Preservative-free 0.01% bimatoprost versus other formulations for glaucoma and OHT: Systematic review and meta-analysis of efficacy and safety

Abstract

Purpose

To demonstrate non-inferiority on efficacy and safety of preservative-free bimatoprost 0.01% versus preserved and higher-concentration bimatoprost formulations for the treatment of glaucoma or ocular hypertension (OHT), using a network meta-analysis (NMA).

Methods

A systematic review and Bayesian NMA were conducted according to a pre-registered protocol (PROSPERO: CRD420250650769) and PRISMA-NMA guidelines. Randomised controlled trials comparing monotherapies of bimatoprost (preserved or preservative-free) in adults with glaucoma or OHT were included. The primary endpoint was mean intraocular pressure (IOP) at Week 12. Conjunctival hyperaemia, were descriptively analysed. Risk of bias was assessed using the Cochrane RoB 2 tool.

Results

Four studies were included. The Bayesian NMA demonstrated that preservative-free bimatoprost 0.01% was non-inferior to all other bimatoprost formulations in lowering IOP, with all 95% credible intervals falling within the 1.5 mmHg non-inferiority margin. Sensitivity analyses using a frequentist model confirmed these findings. Safety analysis indicated that preservative-free bimatoprost formulations were associated with lower rates of hyperaemia compared to preserved versions. At Week 12, more patients had no hyperaemia with preservative-free bimatoprost 0.01% (57.3% vs. 43.6%) and fewer showed worsening scores (18.3% vs. 30.4%). Hyperaemia reported as an adverse event was also less frequent with bimatoprost 0.01% than 0.03% (28.6% vs. 37.4%).

Conclusions

Preservative-free bimatoprost 0.01% provided IOP-lowering efficacy similar to preserved or higher-dose bimatoprost, with less occurrence of conjunctival hyperaemia. Further long-term studies are needed to confirm these benefits.

Introduction

Glaucoma is a group of progressive optic neuropathies marked by the degeneration of retinal ganglion cells, ultimately resulting in vision loss and blindness. ¹ To date, elevated intraocular pressure (IOP) is the only modifiable risk factor associated with glaucoma. ¹ To manage elevated IOP, a range of topical ocular hypotensive agents are used, with prostaglandin analogs being among the most commonly prescribed first-line treatments.^1,2

Traditionally, preservatives such as benzalkonium chloride (BAK) have been added to topical antiglaucoma medications to maintain sterility. BAK may also enhance corneal penetration, associated with the hope to increase the efficacy of active pharmaceutical ingredients. ³ However, not all studies found increased penetration ⁴ and multiple clinical trials have demonstrated that preservative-free formulations offer comparable IOP-lowering efficacy with reduced ocular surface toxicity.^5–7 It is now well established that BAK can disrupt the lipid layer of the tear film, cause conjunctival inflammation, and lead to epithelial cell damage with chronic use due to its detergent properties.^8,9 In addition, it is well established that the lowest effective dose of an anti-glaucoma drug should be used providing the target therapeutic effect while minimizing adverse events. ¹⁰ Overall, the adverse effects related to BAK, other preservatives, and/or high concentration of active ingredients can reduce patient comfort, induce poor adherence, and decrease quality of life, critical concerns in the chronic management of glaucoma.^11,12

Since poor treatment adherence leads to poor IOP control,^13–15 efforts have been made to address these tolerability issues. This includes removing the preservative and lowering the concentration of active ingredients in topical therapies. Bimatoprost, a synthetic prostamide analogue related to prostaglandin F2α, is widely used for lowering IOP.^16–18 The original formulation, bimatoprost 0.03% preserved with BAK, demonstrated robust IOP-lowering efficacy (25–30% from baseline), but was frequently associated with adverse events such as conjunctival hyperaemia.^19–21 To improve tolerability without sacrificing efficacy, a series of reformulations were developed: first, a preservative-free formulation of bimatoprost 0.03%, and later, a lower concentration formulation (0.01%), but containing higher preservative concentration than preservative-free bimatoprost 0.03%. Clinical studies confirmed that the preserved 0.01% formulation maintained comparable efficacy while significantly reducing side effects.^5–7 In this context, a new preservative-free formulation of low-concentration bimatoprost 0.01% has been introduced.

Despite these developments, there remains a lack of comprehensive data directly comparing the efficacy and safety of the preservatove-free 0.01% formulation with other available bimatoprost formulations. This evidence gaps creates uncertainty for clinicians when selecting the most appropriate therapy, particularly for patients at risk of ocular surface disease.

To address this need, we performed a network meta-analysis (NMA) to evaluate whether preservative-free bimatoprost 0.01% demonstrates non-inferiority of the efficacy, and safety compared with other monotherapies of bimatoprost at different concentrations and formulations available for the treatment of glaucoma or ocular hypertension (OHT). The results of this analysis aim to provide evidence-based guidance for clinicians when considering preservative-free or low-concentration options for patients at risk of ocular surface complications.

Methods

Protocol and registration: This work was conducted according to a predefined protocol registered in PROSPERO (CRD420250650769) and following the guideline for network meta-analyses published by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA-NMA). ²²

Eligibility criteria: The review question was established using the population, intervention, comparators, and outcomes (PICO) framework ( Table 1 ). Briefly, the population included adult patients with glaucoma or OHT who were users of a bimatoprost monotherapy (preserved or preservative-free) for at least 3 months. Included studies could be randomized controlled trials, randomized parallel, or cross-over studies, in French or English, and without time restriction.

Table 1.

Eligibility criteria were based on PICO method.

Criteria	Inclusion	Exclusion
Population (P)	Patients with primary open-angle glaucoma, exfoliative glaucoma, pseudoexfoliative glaucoma, pigmentary glaucoma, primary angle-closure glaucoma, ocular hypertension. Relevant demographic factors: Adults (18 years or older) Setting: home use	Other secondary open-angle glaucoma Secondary angle-closure glaucoma, acute angle-closure glaucoma.
Intervention (I)	All monotherapies with bimatoprost (all concentrations, with or without preservative) All posology schemes according to the product leaflet Treatment duration: at least up to Week 12/Month 3	Combined therapies Implant of bimatoprost Other active treatments
Comparator (C)	All monotherapies with bimatoprost (all concentrations, with or without preservative)	Combined therapies Implant of bimatoprost Other active treatments
Outcome (O)	At least 1 of the following outcomes: IOP lowering effects (mean change from baseline, or mean intraocular pressure—depending on data availability) Conjunctival hyperaemia Adverse effects Follow-up: at least at week 12/month 3	/
Study design (SD)	Randomized controlled trials, randomized parallel or cross-over studies	Case reports, case series, preclinical studies, observational prospective non-comparative studies, prospective switch studies, observational parallel studies, retrospective studies, reviews of literature. Studies without wash-out period.
Type of document (T)	English, French, other languages if translation available No time restriction Journal article, congress communication (abstract, poster), regulatory documents from health authorities	/

Information sources: A systematic search of publications was conducted on EMBASE, MEDLINE (through PubMed), CENTRAL (through Cochrane library), ClinicalTrials.gov (www.clinicaltrials.gov), and ICTRP. Handsearching of references cited in individual articles was also performed.

Search: For MEDLINE/PubMed, the algorithm used was as follow: “bimatoprost”[All Fields] AND (“open*angle glaucoma”[All Fields] OR “OAG”[All Fields] OR “Glaucoma, Open-Angle”[Mesh] OR “angle closure glaucoma”[MeSH Terms] OR “ACG”[All Fields] OR “glaucoma”[All Fields] OR “ocular hypertension”[All Fields] OR “OHT” [All Fields] OR “Ocular Hypertension”[Mesh]) AND (“intraocular pressure”[All Fields] OR “efficacy”[All Fields] OR “safety”[All Fields] OR “adverse event”[All Fields]).

Study selection and data collection: Literature search results were uploaded to Citavi 6 Software. Two evaluators independently screened articles obtained from the searches in two steps in Rayyan platform. ²³ The first step included a review of the title and abstract while the second step was based on a full-text assessment. Only records following the predefined eligibility criteria ( Table 1 ) were selected. Two reviewers independently extracted data related to study design and methods, participant characteristics, interventions, and outcomes. Data extraction forms were edited in Excel. Multiple reports of a single study were identified based on data concerning the project or public registration identifier, study design, number of patients included, sponsorship and financial resource. Any discrepancies during screening and data extraction were resolved by discussion or with the help of a third reviewer.

Outcome measures: The primary efficacy outcome was the mean diurnal IOP at week 12, with standard deviation. If not reported, we considered the weighted mean and pooled standard deviation of the three timepoints during the day (8:00 AM, 10–12:00 AM, and 4:00 PM). The reference treatment was set to preservative-free bimatoprost 0.01% eye gel. Secondary outcomes included mean IOP at different timepoints (8:00 AM, 10–12:00 AM, and 4:00 PM), conjunctival hyperaemia, collected directly or as an adverse event, and adverse events, including dry eye, eye pruritus, and punctate keratitis. For all outcomes, data from the intention-to-treat (ITT) population were considered, or if unavailable, replaced by those from the modified ITT, Full Analysis Set, or Per Protocol (PP) populations. The variances of these measurements were aggregated to provide a pooled variance using a decomposition formula available in the R package utilities.

Statistical analysis: A Bayesian NMA was conducted using the gemtc package (version 1.0–2) in R (R Foundation for Statistical Computing, Vienna, Austria). A fixed-effect model was used, assuming a common treatment effect across studies based on expected clinical and methodological homogeneity. Due to the limited number of included studies and the fact that each comparison was informed by a single trial, between-study heterogeneity could not be reliably estimated, and the results are equal to the fixed effects model. The NMA was implemented via Markov chain Monte Carlo simulation using three parallel chains, each with 10,000,000 iterations. A burn-in of 5,000 iterations was applied, and a thinning interval of 2 was used to reduce autocorrelation. Non-informative priors were specified for all model parameters. Convergence was assessed using the Gelman-Rubin diagnostic, with a potential scale reduction factor <1.05 considered indicative of convergence. Trace plots were also visually inspected to confirm adequate mixing and stationarity of the chains. Treatment effects were reported as posterior means with corresponding 95% credible intervals. A non-inferiority margin of 1.5 mmHg was pre-specified for the mean difference in IOP at 12 weeks between groups, as well as for the three timepoints during the day (8:00 AM, 10–12:00 AM, and 4:00 PM). A sensitivity analysis was performed using a frequentist NMA to confirm the main results. Both fixed-effect and random-effect models were computed. Heterogeneity could not be assessed further due to the lack of data and studies available. Consistency, defined as the statistical agreement between direct and indirect comparisons, can be assessed in case of loops in the network. However, due to the anticipated small number of studies, assessing consistency may be challenging.

Risk of bias assessment: Version 2 of the Cochrane risk-of-bias tool for randomized trials (RoB 2) ²⁴ was used to assess the risk of bias in studies included in this meta-analysis.

Results

Study selection: The literature search retrieved 1,585 articles from databases and 159 articles from registers ( Figure 1 ). After removing duplicates and resolving screening decisions through agreement between two evaluators, four studies were included in the analysis.

Figure 1.

PRISMA flow .

diagram of the systematic screening process.

Study characteristics and validity assessment: The characteristics of included studies are shown in Table 2 . The feasibility assessment showed good concordance among three published trials in terms of inclusion/exclusion criteria, interventions, and population characteristics.^6,16,25 These studies focused on patients diagnosed with primary open-angle glaucoma, exfoliative glaucoma, pseudoexfoliative glaucoma, pigmentary glaucoma, primary angle-closure glaucoma, and OHT. Katz et al. and Day et al. included patients with chronic angle closure glaucoma with patent iridotomy or iridectomy, or pseudo exfoliative glaucoma.^6,25 Most of included patients have received ocular hypotensive medication before starting the studies.^6,16,25,26 While IOP at Week 12 was reported in three studies, only two provided change-from-baseline values.^6,16 Figus et al. lacked clarity on washout period use, did not report efficacy outcomes, and had lower baseline IOP ( Table 2 ) with missing confounder data. ²⁶ Two studies added details on central corneal thickness of patients at baseline, ranging from 500 to 600 µm.^6,16 Safety outcomes were heterogeneous across all studies in terms of timepoints and measurement methods. A descriptive analysis was performed to describe safety data.

Table 2.

Description of included studies.

	Day et al. 20136		Figus et al. 201425		Katz et al. 201026			Munoz-Negrete et al. 202416
Study characteristics
Design	Randomized controlled trial, prospective, multicentre, double masked, parallel-group, 12-week study		Randomized controlled trial, prospective, parallel-group, open-label, cohort study		Randomized controlled trial, prospective, double-masked, parallel-group, phase III study			Randomized controlled trial, phase III, international, multicentre, 2-parallel group, investigator-masked, 3-month treatment duration
Location	USA (36 sites)		Italy (1 site)		USA (32 sites)			Europe (90 sites)
Folow-up visits	Weeks 2, 6, 12		Months 6 and 12		Weeks 2, 6, months 3, 6, 9, 12			Weeks 6 and 12
Bimatoprost treatment
Dose	0.03%	0.03%	0.01%	0.03%	0.01%	0.0125%	0.03%	0.01%	0.01%
Preservative-free	Yes	No	No	No	No	No	No	Yes	No
Total number of randomized subjects
	302	295	30	30	186	188	187	236	249
Age in years, mean (SD1)
	64.6 (29–91)	65.0 (29–92)	NA	NA	61.6 (11.7)	64.7 (10.9)	64.2 (11.6)	63.0 (11.9)	63.7 (10.9)
Gender, % males
	43.7	38.6	NA	NA	42.5	38.3	47.6	40.3	38.6
Corneal thickness in µm at baseline, mean (SD)
	554.76	554.46	NA	NA	559.6 (34.5)	560.3 (38.7)	561.3 (35.6)	547.1 (25.7)	544.8 (24.5)
Patients under hypotensive treatment at screening
			100%	100%	74.7%	71.8%	70.6%	100%	100%
IOP in mmHg, mean (SD)
Baseline	23.8 (2.4)	23.8 (2.3)	14.1 (1.7)	14.3 (2.0)	23.5 (3.2)	23.5 (3.4)	23.5 (3.0)	24.2 (2.4)	24.1 (2.4)
At Week 12	17.0 (2.6)	16.8 (2.6)	NA	NA	16.8 (3.0)	16.9 (3.1)	16.4 (2.9)	14.9 (2.5)	15.0 (2.4)
Main conclusions
Efficacy	PF bimatoprost 0.03% was non-inferior and equivalent to P bimatoprost 0.03% in lowering IOP		Efficacy not assessed		Bimatoprost 0.01% was equivalent to bimatoprost 0.03% in lowering IOP throughout 12 months of treatment			PF bimatoprost 0.01% gel has the same efficacy in lowering IOP as P bimatoprost 0.01%
Safety	PF and P bimatoprost 0.03% had similar safety profile		Conjunctival hyperaemia improved and less ocular discomfort in the bimatoprost 0.01% group at each follow-up visit and compared to bimatoprost 0.03%		Bimatoprost 0.01% demonstrated improved tolerability compared to bimatoprost 0.03%, with conjunctival hyperaemia occurring less frequently and with lower severity.			PF bimatoprost 0.01% demonstrated less aggravation of conjunctival hyperaemia compared with P bimatoprost 0.01%

¹In Day et al. 2013, the minimum and maximum were provided instead of SD.

NA: not available; P: preserved; PF: preservative-free; SD: standard deviation.

Network structure and geometry: A network was connected for the variable “mean IOP at Week 12”^6,16,25 ( Figure 2 ). No network was found for the change from baseline in IOP and safety outcomes because of missing data. Since mean baseline IOP was similar between the three studies (ranging from 23.0 to 23.7 mmHg, Table 2 ), the network on the variable mean IOP at week 12 was built.

Figure 2.

Network for mean intraocular pressure at week 12 (P: preserved; pf: preservative-free)

Risk of bias: There was no evidence of bias arising from the randomization process across the included trials. Similarly, no concerns were identified regarding bias due to missing outcome data or selective reporting of results. Risk of bias related to deviations from the intended interventions was identified in two studies ( Figure 3 ). The trial reported by Figus et al. was conducted as an open-label study, which could induce bias, especially in the assessment of conjunctival hyperaemia for which outcome assessors were aware of treatment allocation. ²⁶ In the trial by Muñoz-Negrete et al., participants could distinguish between treatments due to differences in packaging, with preservative-free bimatoprost 0.01% gel provided in single-dose units, whereas preserved bimatoprost 0.01% was dispensed in multidose packaging. ¹⁶ However, the study was investigator-masked, meaning clinical assessments were conducted without knowledge of the treatment assigned to each patient, minimizing potential bias.

Figure 3.

Risk of bias assessment.

IOP data analysis: As previously mentioned, three out of the four studies included in the network reported IOP results.^6,16,25 Preservative-free bimatoprost 0.01% consistently showed non-inferior IOP efficacy to other formulations at Week 12, with Bayesian fixed-effect NMA confirming all 95% credible intervals within the 1.5 mmHg non-inferiority margin ( Figure 4 ). This was confirmed for the mean IOP measured at three different timepoints during the day (data not shown). Sensitivity analyses employing a frequentist approach yielded consistent findings, further supporting the robustness of these data.

Figure 4.

Treatment estimates – Bayesian fixed effects NMA at week 12 compared to preservative-free bimatoprost 0.01% (CrI: credible intervals; P:preserved; pf:preservative-free)

Conjunctival hyperaemia data analysis: In three studies, conjunctival hyperaemia was assessed using standard photographic grading scales, including the Allergan bulbar hyperaemia scale (0–3) and the McMonnies scale. Conjunctival hyperaemia was assessed in these studies at different time points: Week 12 in Day et al. ⁶ and Munoz-Negrete et al., ¹⁶ and Months 6 and 12 in Figus et al. ²⁶ At Week 12, a higher proportion of patients treated with preservative-free bimatoprost formulations had no conjunctival hyperaemia (grade = 0) compared to those receiving preserved formulations (bimatoprost 0.01%: 57.3% versus 43.6% ¹⁶ ; bimatoprost 0.03%: 26.1% versus 23.9% ⁶ ). Additionally, fewer patients in the preservative-free groups experienced worsening of conjunctival hyperaemia scores at Week 12 compared to preserved groups with bimatoprost 0.01% (18.3%versus 30.4% ¹⁶ ).

When reported as an adverse event, conjunctival hyperaemia occurred significantly less frequently in patients treated with bimatoprost 0.01% than in patients receiving bimatoprost 0.03% (28.6% versus 37.4%, p = 0.044). ²⁵ Preservative-free formulations also showed a trend towards reduced conjunctival or ocular hyperaemia compared to preserved formulations (bimatoprost 0.01%: 6.4% versus 10.4% ¹⁶ ; bimatoprost 0.03%: 23.9% versus 26.1% ⁶ ).

Exploration for inconsistency: Inconsistency could not be assessed because only one study connected each formulation.

Discussion

This NMA included four randomized controlled trials evaluating formulations of bimatoprost (preserved and preservative-free bimatoprost from 0.01% to 0.03%) in patients with glaucoma or OHT treated with a bimatoprost monotherapy for at least 3 months. This time frame was chosen to ensure adequate assessment of IOP efficacy and conjunctival hyperaemia, as 12 weeks is a standard duration in clinical trials for evaluating the sustained effects of glaucoma therapies.^27,28

Efficacy was assessed in three out of the four studies, with one study per comparison. The NMA of mean IOP at Week 12 suggests that preservative-free bimatoprost 0.01% gel has similar IOP-lowering effects to other formulations, including preserved and preservative-free 0.03% bimatoprost, although these results are based on limited availability of evidence. The fact that this was confirmed for the mean IOP measured at three different timepoints during the day suggests that bimatoprost dosed once daily is enough to control IOP over 24 h. A non-inferiority margin of 1.5 mmHg was selected based on prior regulatory guidance and clinical trials, where differences below this threshold are not considered clinically meaningful for IOP efficacy in patients with glaucoma or OHT.^29–31 Overall, the IOP results reinforce previous evidence that reducing the concentration of the active ingredient from 0.03% to 0.01% may maintain efficacy,^25,32,33 aligning with the clinical need to balance IOP reduction and tolerability.

Our analysis supports that preservative-free formulations offer comparable efficacy with an improved safety profile. This included a lower incidence of conjunctival hyperaemia, a common adverse effect associated with prostaglandin analogues, which may be exacerbated by preservatives such as BAK. Conjunctival hyperaemia and other ocular surface symptoms are known contributors to treatment discontinuation and poor adherence.^11,12 Preservatives such as BAK can damage the ocular surface by disrupting the tear film, causing inflammation and epithelial harm, especially with long-term use.^34,35 Preservative-free formulations were developed to reduce these effects while maintaining IOP-lowering efficacy. Our analysis confirmed that across included studies, a higher proportion of patients receiving preservative-free bimatoprost had no conjunctival hyperaemia, and fewer experienced worsening of conjunctival hyperaemia compared to those treated with preserved formulations.^6,16 This trend was shown at Week 12 and was consistent regardless of bimatoprost concentration and of the assessment methods, including photographic grading scales and through adverse event reporting. Additionally, low-concentration bimatoprost formulations (0.01%) were associated with less conjunctival hyperaemia compared to the 0.03% formulations. The preservative-free bimatoprost 0.01% gel is formulated with a near-neutral pH, which helps minimize ocular irritation commonly associated with more acidic or alkaline solutions.^36,37 Additionally, its increased viscosity contributes to prolonged corneal contact time, allowing for sustained drug release while reducing the frequency and intensity of adverse sensations such as burning, stinging, and blurred vision.^38,39 The reduced occurrence of conjunctival hyperaemia is clinically significant because it has the potential to enhance patient comfort and adherence to treatment, thereby improving long-term glaucoma management outcomes. These findings are corroborated by real-world data,^{7,32,33,40–42} which consistently reported less conjunctival hyperaemia with preservative-free and low-concentration bimatoprost treatments. In these real-world studies, other safety assessments were performed, (e.g., such as ocular surface disease index, tear break-up time or fluorescein staining ⁷ ), and preservative-free and lower-concentration bimatoprost consistently demonstrated a similar or more favourable safety profile, reinforcing the robustness of this observation across varied clinical settings.^{7,32,33,40–42}

Several limitations temper the interpretation of these findings. The analysis included a limited number of studies, with only one study informing each comparison between formulations, precluding the assessment of heterogeneity or inconsistency within the network. Nevertheless, the included studies had adequate sample sizes for efficacy outcomes. Due to missing or inconsistently reported data, it was not possible to construct a connected network for change-from-baseline IOP or safety endpoints, limiting the scope of comparative safety conclusions. As a result, the impact of tolerability on discontinuation and adherence could not be assessed, and the absence of these data further constrained our ability to build an appropriate and informative safety network. Furthermore, safety outcomes primarily relied on subjective grading scales and were evaluated over a relatively short follow-up period of three months. This duration may be insufficient to fully capture the impact of long-term therapy on ocular surface health. The open-label design of some studies and unclear methodological aspects, such as the washout period in the Figus et al. trial ²⁶ and missing outcome data in others, introduce potential biases that may affect the reliability of the results.

This NMA suggests that preservative-free bimatoprost 0.01% may be associated with comparable IOP-lowering effects and a trend toward lower conjunctival hyperaemia compared with higher-dose or preserved formulations. These findings could support its use as an effective and well-tolerated treatment option for patients with glaucoma and OHT in clinical practice, particularly for patients requiring long-term topical therapy. Enhanced tolerability could be relevant for patient comfort and may have implications for treatment adherence, although this was not directly assessed in the included studies. While preservative-free, lower-concentration bimatoprost formulations may be considered as an option for patients, particularly those with ocular surface concerns, clinical decisions should be individualized. Further adequately powered, long-term studies are needed to confirm these observations and to better characterize their impact on safety, adherence, and long-term clinical outcomes.