA vs control (P = 0.027).
	Lid margin abnormality	Significant difference was found in PC-PGA vs PF-PGA and PC-PGA vs control (P = 0.015 and P = 0.021, respectively). No significant difference was found between PF-PGA and control.
Lee et al. (2019)	Schirmer test	No statistically significant difference between the four TF groups.
	TBUT	Significant differences between 24-month PF-TF and 24-month PC-TF users (P = 0.022)
	OSS (Oxford scale)	No statistically significant differences between the four TF groups
	MG dropout rate (LipiView II)	Significant differences between 6-month PF-TF and 24-month PC-TF and 24-month PF-TF users. There were also significant differences between 6-month PC-TF and 24-month PC-TF and 24-month PF-TF users. There were significant differences between 24-month PF-TF and 24-month PC-TF users.
	Lid margin abnormality	Significant differences between the 6-month PF-TF and 24-month PC-TF groups (P = 0.048)
	Meibum quality	No statistically significant difference between the four TF groups
	Meibum expressibility	Significant differences between the 6-month PF-TF and 24-month PC-TF groups (P = 0.022)
	Meiboscale	Significant differences between the 6-month PF-TF and 24-month PF-TF and 24-month PC-TF groups. There were significant differences between the 6-month PC-TF and 24-month PC-TF and 24-month PF-TF groups. There were significant differences between the 24-month PF-TF and 24-month PC-TF groups.
Sedlak et al. [25]	TP concentration	The TP concentration was higher in PC-LT, but only significant when compared with the control group (P < 0.001). The PF-LT group had a significantly higher TP level than the control group (P < 0.01). The TP value for the PC-BM group was significantly higher compared with the control group (P < 0.001), although no significant differences were noted when compared with the PF-LT and PC-LT groups.
	Activity of superoxide dismutase	No significant differences between the control, PF-LT, and PF-TF groups. In the groups using PC medication, the results were significantly higher than in the control group (P < 0.001).
	Catalase	No significant differences between the control, PF-LT, and PF-TF groups. In the groups using PC medication, the results were significantly higher than in the control group (PC-LT, P < 0.5; PC-BM, P < 0.1).
	Glutathione peroxidase	There were statistically significant differences between the preservative-containing and preservative-free groups. Any of the groups showed significant differences in glutathione peroxidase activity when compared with the control group.
	Advanced oxidation protein products content	All the study groups showed significantly higher values than the control group (PF-LT, P < 0.05; PC-LT, P < 0.01; PF-TF, P < 0.01; PC-BM, P < 0.01). No significant difference between the medication groups.
	Total sulfhydryl	All the study groups showed significantly lower results than the control group (PF-LT, P < 0.05; PC-LT, P < 0.01; PF-TF, P < 0.05; PC-BM, P < 0.05). No significant differences between the medication groups.
	Total oxidant status	When compared with control, all groups showed higher results (PF-LT, P < 0.01; PC-LT, P < 0.001; PF-TF, P < 0.001; PC-BM, P < 0.001).
	Total antioxidant response	No significantly different results were found between the study and control groups.
	Oxidative stress index	Only patients using PF-LT obtained significantly higher results when compared with controls; for the rest, the results were as follows: PC-LT, P < 0.001; PF-TF, P < 0.01; PC-BM, P < 0.05.
Kumar et al. [21]	Not performed
Ruangvaravate et al. [26]	Schirmer test	The test results showed minimal improvement in the PF-TF group, with no statistically significant difference from baseline. There was also no statistically significant difference in clinical sign test results between the PC-TF and PF-TF groups.
	TBUT	Significant increase in both groups at week 24 (PF-TF, P = 0.002; PC-TF, P = 0.004)
	OSS (Oxford score)	No significant difference was found when compared with baseline values or when compared between groups
	Conjunctival hyperaemia	No significant difference was found when compared with baseline values or when compared between groups
	Lid inflammation	No significant difference was found when compared with baseline values or when compared between groups
Guo et al. [27]	Meibomian gland dropout (K5M)	MGL increased significantly at 9 and 12 months in the PC-PGA group (P < 0.001). No significant differences in MGL were found in the PF-PGA and PC-PGA + DQS groups at 9 or 12 months.
Erb et al. [28]	TBUT	Lower than 5 s for 10.8% and lower than 10 s for 52.3% of the eyes
	OSS	No statistical difference between the individual therapy groups. The results were slightly better in treatment-naïve patients.
	Conjunctival hyperaemia	No statistical difference between the individual therapy groups. The results were slightly better in treatment-naïve patients.
	OSD	Lid redness was the most common ocular sign among patients with OSD. The different ocular surface parameters showed no statistically significant difference between patients.
Hagras et al. [29]	TBUT	The TBUT significantly worsened after PC-LT at 2 and 5 months (P < 0.001 and P = 0.026, respectively), but remained nonsignificant after PF-TF treatment.
	Conjunctival hyperaemia	Significantly higher in PC-LT-treated eyes
	Corneal erosion	Significantly higher in PC-LT-treated eyes
Seong et al. [30]	TBUT	Results when switching to PF-LT after 90 days were not significant (P = 0.369).
	OSS	Significant differences were found in corneal and conjunctival staining after 90 days, with an improvement of P < 0.001 and P = 0.02, respectively.
	Hyperaemia: Bulbar injection+Limbal injection	Results for hyperaemia improved with both bulbar and limbal injections, although only bulbar injection obtained significant results (P = 0.025).
Muz et al. [31]	Schirmer test	The results were better for PQ-preserved TV treatment with significant differences compared to BAK-preserved LT at baseline (P = 0.014) and also after 12 months (P = 0.015).
	TBUT	The results were better for PQ-preserved TV treatment compared to BAK-preserved LT, but they were not significant after 12 months (P = 0.475).
	OSS (Oxford scale)	The results were similar with both medications, and the results were not significant.
Wu et al. [41]	Schirmer test	After 4 months, the results between the PF-BM and PC-LT groups were not clinically significant (P = 0.0529).
	NIKBUT (K5M)	Comparing the average NIKBUT after 4 months, the results were not clinically significant (P = 0.7749).
	OSS (Oxford scale)	After 4 months, the results between the PF-BM and PC-LT groups were not clinically significant (P = 0.5897).
	Conjunctival redness or hyperaemia score (R score) (K5M)	After 4 months, the results between the PF-BM and PC-LT groups for limbal temporal and bulbar temporal hyperaemia were not clinically significant (P = 0.1751 and P = 0.3309, respectively). Limbal nasal and bulbar nasal hyperaemia value compared between the PF-BM and PC-LT groups was clinically significant, with better results for the PF-BM group (P = 0.0035 and P = 0.0068, respectively).
	TMH (K5M)	The results were not significant between the PF-BM and PC-LT groups after 4 months (P = 0.4663).
Kim et al. [39]	TBUT	The values decreased in both groups during the visits, although the results between them were not clinically significant (P = 0.458 at visit 4).
	OSS (Oxford scale)	The values were similar in both groups at visit 3. No clinically significant results were obtained when the groups were compared (P = 0.399 for corneal staining; P = 0.292 for conjunctival staining).
	Hyperaemia value (Efron scale)	Only the bulbar hyperaemia value of the PF group at 12 weeks was significantly lower than that of the PC group (P = 0.037).
Kim et al. [40]	TBUT	No significant difference between the PF-LT, PC-LT, and PF-TF groups at study completion.
	OSS	There were significant differences with lower results in the PF-LT group alone. The results in the PF-TF and PC-LT groups remain the same, with no significant changes.
	Hyperaemia score	No significant differences between the PF-LT, PC-LT, and PF-TF groups at study completion.

ANOVA analysis of variance, BAK benzalkonium chloride, BM bimatoprost, BUT break up time, DEWS dry eye workshop, GL green lisamine, K5M keratograph 5M, LT latanoprost, MG meibomian gland, MGL meibomian gland loss, NIKBUT not invasive break up time, OSD ocular surface disease, OSS ocular surface staining, PC preservative-container, PGA prostaglandin analogue, PF preservative-free, PQ polyquad, TBUT tear break up time, TMH tear meniscus height, TV travoprost, TF tafluprost, TP total protein, TF-BUT tear film break-up time, KEP keratoepitheliopathy.

Table 4.

Symptoms reported on patients and outcomes.

Author (Year)	Assessments	Outcomes
Rossi et al. [32]	OSDI	OSDI mean change was −7 points (P = 0.0027).
Lopes et al. [33]	Ocular discomfort score	Average score for ocular discomfort was low (1.83). Most patients preferred the new study medication (81.5%) over the previous one.
Walimbe et al. [34]	OSDI	Baseline score was 18.09 ± 18.61. On day 56, it decreased to 7.06 ± 10.75 (P < 0.0001).
Moussa et al. [35]	OSDI	PC (PQ) TV demonstrated the highest tolerance. Results between the other (PC-BM, PC-LT, and PF-TF) groups had no statistical relevance.
Rahmetnejad et al. [36]	OSDI	No significant difference between or within groups during the 2-month follow-up.
Wong et al. [37]	OSD symptoms	Foreign body sensation, irritation, itching, and photophobia showed significant changes.
Diagourtas et al. [38]	OSDI	The worst results were found in Xalaprost users. The most well-tolerated was Xalatan, with a mean score of 2.
Kumar et al. [20]	Not performed
Lee et al. [42]	OSDI	No significant difference was found between those in the control group and PGA users.
Zaleska-Żmijewska et al. [16]	McMonnies scale (reported by patient)	Seventy percent of individuals in the control group and 43.3% in the preservative-free group reported Grade 0 scores on the McMonnies scale, indicating no ocular irritation. In contrast, only 26.7% of individuals in the PC treatment group scored Grade 0
	OSDI	Individuals using preservative-free treatments rated their condition as having the lowest impact on their quality of life. However, these differences were not significant, as indicated by Levene’s test (P = 0.218) and ANOVA (P = 0.374).
Misiuk-Hojlo et al. [17]	Subjective symptoms evaluation	Patient subjective assessments of tolerability improved significantly, with the mean score decreasing from 5.3 (SD = 2.2) at baseline to 1.9 (SD = 1.7) at day 90 (P < 0.0001).
	VAS scale	The mean VAS score decreased from 5.3 (SD = 2.2) at baseline (D0) to 1.9 (SD = 1.7) at day 90 (P < 0.0001). A lower VAS score indicates better tolerability.
Lopes et al. [18]	OSDI	OSDI questionnaire mean score: 35.27 ± 10.67. The values after 6 weeks are not provided, although they decreased, indicating an improvement in symptoms (P < 0.0001, statistically significant).
Kumar et al. [19]	OSDI	Significant differences were observed in the OSDI scores between the BAK and PQ groups (16.63, P < 0.05). The mean difference in the OSDI scores between the BAK and control groups was also significant (18.96, P < 0.05). However, the OSDI scores between the PQ and control groups were similar (P > 0.05).
El Ameen et al. [22]	Not performed
Ha et al. [23]	OSDI	Significant values were found when comparing PC-PGA vs PF-PGA (P = 0.003) and PC-PGA vs control (P = 0.001).
Lee et al. (2019)	OSDI	No significant difference between the four studied groups
Kumar et al. [21]	OSDI	The mean difference between the PF and PC groups was significant (P < 0.01). The difference between the control and PF groups was not significant.
	GQL-15	The mean difference between the PF and PC groups was significant (P = 0.047). The difference between the control and PF groups was not significant.
Ruangvaravate et al. [26]	OSD symptoms (itching, burning, redness, tearing)	Over the course of the study, all subjective OSD symptoms improved in both groups, with statistically significant improvements observed for burning and redness symptoms. Other symptoms also improved, although the differences between weeks 0 and 24 and between groups were not statistically significant.
Guo et al. [27]	Not performed
Erb et al. [28]	VAS scale	Patients were significantly (P < 0.0001) more satisfied with the tolerability to PF-LT than with their previous treatment.
Hagras et al. [29]	SPEED questionnaire	Patients reported eye discomfort in the beginning, especially hyperaemia, which improved later. Significant improvement when switching back from PC-LT treatment to PF-TF treatment.
Seong et al. [30]	OSDI	The results after 90 days of treatment improved significantly from a baseline value of 26.4 ± 18.5 to 15.7 ± 15.6 (P < 0.001).
Muz et al. [31]	OSDI	Patients reported similar results with both medications, and the results were not clinically significant.
Wu et al. [41]	OSDI	The results were better for PF-BM users, although the values between groups were not significant after 4 months (P = 0.1575).
Kim et al. [39]	OSDI	No clinically significant results were obtained when the groups were compared at visit 4 (P = 0.282).
Kim et al. [40]	OSDI	The PF-LT group showed significant decreases in ‘stinging/burning’, ‘dryness’, ‘light sensitivity’, and ‘pain or soreness’ symptoms at various visits compared with the PC-LT and PF-TF groups.

BAK benzalkonium chloride, BM bimatoprost, GQL-15 glaucoma quality of life-15, LT latanoprost, OSD ocular surface disease, OSDI ocular surface disease index, PC preservative container, PF preservative-free, PGA prostaglandin analogue, PQ polyquad, TF tafluprost, VAS visual analogue scale.

Discussion

This systematic review evaluated studies on ocular surface disease in patients using prostaglandin analogues, potentially related to dry eye symptoms.

Preservatives are commonly used in eye medications because of their convenience, regulatory obligations, and comparatively higher expenses associated with alternative options [43]. BAK has been a fundamental component of ophthalmology since the 1940s and remains the predominant preservative used in eye drop treatment [43]. Furthermore, BAK has adverse effects on various eye components such as the conjunctival tissue, corneal epithelium, trabecular meshwork, and lens epithelium [44, 45].

Five articles [17, 18, 26, 30, 34] in our systematic review assessed tolerability and changes in the ocular surface after switching from other preservative-containing medications to preservative-free medications. The minimum follow-up for these five studies was 42 days [18]. These studies reported that the assessments tended to improve after the switch; however, some changes were not clinically significant (Tables 3, 4). Diverse outcomes were observed among the five studies. No surface parameter was identified where changes were clinically significant across all included articles. The Schirmer test was only conducted in two of the included articles [18, 26], and no clinically significant results were found when transitioning to preservative-free formulations. Regarding patient tolerance, these studies performed different tests, which resulted in better tolerance and symptomatology after the switch. Some studies [19, 31–33, 35, 36] attempted to use different preservatives to prove whether they can be as effective as BAK without having toxic properties. Rossi et al. [32], Lopes et a. [33], Moussa et al. [35], Kumar et al. [19], and Muz et al. [30] compared polyquad and BAK. The OSDI questionnaire was administered in four of the studies [19, 31, 32, 35], and Lopes et al. [33] studied the ocular discomfort score. Most patients reported better tolerance with polyquad preservative. However, in Muz et al.’s [31] study, no significant differences were reported in the OSDI scores. Various tests conducted for hyperaemia produced values that were not clinically significant [32, 35]. Discrepant results were observed among studies for TBUT and OSS parameters. The Schirmer test was performed only in the study by Muz et al. [31], showing improved results with polyquad.

Rahmatnejad et al. [36] compared SofZia and BAK preservatives. Assessments for ocular surface parameters and patients’ symptoms did not report statistically significant difference.

Diagourtas et al. [38] studied the effect of the same active (latanoprost) from different brands (considering its compounds) on the ocular surface. The only assessment performed was TBUT, which showed no significant differences between formulations. As for patients’ symptoms assessed using the OSDI questionnaire, the brand Xalatan® was the most tolerated.

In this review, one study [27] aimed to treat patients with ophthalmic eye drops using prostaglandin analogues with the preservative DQ, which acts as an agonist of the P2Y2 receptor and stimulates the production of tears and mucin. Currently, DQ is authorised for the management of dry eyes in Japan and South Korea [46], and this study explored MG loss after a follow-up of 12 months, comparing preservative-containing, preservative-free, and preservative-containing + DQ 3% drops. This medication proved to be as safe as preservative-free medications, showing the same or better results than preservative-containing medications.

Seven studies [16, 19, 21, 23–25, 42] conducted assessments on prostaglandin users and compared the results with the ones obtained from a control group. Most assessments in the included studies were the Schirmer test, TBUT, tests for hyperaemia, and ocular surface staining. After reviewing the results of these studies, it was observed that most analysed parameters were altered with prostaglandin use compared with the control group. Worse results were typically observed when preservatives were used. Sedlak et al. [25] measured oxidative stress levels in the tear film and their relationship with the presence or absence of preservatives in prostaglandin analogue medication. Their findings revealed a clinically significant difference in total protein concentration, advanced oxidation protein product content, total sulfhydryl content, and total oxidant status in patients using the active substance with or without preservatives compared with the control group.

Preservative-containing medication showed significantly poorer values than preservative-free medications in terms of glutathione peroxidase and oxidative stress index for preservative-containing latanoprost.

Ha et al. [23] compared the preservative-containing, preservative-free, and control groups; no statistically significant differences were found in the OSDI, TBUT, and lid margin abnormality score between the preservative-free prostaglandin analogue and control groups.

Some studies [20, 22, 39–41] aimed to demonstrate better results in ocular surface disease when preservatives were absent; however, without a control group for comparison, we cannot provide a general view of the harm caused to the ocular surface by the use of preservative-free medication.

Five articles investigated hyperaemia scores, three [22, 40, 41] of which observed significantly improved scores for preservative-free medications than for preservative-containing medications; this was significant when bulbar nasal tissue was analysed. After a 12-week follow-up, Kumar et al. [20] observed no significant differences between groups for hyperaemia scores. Additionally, TBUT assessment was performed in all five articles, although only Kumar et al. [20] showed a significant difference when preservative-containing medication was used.

Only one study [24] compared different statuses of the ocular surface for different medication durations. As patients continued the medication for a longer duration, the results of some tests worsened (TBUT, MG dropout rate, lid margin abnormality, meibum expressibility, and Meiboscale score); the symptoms experienced by the patients also worsened. Additionally, some parameters deteriorated in the presence of preservatives (TBUT, MG dropout rate, and Meiboscale score).

Table 4 shows the various subjective tests (primarily questionnaires) conducted on the participating patients to assess their ocular symptom status, particularly the sensation of dry eyes and quality of life, which are crucial for successful adherence to these treatments. Among all studies included in this review, many showed better results with the use of preservative-free medication. However, there is lack of uniformity between articles and assessment outcomes to corroborate this. To the best of our knowledge, this is the first review to analyse the relationship between ocular surface disease and the use of prostaglandin analogue glaucoma medications. This systematic review on prostaglandin analogues and their impact on ocular surface damage has several limitations that need to be acknowledged. First, the dynamic nature of scientific research implies that some data and findings may be subject to future updates or revisions. Furthermore, the scope of the review was constrained by the depth of available research studies, which in turn affected the breadth of the discussion on certain aspects. Despite these limitations, this review aims to provide a nuanced understanding of the associations between prostaglandin analogues and ocular surface damage, while acknowledging existing constraints and contributing valuable insights to the current body of knowledge in this field.

This review provides valuable information for the evaluation and treatment of patients with glaucoma and ocular hypertension, as well as for understanding the current status of treatments, techniques implemented to enhance treatment adherence, and improvement of patients’ symptoms using preservative-containing medications.

Author contributions

RMC was responsible for designing the review protocol, writing the protocol and report, conducting the search, screening potentially eligible studies, extracting and analysing data, interpreting results, updating reference lists and creating the tables. MCSG was responsible for designing the review protocol, screening potentially eligible studies, writing the protocol and report, conducting the search and creating the tables. MCM contributed to extracting and analysing data, interpreting results and provided feedback on the report. RMC was tasked with formulating the review protocol, composing both the protocol and report, overseeing the search, screening potentially eligible studies, extracting and analysing data, interpreting results, updating reference lists, and generating the ‘Summary of Findings’ tables. MCSG was responsible for devising the review protocol, extracting data, screening potentially eligible studies, and drafting this report. MCM contributed to the report’s composition and provided feedback.

Funding

The authors confirm that they have no affiliations with or participation in any organisation or entity with any financial interest (including honoraria, educational grants, participation in speakers’ bureaus, membership, employment, consultancies, stock ownership, or other equity interests, as well as expert testimony or patent-licensing arrangements) or non-financial interest (such as personal or professional relationships, affiliations, knowledge, or beliefs) related to the subject matter or materials discussed in this manuscript. This study did not receive specific grants from funding agencies in the public, commercial, or non-profit sectors. Funding for the English language revision was provided by RICORS (RD21/0002/0011, funded by MCIN-Instituto de Salud Carlos III and co-funded by European Union–NextGenerationEU; Plan de Recuperación Transformación y Resiliencia). The editing of this manuscript was sponsored by RETICS OFTARED (RD16/0008/0010, funded by Instituto de Salud Carlos III and co-funded by the European Union, ERDF/ESF, ‘Investing in your future’).

Competing interests

The authors declare no competing interests.