Lack of Long-Term Drift in Timolol’s Effectiveness in Patients with Ocular Hypertension

Abstract

purpose. To report the change over time in intraocular pressure in patients treated with topical timolol or placebo.

methods. Ninety patients with untreated ocular hypertension, defined as elevated intraocular pressure and normal visual fields, were randomly assigned to treatment and prospectively observed at 3-month intervals for up to 10 years in a double-masked fashion. The study end point was one eye showing reproducible glaucomatous field defects; patients were also withdrawn if one eye showed intraocular pressure of 35 mm Hg or higher. For each treatment group and visit, the mean and median of all intraocular pressure measurements were calculated. Medians were also calculated for the timolol group, assuming a worst-case scenario in which all patients who reached the end point, or intraocular pressure of 35 mm Hg or more, were assumed to show higher intraocular pressure than those remaining in the study. Distributions of slopes for intraocular pressure over time were compared between treatment groups.

results. Means and medians of follow-up intraocular pressure over time did not differ between timolol- and placebo-treated patients. This was also true when assuming a worst-case scenario. Slopes of intraocular pressure over time did not differ statistically between treatment groups.

conclusions. In agreement with other masked and controlled studies and in conflict with uncontrolled ones, the present study did not demonstrate long-term drift of intraocular pressure in patients with ocular hypertension treated with topical timolol.

Timolol, the first commercially available topicalβ -blocker, was introduced in the 1970s. It produced considerable reduction of intraocular pressure (IOP), had few local side effects, and soon became the most widely used antiglaucoma agent. Some early long-term studies (follow-up between 18 and 35 months) in patients with ocular hypertension and primary open-angle glaucoma ^{1 2 3 4} reported a slow reduction of the IOP-lowering effect of timolol over time. These studies were prospective, but no control groups were used. Similarly, Brubaker et al. ⁵ reported that although a substantial initial reduction of aqueous humor flow was seen after 1 week of timolol treatment in a small number of patients with glaucoma, there was a smaller reduction after 12 months. The mechanism of adaptation is unknown. 

Boger ⁶ reviewed the subject and introduced the term“ long-term drift” to describe the reported diminishing IOP reduction during treatment with timolol. This alleged phenomenon of the diminishing effectiveness of timolol has been discussed in textbooks of ocular pharmacology ^{7 8 9} and is often referred to in printed material distributed by the drug industry. ^{10 11 12}  

Several prospective and controlled studies, however, have been unable to show evidence of any diminishing effectiveness of timolol when compared with control groups. Thus, in a 6-year prospective, controlled study that compared topical timolol and no treatment in patients with suspected glaucoma, Chauhan et al. ¹³ could not find any difference between groups in the course of IOP over time. Instead, both treated and untreated eyes showed a slight increase of IOP over time, but the curves were parallel. In a study of patients with ocular hypertension, Kass et al. ¹⁴ found no difference between timolol- and placebo-treated eyes over a 5-year span, whereas in a similar study, Epstein et al. ¹⁵ found a small trend of diminishing IOP reduction over 5 years when timolol-treated eyes were compared with untreated control eyes. 

As one of the most widely prescribed glaucoma medications, timolol has often been used as the gold standard when the IOP-reducing effect of new medications is evaluated. Because of the lack of consensus about long-term drift, or diminishing effectiveness, of this standard drug, we decided to publish a detailed analysis of the IOP-reducing effect of topical timolol treatment in the Malmö Ocular Hypertension study, ¹⁶ a prospective study with follow-up of up to 10 years. 

Subjects and Methods

The study was designed as a randomized, prospective, double-masked trial comparing topical timolol treatment with placebo treatment in patients with ocular hypertension and at least one additional risk factor. Ocular hypertension was defined as elevated IOP and normal visual fields. Patients seen in clinical practice were recruited consecutively, between 1981 and 1987. A large majority of the patients (79/90) had never been treated with IOP-lowering drugs before entering the study. Among the 11 patients who had been previously treated, the interval between study inclusion and prior treatment ranged between 1 and 10 years. The tenets of the declaration of Helsinki were followed. All patients gave informed consent to participate in the study, and the Ethics Committee of the Medical Faculty of the University of Lund approved the study protocol. 

Patients with untreated IOP of 22 mm Hg or higher, calculated as the mean of a diurnal curve with three readings obtained during office hours, and at least one additional risk factor were included. Risk factors were suspicious disc appearance, defined as localized complete or incomplete rim notch; cup–disc ratios of 0.6 or more; cup–disc discrepancy of 0.2 or more between the two eyes or marked cup asymmetry defined as vertical cup–disc ratio exceeding horizontal cup–disc ratio by 0.2 or more; presence of disc hemorrhage; positive family history of glaucoma; pseudoexfoliation or pigment dispersion syndrome; diabetes; or mean IOP of 27 mm Hg or higher, defined as the mean of three measurements. Visual fields had to be normal as tested with static computerized perimetry plus additional manual, kinetic perimetry in the peripheral and midperipheral fields. Exclusion criteria were mean untreated IOP of 35 mm Hg or more; visual acuity of 0.3 or less; concomitant medication known to affect IOP, including β-blocking drugs, clonidine, and carbonic anhydrase inhibitors; history of intraocular surgery; lens opacities or pupils that would not dilate sufficiently for disc photography; unreliable patient who could not perform adequately during perimetry; history of pulmonary or cardiac disease contraindicating the use of β-blocking agents. 

Patients were randomly assigned to 0.5% topical timolol (Timoptic; Merck, Stockholm, Sweden) or placebo twice daily. The baseline characteristics—age, IOP, and gender—were very similar and were not significantly different between treatment groups. The risk factors were also evenly distributed between patients randomized to placebo and timolol treatment. ¹⁶  

Randomization took place after the baseline visit. Eye drops containing timolol or a placebo were packed in identical bottles marked with a patient allocation number but with no other information about the contents. The bottles were dispensed by an independent pharmacy, which had no information about the identities of the patients. Thus, whether a patient was receiving timolol or the placebo was masked to both investigators and patients. This information was made available from a separate allocation center only when patients left the study. 

The first follow-up was after 1 month and then follow-up visits occurred at 3-month intervals. Office-hour IOP curves, a general ophthalmic examination, computerized static threshold perimetry, and disc photography were performed at each visit. 

The end point for study subjects was the development of reproducible visual field defects. Patients were also withdrawn from the study if the mean IOP of the diurnal curve was 35 mm Hg or more, if visual field defects caused by disease other than glaucoma appeared, or if they began treatment with systemic β-blockers. Some patients were also lost to further follow-up because of death or grave systemic illness, because of moving out of the area, or because of their decision withdraw from the study. 

All patients included in the current analysis were followed up for at least 9 months. For each patient, IOP was calculated as the mean of the office-hours tension curve at every visit. One eye of each patient was included in the analyses. Patients with only one eye that fulfilled the inclusion criteria were, of course, represented by that eye. In patients in whom both eyes fulfilled the inclusion criteria, one eye was randomly chosen for analysis, except eyes in which reproducible glaucomatous visual field defects developed or eyes in which IOP reached 35 mm Hg or more, which were always included in the current analysis. 

The IOPs were analyzed in three ways: (1) Mean IOP for each visit and treatment group was calculated and compared over time. (2) The median (50th percentile) IOP was calculated for each visit and each treatment group. Due to withdrawals and patients’ reaching end points, the number of subjects included in calculations of the median decreased over time. Data from patients who were lost to follow-up for reasons other than development of glaucoma and/or IOP increase to 35 mm Hg or more were omitted from the analysis. Additional medians were calculated for each treatment group, assuming worst-case scenarios in which all patients who left the study because of development of glaucoma or IOP of 35 mm Hg or more were included and assumed to have higher IOP after withdrawal from the study, than all patients remaining in the study. (3) Slopes of IOP over follow-up time were calculated for all patients and compared between treatment groups. 

Results

Ninety patients were divided into two groups at random, 46 to timolol and 44 to placebo treatment. Six of these patients were not included in the long-term analysis, because they left the study after the first 3 months due to death (one patient), because they wanted to withdraw (four patients), or because they began systemic β-blocker treatment (one patient). IOP reduction in these six patients was similar to that of those included in the long-term analysis: 6.1 and 1.7 mm Hg in the timolol- and placebo-treated patients, respectively. Thus, 84 patients, 42 in each of the treatment groups, were included in the present long-term analysis. Mean age at baseline was 63.5 years (range: 40–75) in the timolol group and 61.1 years (range, 31–79) in the placebo group at the time of inclusion. 

After 1 month, the mean IOP reduction compared with baseline was 6.8 mm Hg in the timolol group and 2.0 mm in the placebo group, and the difference in IOP between treatment groups was 4.0 mm Hg on average (P < 0.0001, t-test). This mean IOP difference between treatment groups showed only small fluctuations throughout the study (Fig. 1) . At the 10-year visit, the difference was approximately the same (4.2 mm Hg) but was no longer significant (P = 0.183) and had poor power because of the small number of patients remaining in she study at that time (6 receiving placebo and 13 timolol). Study power, however, was high: better than 90%, up to and including the 90th month, to detect a reduction of differences between treatment groups of 1 mm Hg compared with the initial reduction of IOP. 

During the full 10-year period, 7 timolol- and 15 placebo-treated patients were excluded because of development of glaucomatous visual field defects. Two timolol- and four placebo-treated patients were excluded because IOP increased to 35 mm Hg or more. Study attrition for the two treatment groups is shown in Table 1 . 

The median IOPs for each visit and treatment are shown in Figure 2 . Differences between mean and median values were small, ranging from 0.25 to 0.65 mm Hg. The initial treatment-induced difference between the timolol and placebo groups did not diminish over time. Even the worst-case scenario failed to change the result. It was only after the 99-month visit, when no more than 11 placebo-treated and 15 timolol-treated patients remained in the study, that the worst-case timolol medians shifted in the direction of the medians of the placebo group (Fig. 2 , filled symbols). 

IOP trends for each individual were calculated as slopes of IOP over follow-up time using linear regression analysis (Fig. 3) . There was no evidence of a larger positive tail in the timolol-treated group. 

Discussion

Our results indicate that the IOP-reducing effect of timolol in patients with ocular hypertension plus an additional risk factor was not diminished over a 10-year follow-up when comparisons were made with a placebo-treated control group. We chose to analyze means of IOP reduction over time to facilitate comparison between the results of the present study and earlier studies. But we also compared medians, which allowed us to include patients who left the study because of development of glaucomatous field defects or very high IOP. No evidence of long-term drift of the effects of timolol was seen with either approach. It could be argued that such a trend was seen after 99 months when our timolol worst-case scenario was analyzed (Fig. 2) , but at this time the median was calculated using only a small number of patients remaining in the study. With few patients, the median can be expected to increase in this kind of worst-case scenario analysis. The worst-case scenario for the placebo-treated patients showed a marked drift earlier than the timolol-treated ones, which can be explained by the higher number of worst cases in that treatment group. 

Our comparisons of means or medians could, however, hide a small subgroup with long-term drift. This is why we calculated the slope of follow-up IOP for each patient and compared the distributions of such slopes between treatment groups. In that case, there was still no evidence of such a subgroup with long-term drift, which would have been visible as a long positive tail in the distribution of IOP slopes of the timolol-treated patients shown in Figure 3 . 

A prospective but unmasked study by Gandolfi and Vecchi ¹⁷ compared long-term IOP drift in patients receiving timolol, or timolol interrupted by periods of dipivefrin treatment. They reported long-term drift of IOP in 50% of patients treated with timolol only. In this study, long-term drift was defined as an increase of IOP of 5 mm Hg or more, compared with the first follow-up visit, and after interruption of treatment, the next IOP measurement was allowed to further increase by a maximum of 2 mm Hg. We applied that criterion in our placebo group, but it could not be tested in our timolol group because those subjects were treated without interruption. We found that 50% of the placebo-treated patients fulfilled this long-term drift criterion, indicating that the criterion has very low specificity. 

The notion of long-term drift may have arisen from previously published uncontrolled and/or unmasked studies. Another possible explanation could be publication bias, which has been recognized as “a large problem that may seriously distort attempts to estimate the effect under investigation.” ¹⁸ Questions of statistical power can always be raised when results are negative and may make publication more difficult than if positive results are found, and it has also been shown that investigators often decide not to submit reports of studies with negative results. ¹⁹ Certain actions have been proposed ²⁰ or taken to reduce such publication bias. ^{21 22}  

Thus, in the present study involving patients with ocular hypertension, there was no evidence of development of subsensitivity to timolol. These results are in agreement with other controlled studies, although several uncontrolled studies have suggested such long-term drift. 

 Supported by Merck, Sharpe & Dohme, Sweden, and by funds administered by Malmö University Hospital.

 Submitted for publication February 16, 2001; revised May 22, 2001; accepted July 18, 2001.

 Commercial relationships policy: F.

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked“ advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

IOP over time calculated as the mean for each visit and treatment group. At baseline mean IOP was slightly lower (0.8 mm Hg) in the placebo group than in the timolol group. After 1 month, both groups showed reduced IOP, indicating that some of the effect was due to regression to the mean. The initial difference in the IOP-lowering effect between treatment groups was stable over time.

Table 1.

 

Study Attrition

	Months of Follow-Up
		0		12		24		36		48		60		72		84		96		108		120
Timolol group
Mean IOP ≥35 mm Hg								1		1
Glaucomatous VFD		1		1		1		1		1						1				1
Causes unrelated to glaucoma or IOP		6		4		5				2		2		2		1		2
Patients	46		39		34		28		26		22		20		18		16		14		13
Placebo group
Mean IOP ≥35 mm Hg				1		1						1						1
Glaucomatous VFD				3				2		3		2				2		2		1
Causes unrelated to glaucoma or IOP		3						3		6		2		2		1				2
Patients	44		41		37		36		31		22		17		15		12		9		6

 Data are number of patients or control subjects. VFD, visual field defects.

Figure 2.

 

IOP over time calculated as the median (50th percentile) for each visit and treatment group. The IOP curves for the two treatment groups were parallel over the 10 years of follow-up. The worst-case-scenario medians—that is, assuming that all patients excluded because of development of glaucomatous field defects or IOP of 35 mm Hg or more would show higher IOP than all remaining patients—showed only a slight upward drift after 99 months in the timolol group and a more marked one after 81 months in the placebo group.

Figure 3.

 

IOP slopes over time. Both distributions are well centered around 0, and there is no evidence of a larger positive tail in the timolol group. There is no significant difference between the two distributions (P = 0.33, Student’s t-test).