Inclusion Criteria

Inclusion criteria included diagnosis of primary open-angle glaucoma based on the European Glaucoma Society Guidelines criteria, treatment with nonfixed combination of latanoprost and timolol for at least 3 months or latanoprost plus timolol fixed combination for at least 6 months, IOP less than 21 mm Hg at the baseline 8 am time point on latanoprost plus timolol fixed combination treatment, and target IOP not reached as set by the treating physician.

Exclusion Criteria

Exclusion criteria included contraindications to β-blockers; closed or barely open anterior chamber angles or a history of acute angle closure; ocular surgery or argon laser trabeculoplasty within the previous 3 months; ocular inflammation or infection occurring within 3 months before the pretrial visit; neovascular patients (including diabetic patients with neovascular glaucoma); concomitant topical ocular medications that can interfere with study medications; hypersensitivity to benzalkonium chloride or to any other component of the trial drug solutions; other abnormal condition or symptom preventing the patient from entering the trial, according to the investigator’s judgment; any history of refractive surgery, pregnancy, breastfeeding, or childbearing potential without adequate contraception; inability to adhere to the treatment and visit plan; and participation in any other clinical trials within 3 months before the pretrial visit.

Study Plan

Patients treated with latanoprost plus timolol with an IOP of less than 21 mm Hg but requiring further IOP reduction were enrolled in the study. Patients receiving the nonfixed combination were checked for eligibility and scheduled for a wash-in period of 6 weeks with latanoprost plus timolol fixed combination at the prescreening visit.

At the baseline visit, patients were randomized to either bimatoprost plus timolol or travoprost plus timolol and were treated for 3 months. After 3 months, the patients randomized to bimatoprost plus timolol were switched to travoprost plus timolol and patients randomized to travoprost plus timolol were switched to bimatoprost plus timolol. There was no wash-out between the 2 treatments.

Assessment of baseline data was conducted before randomization and assessment of IOP and tolerability were carried out at baseline, 1 month, 3 months, 4 months, and 6 months after randomization. The investigator could choose to have a nonscheduled safety visit in between the scheduled visits. At the baseline visit, patients’ ophthalmic and systemic history were recorded and gonioscopy, pachymetry, and standard automated achromatic perimetry were performed. Patients then were randomized to be treated with either bimatoprost plus timolol or travoprost plus timolol, both dosed in the evening at 9 pm . A complete ophthalmologic examination, including visual acuity and refraction and lid and slit-lamp examination, was performed at the baseline, month 3, and month 6 visits. Additionally, IOP was measured at 8 am , 11 am , 2 pm , 5 pm , and 8 pm by Goldmann applanation tonometry. Month 1 and 4 visits were safety visits; IOP was measured at 11 am , and adverse events were recorded.

Study Outcomes

The primary outcome was to compare the mean diurnal IOP from the daily curve after 3 months of bimatoprost plus timolol and travoprost plus timolol treatment. Secondary outcomes were to compare between groups the diurnal IOP change from baseline 3 months after the start of each treatment (for each time point separately and for the mean IOP), mean morning (8 am , 11 am ), and mean afternoon IOP (2 pm , 5 pm , 8 pm ), IOP change from baseline at 1 month after the start of each treatment, diurnal IOP change from baseline after switch to the other therapy, distribution of IOP reductions from baseline expressed as percentages, and tolerability.

Double masking was performed by relabeling and repackaging the study drugs in identical boxes. Each patient received after randomization an identical masked box containing 4 smaller masked boxes, each containing 1 masked bottle with the first treatment (1 bottle per month of treatment plus 1 extra safety bottle). At the 12-week visit (crossover visit), a second box containing 4 packed bottles of the second treatment were dispensed to each patient. Each patient eligible for the trial was allocated to 1 of the 2 treatment sequences (either first bimatoprost plus timolol for 3 months and thereafter travoprost plus timolol for 3 months or vice versa) according to a per-center computer-generated randomization code list.

Clinical Tolerability Assessments

Best-corrected visual acuity (Snellen chart), biomicroscopy, and ophthalmoscopy were recorded at the pretrial visit and at each follow-up visit. Any kind of adverse event was recorded. Changes of conjunctival hyperemia were recorded at the slit lamp using a standard scale (0 to 3, where 0 is none, 1 is mild, 2 is moderate, and 3 is severe) with the help of a standardized photographic chart. Superficial keratitis, defined as the presence of small circular epithelial erosions in the cornea, was also graded as none, mild, moderate, and severe (none = no staining; mild = rare stained erosions localized close to the lid margins; moderate = rare stained erosions localized in the 4 quadrants; diffuse = diffuse stained erosions involving the 4 quadrants).

Analysis

Data are presented as mean and 95% confidence intervals for continuous variables and frequencies for categorical variables. For the IOP recordings, the mean value of 2 measurements at each time point (or the median of 3 readings if the first 2 were not within 2 mm Hg) was used in the calculations. In addition, if both eyes fulfilled the eligibility criteria, the worse eye (i.e., with higher IOP) was used in the analysis. An intent-to-treat approach was used to analyze the IOP variables, and in case of missing data, the last observation available was carried forward. A mixed-model analysis of variance model was used to compare the IOP variables between groups, including the following effects: patients as random effect, and drug and interaction between drug and treatment sequence as fixed effects (to explore potential carryover effects). Categorical variables such as proportions and tolerability variables were analyzed using the Pearson chi-square test or Fisher exact test as appropriate.

The sample size estimate was based on a test for superiority for a crossover study. Setting the α error to 5%, 85 patients per group would enable to detect a difference of 1 mm Hg, assuming a standard deviation of 2.3 mm Hg (based on previously published data ) with a power of 80%.