Diurnal IOP Fluctuation – Not an Independent Risk Factor for Glaucomatous Visual Field Loss in High Risk Ocular Hypertension Authors:

Purpose: To study whether intraocular pressure (IOP) fluctuations contribute to the risk of developing glaucoma in patients with high-risk ocular hypertension. Methods: Ninety patients included in the Malmö Ocular Hypertension Study were examined every 3 months with office hour diurnal tension curves and computerised perimetry. Patients were followed prospectively for 10 years or until glaucomatous visual field loss could be demonstrated. Post-study data was included in the analyses, extending maximum follow-up to 17 years. Results: After 17 years 37 patients had developed glaucomatous visual field defects. When applying univariate Cox regression analyses, mean IOP of all measurements during the prospective part of the study was a significant risk factor for developing glaucoma (95% confidence interval (CI): 1.08 —1.39), while IOP fluctuations were almost significant (95% CI: 0.98—1.93). When separating effects of mean IOP level and mean IOP fluctuation using Cox multiple regression analysis, only IOP level came out as significant (95% CI: 1.09–1.38), and IOP fluctuations did not contribute to the risk (95% CI: 0.80—1.60). IOP fluctuation depended linearly on IOP level (p<0.0001), i.e. IOP fluctuation was larger in eyes with higher IOP levels Conclusion: IOP fluctuations were not an independent risk factor for the incidence of glaucomatous visual field loss in subjects with ocular hypertension. Bengtsson & Heijl: IOP fluctuations and risk for GLVFL 3 Introduction Increased intraocular pressure (IOP) is a well-known risk factor for development of glaucoma [17] and also for glaucoma progression [14]. IOP is a dynamic parameter, i.e. IOP values and measurements fluctuate during the 24 hours. Fluctuations of IOP have been the subject of many studies, but most describe diurnal patterns of IOP. Maslenikow [21] was one of the first in 1904. Much later Duke Elder (1952) reported that diurnal variation seldom exceeded 5 mmHg in healthy subjects [9]. This finding was confirmed by Drance (1960), who also showed that a majority of untreated glaucoma patients had IOP fluctuations larger than 5 mmHg [7]. At that time increased IOP fluctuations were thought of as an indicator of early glaucoma [8, 18], although it was also known that IOP fluctuations are proportional to IOP levels [19, 24], i.e. that patients with higher IOP also have larger IOP variation. Later, De Vivero and co-workers (1994), could not demonstrate higher IOP variation in normal tension glaucoma patients than in normal subjects [6]. Today reports are available, where the effects of IOP level and IOP fluctuations as risk factors for glaucoma and glaucoma progression have been separated. Asrani et al. reported that increased IOP fluctuations, as revealed by diurnal tension curves using home tonometry, increased the risk for glaucoma progression [2]. Only patients with IOP < 24 mm Hg during follow-up were included in the analysis. In a study by Ishida and co-workers [16] results depended upon which criteria for visual field progression that were applied. In the study by Daugeline et al 1999 [5] IOP variability was not an independent risk factor. Both the latter studies included patients with normal tension glaucoma only. Effects of IOP fluctuations on visual fields have also been studied in glaucoma patients with elevated pressures. Smith (1985) found that diurnal fluctuations were as large in patients with elevated IOP without field loss as in glaucoma patients with field defects [25], while Bergeå Bengtsson & Heijl: IOP fluctuations and risk for GLVFL 4 et al. (1999) [4] reported that glaucoma patients with smaller IOP variations progressed less often than those with larger variations. In a longitudinal study of patients with ocular hypertension Gonzales and co-workers (1997) reported that increased IOP variation at one baseline diurnal tension curve was a risk factor for development of glaucomatous field defects using manual perimetry [10]. A couple of studies, reporting IOP fluctuations to be an independent risk for glaucoma and glaucoma progression [2, 4] have analyzed IOP measurements obtained at different times, e.g. baseline IOP fluctuations and follow-up IOP levels, or vice versa. This approach is less than optimal to determine whether IOP fluctuations are an independent risk factor, i.e. whether they predict future risk better than the IOP level itself, e.g. mean IOP. The alleged proof for IOP fluctuations as an independent risk factor for glaucoma or glaucoma progression is thus rather weak. Despite this, increased diurnal fluctuation is frequently referred to as an important risk factor for glaucoma progression, e.g., in printed and web based material provided by the drug industry. Our aim was to study the effect of IOP fluctuations on the incidence of glaucomatous visual field loss in patients with ocular hypertension, who were prospectively followed with diurnal office hour tension curves and computerized perimetry every 3 months for up to 10 years, and subsequently followed for a maximum of 17 years as clinical patients. We wanted to separate the effects of IOP variations and IOP level, in order to study whether IOP fluctuations were an independent risk factor. Bengtsson & Heijl: IOP fluctuations and risk for GLVFL 5 Subjects and Methods The prospective study Ninety patients with ocular hypertension were included in the prospective Malmö Ocular Hypertension study. Detailed inclusion criteria have been published before [11, 3], but briefly patients had to have an untreated IOP ≥ 22 mm Hg, defined as the mean of three measurements with the Goldmann tonometer at, 8.00 11.30 and 15.30 hours, and normal visual fields. Prior to inclusion, visual fields were tested with static threshold computerised perimetry with the Competer perimeter [12] and with peripheral and nasal mid-peripheral kinetic testing on the Goldmann perimeter. At least one additional risk factor was required for inclusion: suspect disc topography or disc haemorrhage, positive family history of glaucoma, pseudo-exfoliation or pigment dispersion syndrome, diabetes, or mean IOP of ≥27 mm Hg on the baseline diurnal tension curve. Patients were recruited consecutively between 1981 and 1987, and gave informed consent to participate. The tenets of the declaration of Helsinki were followed and the study was approved by the Ethics Committee of the Medical Faculty of Lund University. The study was designed as a randomized, prospective, double masked trial comparing topical timolol treatment with placebo treatment. Patients were prospectively followed every 3 months in the study for 10 years or until they reached the outcome, which was development of reproducible glaucomatous visual field loss. Some patients also left the prospective study because of: mean IOP ≥35 mmHg, appearance of non-glaucomatous visual field defects caused by neurological or retinal conditions, dense cataract, adverse reactions, systemic ßblocker treatment, serious illness, or because they wished to withdraw. Detailed information Bengtsson & Heijl: IOP fluctuations and risk for GLVFL 6 about attrition has been published before [11]. These patients were included in the retrospective follow-up part. All patients underwent diurnal Goldmann tonometry (8.00 11.30 and 15.30 hours), computerized threshold perimetry with the Competer perimeter, disc photography and a general eye examination at each visit every three months while being followed in the prospective part of the study. During this part 22 patients developed visual field defects, defined as reproducible clusters of depressed test point locations as compared to the mode value of the test. Seven patients died and 3 were lost to follow-up. The retrospective follow-up part Also after leaving the prospective part of the study all living patients, but the 3 lost for followup during the prospective part, were followed at our department. For the purpose of this paper, all available post-study patient records were searched except those of patients who had developed reproducible glaucomatous field loss, had been lost to follow-up, or died during the prospective part of this study. Throughout the post study phase most patients were changed from computerized threshold perimetry on the Compter perimeter the to the 30-2 Full Threshold test of the Humphrey Field Analyzer 640 (Carl Zeiss Meditech, Dublin, CA, USA). Thus, those patients who developed glaucoma field loss, defined as reproducible cluster of significantly depressed points in pattern deviation probability maps [15], during the retrospective follow-up part were examined with the Full Threshold program of the Humphrey Field Analyzer. Bengtsson & Heijl: IOP fluctuations and risk for GLVFL 7 IOP parameters All IOP data included in our analyses were those collected using diurnal tension curves during the prospective part of the study. Individual levels of IOP were computed in different ways: 1. as the mean of all measurements 2. as the mean of the maximum IOP measurement of each diurnal tension curve 3. as the mean of a randomly chosen value of the three IOP values of each tension curve Individual IOP variation was also computed in different ways: 1. as the mean of the range of each diurnal tension curve 2. as the maximum range in all tension curves 3. as the difference between the lowest and highest IOP values measured during the whole prospective part of the study All analyses were based on patients rather than eyes. When both eyes of a patient were eligible, the first eye to reach the outcome was included in the analyses. When visual field remained normal, one eye was randomly selected. Statistical analyses Univariate Cox regression analyses were performed to evaluate the different parameters of IOP levels and IOP variability as risk factors for developing glaucomatous visual field defects. A multivariate Cox regression analysis was performed to evaluate the simultaneous effect of IOP fluctuations and IOP level. All Cox analyses were performed using treatment group, timolol or placebo, as a stratification variable, i.e. effects of treatment on IOP were controlled for. Bengtsson & Heijl: IOP fluctuations and risk for GLVFL 8 Results Fifty-two females and 38 males were included in the study. Mean age at baseline was 62 years, ranging from 48 to 79. Including post study data increased follow-up time from up to 10 years to a maximum of 17.5 years. Mean follow-up time was 8.5 years. Forty-four patients were randomised to treatment with placebo and 46 to timolol. Thirty-seven patients developed reproducible glaucomatous visual field defects, 18 initially randomised to timolol , and 19 to placebo. The mean follow-up IOP was 22.7 mmHg ranging from 15.3 to 33.0, and mean IOP fluctuations, calculated as the mean diurnal range, was 3.7 mmHg ranging from 2.0 to 8.2. Using univariate analyses all parameters describing IOP level came out as highly significant risk factors, Table I. The risk for developing glaucoma increased with slightly more than 20% for each mmHg of pressure. Mean IOP of all measurements showed somewhat higher risk than mean of the maximum or of a randomly chosen IOP value from every tension curve. Table I Cox univariate analyses – different parameters for IOP level Explaining variable risk 95% confidence interval p-value Mean of all IOP measurements 1.226 1.08 – 1.39 0.0013 Mean of maximum IOP 1.203 1.07 – 1.36 0.0027 Mean of random IOP 1.216 1.08 – 1.38 0.0017 When analysing IOP fluctuations using univariate models, the mean of the daily range came out as an almost significant risk factor (p=0.063), Table II. The highest daily range and the highest total range was not associated with any increase of risk. Bengtsson & Heijl: IOP fluctuations and risk for GLVFL 9 Table II Cox univariate analyses – different parameters for IOP variability Explaining variable risk 95% confidence interval p-value Mean of daily range 1.379 0.98 – 1.93 0.063 Maximum of all daily ranges 1.039 0.91 – 1.19 0.577 Range between minimum and maximum IOP of all measurements 1.021 0.95 – 1.09 0.56 When applying a multivariate analysis including both IOP level and IOP fluctuations, only mean IOP came out as a significant risk. Thus, the effect of IOP fluctuations was nonsignificant (p=0.49), when separated from IOP level using multivariate analysis, Table III. Table III Cox multivariate analysis – IOP level and IOP variability Covariates risk 95% confidence interval p-value Mean of all IOP of all measurements 1,209 1.09-1.38 0.005 Mean of daily range 1.13 0.80-1.60 0.49 Discussion In this group of patients with high risk ocular hypertension, IOP fluctuations were not an independent risk factor for glaucoma development. Mean IOP was a strong and highly significant risk factor both in uniand in multivariate analyses, while IOP fluctuations were almost significant but only when IOP level not was taken into account. These results could be explained by the dependence of IOP fluctuations on IOP level; IOP fluctuations increased with 0.17 mmHg for each mm increase in IOP level (linear regression analysis p<0.0001). Thus, IOP fluctuations were proportional to mean IOP, Fig.1. Bengtsson & Heijl: IOP fluctuations and risk for GLVFL 10 Fig.1 IOP fluctuations as percentages of mean IOP. Fluctuations relative to IOP level are of similar magnitude all over the range of mean IOP values. Top: filled symbols indicate timolol treated patients; open symbols indicate placebo treated patients. Mean IOP was lower in the timolol group, but fluctuations were similar among timolol and placebo treated patients. Bottom: filled symbols indicate patient who developed glaucomatous field defects; open Bengtsson & Heijl: IOP fluctuations and risk for GLVFL 11 symbols indicate patients with no visual field defects after 17 years of follow-up. IOP fluctuations were similar in the two groups. Gonzales and co-workers [10] found that IOP fluctuations were a significant risk factor for glaucomatous visual field defects in patients with ocular hypertension. They compared development of field loss in two groups of patients: one with IOP fluctuations > 5mmHg and another with such fluctuations ≤5 mmHg at a baseline tension curve. Thus, the importance of IOP level was not considered at all. Our univariate analysis also suggested that increased IOP fluctuations may be a risk factor, but when taking IOP level into account IOP fluctuations lost importance totally. Most studies have looked at IOP fluctuations as a risk factor for progression of glaucomatous visual field loss. Such progression may be more difficult to estimate than onset of glaucomatous damage because of the much larger perimetric test-retest variability in diseased eyes as compared to normal ones. Different criteria for field progression could then yield different study results. Thus, Ishida and co-workers found that increased IOP fluctuations was a risk for visual field progression when defining progression as pointwise deterioration of threshold values, but not when defining progression as deterioration of the global Mean Deviation values [16]. In our study all patients were tested with the threshold program of the computerized static Competer perimeter during the prospective phase. During the clinical follow-up the patients switched to the 30-2 Full Threshold program of the Humphrey Field Analyzer (Carl Zeiss Meditech Inc, Dublin, Calif). This means that the patients who developed glaucoma during the prospective part were diagnosed using a different instrument than those who developed Bengtsson & Heijl: IOP fluctuations and risk for GLVFL 12 reproducible field loss during the retrospective follow-up part. We do not believe, however, that there were any important differences between the two instruments in their ability to identify repeatable glaucomatous field loss. Some early studies reported similar sensitivities and specificities between the Competer and the Octopus perimeters (Haag-Streit AG Switzerland), and between the Octopus and Humphrey Field Analyzer perimeters [13,22]. Asrani et al found that IOP fluctuations, as measured with self tonometry at home, was a significant risk factor for progression independent of IOP level [2]. This report has often been cited as the proof for variability as a risk factor for glaucoma progression. Others have pointed out, however, that the Asrani study lacked well-defined criteria for visual field glaucoma progression, and also doubts have been raised about patients’ ability to perform the self tonometry making the significance of the results less reliable [26]. Further, the analysis was limited to patients with normal or only slightly increased IOP. More than 60% of the subjects originally included were excluded from the analysis because they showed IOP > 24mmHg once or more during the 5 year follow-up. The conclusions of Asrani and coworker’s study could therefore not be considered as relevant for the large majority of clinical glaucoma patients having increased IOP. The patients included in Asrani’s study were quite different from the ones included our study. We included patients with increased IOP only, while such patients were excluded from the analyses in Asrani’s study, hence, there is little similarity between patients analysed in the present study and their paper. In the study by Bergeå et al. also multivariate analyses including several different IOP parameters were performed [4]. Prior to the risk analyses they carried out principal component analysis to avoid multicollinearity, i.e to avoid highly inter-correlated explanatory variables in the same model. This resulted in a number of multivariate regression models Bengtsson & Heijl: IOP fluctuations and risk for GLVFL 13 including different IOP parameters. Two of them included IOP range, calculated in the same way as our “mean of daily range”. In these two regression models IOP range was combined with: untreated baseline IOP, and with IOP change, i.e. treatment effect calculated as the difference between untreated baseline IOP and mean of treated follow-up IOP. Effects of follow-up IOP level and follow-up IOP fluctuations were not simultaneously tested in the same model. Thus, this study neither showed, nor disapprove, that IOP fluctuations were a risk factor for glaucoma progression independent of IOP level. A recent paper by Nouri-Mahdavi and co-workers [23] reports that IOP fluctuations, calculated as the standard deviation of single IOP measurements measured at different days during the follow-up of the Advanced Glaucoma Intervention Study and analysed using multivariate technique, was a significant risk factor for visual field deterioration. Surprisingly, and in contrast with an earlier AGIS report [1], mean IOP did not increase risk for field worsening. While the apparent differences between our results and several studies discussed above, most likely depend on whether both IOP level and IOP variability have been independently analysed in a multivariate way. We should point out that we measured IOP during office hours only. More measurements outside office hours would of course improve estimates of fluctuations, but retention was also important. It is not likely that patients would have stayed for 10 years in the prospective part of the study if we had asked them to come for measurements more than three times per day every three months. Bergeå also measured office hour curves 3 times per day[4], while the patients included in Asrani and co-workers’ study [2] measured their own pressure 5 times a day, from early morning to bedtime. In this context it could be of interest to mention a more recent study by Liu and co-workers [20], where IOP Bengtsson & Heijl: IOP fluctuations and risk for GLVFL 14 was measured every 2 hours during a 24-hour period in patients with untreated glaucoma and in age-matched healthy subjects. They found that diurnal IOP was higher in the glaucoma eyes, but that the diurnal to nocturnal IOP range was larger in the healthy eyes. This result could be interpreted as high diurnal IOP is a risk for glaucoma, but not large IOP fluctuation. In summary in the present study IOP fluctuations were not an independent risk factor for the incidence of glaucomatous visual field loss. Acknowledgements This study was supported by grant K2002-74X-10426-10A from the Swedish Research Council, by the Järnhardt foundation, and by funds administered by Malmö University Hospital. Bengtsson & Heijl: IOP fluctuations and risk for GLVFL15 References1. AGIS investigators. The Advanced Glaucoma Intervention Study (AGIS) (2000) Therelationship between controil of intraocular pressure and visual field deterioration. Am JOphthalmol 130:429-40. 2. Asrani S, Zeimer R, Wilensky J, Gieser D, Vitale S, Lindenmuth K (2000) Large diurnalfluctuations in intraocular pressure are an independent risk factor in patients with glaucoma. JGlaucoma 9: 134-142. 3. Bengtsson B, Heijl A (2001) Lack of long-term drift in timolol’s effectiveness in patientswith ocular hypertension. Invest Ophthalmol Vis Sci 42:2839-2842.4. Bergeå B, Bodin L, Svedbergh B (1999) Impact of intraocular pressure regulation on visualfields in open-angle glaucoma. Ophthalmology 106:997-1005.5. Daugeliene L, Yamamoto T, Kitazawa Y (1999) Risk factors for visual field damageprogression in normal-tension glaucoma eyes. Graefes Arch Clin Exp Ophthalmol 237:105-108. 6.. De Vivero C, O ́Brien C, Lanigan L, Hitching R (1994) Diurnal intraocular pressurevariation in low-tension glaucoma. Eye 8:521-523.7. Drance SM (1960) The significance of the diurnal tension variations in normal andglaucomatous eyes. Arch Ophthalmol 64:494:501.8.. Drance SM (1963) Diurnal variation of intraocular pressure in treated glaucoma.Significance in patients with chronic simple glaucoma. Arch Ophthalmol 70:302-311.9. Duke Elder S (1952) The Phasic variations in the ocular tension in primary glaucoma. Am JOphthalmol 35:2-8.10. Gonzalez I, Pablo IE, Pueyo M, Ferrer B, Melcon E, Abecia E, Honrubia FM (1997)Assessment of diurnal tension curve in early glaucoma damage. Int Ophthalmol 20: 113-115. Bengtsson & Heijl: IOP fluctuations and risk for GLVFL16 11. Heijl A, Bengtsson B (2000) Long-term effects of timolol therapy in ocular hypertension adouble-masked, randomised trial. Graefes Arch Clin Exp Ophthalmol 238:877-883. 12. Heijl A, Drance SM, Douglas GR. Automatic perimetry (Competer). (1980) Ability todetect early glaucomatous field defects. Arch Ophthalmol 98: 1560-1563. 13. Heijl A, Drance SM (1981) A clinical comparison of three computerized automaticperimeters in the detection of glaucoma defects. Arch Ophthalmol 99:832-836. 14. Heijl A, Leske MC, Bengtsson B, Hyman L, Bengtsson B, Hussein M (2002) Reductionof intraocular pressure and glaucoma progression. Results from the Early Manifest GlaucomaTrial. Arch Ophthalmol 120:1268-1279.15. Heijl A, Lindgren G, Olsson J (1986) A package for the statistical analysis of visual fields.Doc Ophthalmol Ser 49: 153-168.16. Ishida K, Yamamoto T, Kitazawa Y (1998) Clinical factors associated with progression ofnormal-tension glaucoma. Journal of Glaucoma 7:372-377.17. Kass MA, Heuer DK, Higginbotham EJ, Johnson CA, Keltner JL, Miller JP, Parrish RK,Wilson MR, Gordon MO (2002) The Ocular Hypertension Treatment Study: a randomisedtrial determines that topical ocular hypotensive medication delays or prevents the onset ofprimary open-angle glaucoma. Arch Ophthalmol 120:701-713. 18. Katavisto M (1964) The diurnal variations of ocular tension in glaucoma. ActaOphthalmol 46: Suppl 78: 1-130. 19. Langley D, Swanljung H (1951) Ocular tension in glaucoma simplex. Br J Ophthalmol35:445-458. 20. Liu JHK, Zhang X, Kripke DF, Weinreb RN (2003) twenty-four-hour intra ocularpressure pattern associated with early glaucomatous changes. Invest Ophthalmol Vis Sci44:1586-1590 Bengtsson & Heijl: IOP fluctuations and risk for GLVFL17 21. Maslenikow A, Z (1904) Ueber Tagesschwankungen des intraokularen Druckes beiGlaukom. Augenheilk 11:564, 1904 22. Mills RP, Hopp RH, Drance SM (1986) Comparison of quantitative testing with theOctopus, Humphrey, and Tubingen perimeters. Am J Ophthalmol 102: 496-504. 23. Nouri-Mahdavi K, Hoffman D, Coleman AL, Liu G, Li G, Gaasterland D, Caprioli J(2004) Predictive factors for glaucomatous visual field progression in the AdvancedGlaucoma Intervention Study. Ophthalmology 111: 1627-1635. 24. Saccà SC, Rolando M, Marletta A, Cerqueti P, Ciurlo G (1998) Fluctuations of intraocularpressure during day in open-angle glaucoma, normal-tension glaucoma and normal subjects.Ophthalmologica 212:115-119.25. Smith J (1985) Diurnal intraocular pressure. Correlation to automated perimetry.Ophthalmology 92; 858-861.26. Wax MB, Camras CB, Fiscella RG, Girkin C, Singh K, Weinreb R (2002) Emergingperspectives in glaucoma: optimizing 24.hour control of intraocular pressure. Am JOphthalmol suppl.133:S1-S10