ORIGINAL STUDY

Predictive Value of the Water Drinking Test and the Risk of Glaucomatous Progression

Carlos G. De Moraes, MD, MPH,*w Remo Susanna, Jr, MD,w Lisandro M. Sakata, MD, PhD,z and Marcelo Hatanaka, MD, PhDw

Key Words: , water drinking test, visual fields, progression, Purpose: The purpose of this is to test the hypothesis the intra- risk factors ocular pressure (IOP) peaks during a stress test [the water drinking test (WDT)] can estimate the risk of future visual field progression (J Glaucoma 2017;26:767–773) in treated primary open-angle glaucoma (POAG) patients. Patients and Methods: Design: Prospective, longitudinal study. Setting: Clinical practice. Study population: Treated POAG he goal of glaucoma therapy is to prevent or minimize patients whose IOP was r18 mm Hg and who had no IOP- Tthe decay of vision-related quality of life during lowering interventions between the date of the WDT and the last patients’ lifespan.1 The irreversibility of glaucomatous eligible visual field. Intervention: At baseline examination, patients damage argues for estimation of progression before it underwent the WDT and were then followed at regular intervals occurs, rather than after, to modify the treatment approach with office-based IOP measurements and visual field testing. Main before progression. In addition, the more damaged the outcome measure: Cox-proportional hazards survival analysis testing the predictive value of IOP peaks during the WDT versus , the lower the intraocular pressure (IOP) IOP measurements during office hours on visual field progression. required to slow progression and the higher the risk of blindness.2 A number of studies have investigated the main Results: A total of 144 of 96 patients with baseline visual field risk factors for future glaucomatous visual field pro- damage ranging from mild to severe followed for a mean of 28 gression, of which the major randomized clinical trials have months were analyzed. In the multivariable analysis adjusting for played a major role.3–6 Although their results were not potential confounders, higher IOP peaks during the WDT were 7–9 10,11 predictive of future visual field progression (hazard ratio = 1.11; always identical, reviews and consensus guidelines in 95% confidence interval, 1.02 to 1.21; P = 0.013). The average and open-angle glaucoma have underscored the importance of peak IOP during office hours over the same follow-up period were elevated IOP, decreased central corneal thickness, older not significantly associated with progression (P = 0.651 and 0.569, age, worse baseline damage, disc hemorrhages, and low respectively). systemic blood pressure, among others. Nonetheless, with the advent of new technologies and Conclusions: IOP peaks detected with the WDT were predictive of a better understanding of the pathogenesis of glaucoma, future visual field progression in a treated POAG population. This 12 stress test could be a useful tool for risk assessment in daily new emerging risk factors have been reported and may practice. aid clinicians when stratifying patients based upon their risk. Some of these emerging risk factors (eg, corneal hys- teresis; lamina cribrosa defects seen with high-resolution optical coherence tomography (OCT); 24-hour IOP-related measurements with a contact lens sensor)12 require tech- Received for publication December 20, 2016; accepted June 12, 2017. From the *Bernard and Shirlee Brown Glaucoma Research Labo- nologies not always available in clinical practice or have not ratory, Department of , Columbia University yet been evaluated with sufficient scientific thoroughness to Medical Center, Edward S. Harkness Institute, New York, NY; define a level of evidence comparable with those reported in w Department of Ophthalmology, University of Sao Paulo School of clinical trials. Medicine, Avenue Dr. Ene´as de Carvalho Aguiar, Sa˜ o Paulo, SP; and zDepartment of Ophthalmology, Federal University of Parana Although there is a high-level of evidence that elevated School of Medicine, Rua Gen. Carneiro, Curitiba, PR, Brazil. mean IOP is a risk factor for glaucoma onset and pro- Supported by unrestricted departmental grant from Research to Pre- gression, such evidence is based upon longitudinal collec- vent Blindness, New York, NY (Department of Ophthalmology, tion of IOP measurements over months or years,3–9 which Columbia University Medical Center) and the Research Fund of the Department of Ophthalmology, University of Sao Paulo School can be influenced by the interval between visits and the of Medicine, Sao Paulo, Brazil. number of IOP measurements. Longitudinal studies and C.G.D.M.: Grants—National Institutes of Health, National Eye clinical trials have failed to show that a single baseline IOP Institute, Bethesda, MD; Consultant—Sensimed, AG, Lausanne, measurement can differentiate treated glaucoma patients at Switzerland; Honoraria, Allergan Inc., Sao Paulo, Brazil; L.M.S.: 13–16 Consultant—Allergan Inc., Sao Paulo, Brazil; Mundipharma Inc., low-risk versus high-risk of future progression. This is Sao Paulo, Brazil; M.H.: Consultant—Allergan Inc., Sao Paulo, in part because of the high IOP variability within-days and Brazil; Mundipharma Inc., Sao Paulo, Brazil; Alcon Inc., Sao between-days, so that there is an equal chance of detecting Paulo, Brazil; R.S.: Consultant—Allergan Inc., Sao Paulo, Brazil; the IOP at its peak or trough with a single snap-shot Mundipharma Inc., Sao Paulo, Brazil; Alcon Inc., Sao Paulo, Brazil. measurement. This limitation ultimately affects the pre- Reprints: Carlos G. De Moraes, MD, MPH, Edward S. Harkness dictive value of single IOP measurements when assessing Eye Institute, Columbia University Medical Center, 635 West the risk visual field progression. 165th Street, Box 69, New York, NY 10032 In an attempt to overcome the limitations of single- (e-mail: [email protected]). Copyright r 2017 Wolters Kluwer Health, Inc. All rights reserved. visit IOP measurements, numerous studies have inves- DOI: 10.1097/IJG.0000000000000725 tigated the role of within-day IOP variability on the risk of

J Glaucoma Volume 26, Number 9, September 2017 www.glaucomajournal.com | 767 Copyright r 2017 Wolters Kluwer Health, Inc. All rights reserved. De Moraes et al J Glaucoma Volume 26, Number 9, September 2017 future progression.17–20 In that context, studies investigat- hemifield test outside normal limits. These results had to be ing IOP stress tests have been conducted to evaluate confirmed on at least 2 consecutive examinations. All their repeatability and validity, a first step towards defining included eyes had a best-corrected visual acuity of 20/40 or their role as predictors of glaucoma progression.21,22 In better during the entire period of the study, spherical other words, by challenging the eye’s outflow facility, stress refraction within ± 5.00 diopters, and cylinder correction tests may reveal IOP peaks that could be predictive of within ± 3.00 diopters. future visual field progression. Similar approaches have Participants were excluded if they had a closed or nar- been employed in different fields of medicine. In cardiology, row angle upon gonioscopic examination, nonglaucomatous for instance, even though a single blood pressure meas- optic neuropathy, secondary glaucoma, retinal disease, or urement at rest may not be indicative of ischemia and the any other abnormality that could interfere with visual future risk of myocardial infarction, when undergoing a field testing. stress test (pharmacologically or through exercise) clini- All study patients were on antiglaucoma topical cians can detect signs that allow a more accurate estimation medications or had undergone laser or of future events.23 The same is true in diseases like diabetes before enrollment. All IOP measurements (ie, glucose tolerance test)24 and asthma (ie, methacholine were r18 mm Hg based upon office-based measurements challenge test).25 during follow-up. IOP measurements were performed with One such IOP stress test, the water drinking test a calibrated Goldmann applanation tonometer (Haag- (WDT), has shown good correlation and agreement with Streit, GmbH, Switzerland). The results of the WDT physiological IOP peaks seen in treated and untreated performed at the beginning of follow-up—before the first open-angle glaucoma patients.26–28 In addition, in patients visual field test of the sequence evaluated for progression— with bilateral glaucoma, eyes with higher IOP peaks during were considered for analyses (herewith called the “Baseline the WDT (despite similar baseline measurements) have WDT”). Only office-hour IOP measurements and visual worse visual field damage than their fellow eyes.29 Given field information up to the date of confirmed visual field the lack of large longitudinal studies assessing the role of progression and/or IOP-lowering interventions (eg, adding the WDT as a predictor of visual field progression, the new drugs or changing medications, laser trabeculoplasty, recent consensus of the World Glaucoma Association or incisional ) were included in the anal- stated that “there is insufficient evidence supporting the role yses. All data after this point in time were censored. of provocative tests, such as the WDT, as providing inde- The WDT consisted in obtaining a baseline IOP meas- pendent contribution to assess risk of glaucoma develop- urement, followed by the ingestion of 800 mL (27 ounces) of ment and progression” and that “prospective studies are potable water in 5 minutes. IOP was then measured 3 times at necessary to clarify whether the water-drinking provocative 15-minute intervals by the same examiner, with the same test can provide additional information over office-based tonometer. The maximum value of the 3 measurements IOP measurements in establishing risk of glaucoma devel- defined the IOP peak (herewith called “WDT IOP peak”). To opment or progression.”10 Nonetheless, Goldberg and minimize the effect of circadian IOP variation, all WDTs were Clement30 recently discussed in an editorial the usefulness performed between 4 and 5 pm. of the WDT to assess IOP fluctuation and to detect the Standard achromatic perimetry was performed with circadian IOP peak, as well as to assess the efficacy of the Humphrey Visual Field Analyzer (24-2 SITA-Standard; treatment. Carl Zeiss Meditec Inc., Dublin, CA). All patients were In the present prospective, longitudinal study, we experienced with visual field testing and only reliable exams investigated the predictive value of the results of the WDT (< 20% fixation losses, <33% false-positive, and false- in assessing the risk of glaucoma progression when com- negative rates) were analyzed. The last reliable visual field pared with office-based IOP measurements in a population test performed before the WDT was considered as the with established and treated open-angle glaucoma. baseline examination. Visual field progression was deter- mined based upon established progression criteria,31 as defined by the development of a new defect or worsening of METHODOLOGY a previous defect. A new defect was defined by the presence The study protocol adhered to the tenets of the Dec- of a new cluster of Z3 nonedge points with sensitivities at laration of Helsinki and was approved by the local Com- P < 5% on the pattern deviation plot with at least 1 of the mittee of Ethics. Study participants are part of a larger points at P < 1%. Worsening of a previously damaged cohort of patients with primary open-angle glaucoma who region was defined based upon the deterioration of Z3 have been undergoing the WDT at all office visits in a points in that region by at least 10 dB. Progression had to glaucoma referral practice. Consecutive patients meeting be confirmed in a subsequent test performed within 1 to the inclusion and exclusion criteria described below were 2 months after progression was suspected. selected for the present study. All participants signed a The endpoint for eyes with visual field progression was written informed consent to participate in the study. considered the day of the first visual field suggesting pro- Primary open-angle glaucoma patients were included gression. For patients without visual field progression all if they had a glaucomatous appearing optic disc—as visual field data collected until their last examination were defined by a senior fellowship trained glaucoma specialist analyzed. (R.S.J.)—associated with glaucomatous visual field loss seen on 24-2 standard automated perimetry. Visual field Statistical Analysis loss was defined according to the modified Anderson’s Continuous data are presented as mean, SD, and criteria: (i) the presence of a cluster of Z3 nonedge points range whereas categorical variables are described as pro- with sensitivities at P < 5% on the pattern deviation plot, portions (%). The main outcome measure was the associ- with at least 1 point at P < 1%, or (ii) the pattern standard ation between IOP parameters as independent variables deviation (PSD) value at P < 5%, or (iii) the glaucoma and visual field progression as dependent variable. The

768 | www.glaucomajournal.com Copyright r 2017 Wolters Kluwer Health, Inc. All rights reserved. Copyright r 2017 Wolters Kluwer Health, Inc. All rights reserved. J Glaucoma Volume 26, Number 9, September 2017 Water Drinking Test and Visual Field Progression following IOP parameters were investigated: WDT IOP peak, the average and peak IOP from office-based meas- TABLE 1. Demographic Characteristics of the Study Sample urements during the follow-up period (herewith called Clinical Parameters Mean (SD) [Range] “office mean IOP” and “office peak IOP,” respectively). Age (y) 67.8 (10.9) [39-88] Risk factors for glaucoma progression were tested with Sex (male) 53%* Cox-proportional hazards regression with visual field pro- Race (white) 88%* gression as the binary outcome. A robust variance estimate CCT (mm) 534.2 (37.2) [412-635] that adjusts for within-cluster correlations was used to No. visual fields 5.4 (2.8) [4-16] estimate standard errors taking into account the correlation Follow-up time (months) 28.1 (15.8) [6-83] between fellow eyes within the same subject. First, we tested No. antiglaucoma medications 2.3 (1.2) [0-4] w whether the proportional hazards assumption held true Prostaglandin analogs 73.6% with the method of Schoenfield’s residuals, which was Baseline MD (dB) 8.8 (6.4) [28.9 to 1.0] Final MD (dB) 9.6 (6.4) [27.2 to 0.1] nonsignificant (P = 0.9981), suggesting that the assumption WDT baseline IOP (mm Hg) 12.7 (2.4) [7-17] was not violated. The following potential confounding WDT peak IOP (mm Hg) 15.8 (3.2) [10-23] variables were also investigated: age, central corneal Office mean IOP (mm Hg) 13.1 (2.6) [6-16] thickness, baseline mean deviation (MD), number of anti- Office peak IOP (mm Hg) 14.0 (2.5) [8-18] glaucomatous drugs, and use of prostaglandins. First, univariable models were performed with each predictor and *Proportion of patients. wProportion of eyes. only those with P < 0.20 were then tested in the multi- CCT indicates central corneal thickness; IOP, intraocular pressure; MD, variable analysis. Tests of association between variables mean deviation of the 24-2 visual field test; WDT, water drinking test. were conducted with mixed-effects linear models taking into account random effects associated with the inclusion of both eyes of the same patient. progress during follow-up (log-rank test, P = 0.014, Fig. 1). Measures of goodness-of-fit for each model were Given the strong correlation between baseline and peak computed with Bayesian (BIC) and Akaike information IOP, we did not include both parameters in the same model criteria (AIC). When fitting models, it is possible to increase because of collinearity. Instead, we ran a model that the goodness-of-fit by adding parameters, but doing so may included the interaction term “BaselineIOP*PeakIOP,” result in overfitting. Both BIC and AIC address this which tested whether the role of peak IOP on progression problem by introducing a penalty term for the number of depended upon the starting baseline IOP. This interaction parameters in the model. The model with the lowest BIC or term was not significant (P = 0.997). In addition, we cal- AIC values suggests better goodness-of-fit. Statistical culated the predictive value of IOP fluctuation during the analyses were performed with commercial software WDT (defined as ‘Peak IOP’—‘Baseline IOP’) on future (STATA, version 14; StataCorp LP, College Station, TX). visual field progression. Despite statistical significance in Statistical significance was defined at P < 5%. the univariable analysis (HR = 1.12; 95% CI, 1.01 to 1.26; P = 0.039), it lost significance in the multivariable analysis (HR = 1.10; 95% CI, 0.99 to 1.23; P = 0.074). RESULTS When investigating the predictive value of office-based A total of 144 eyes of 96 patients were analyzed. Most mean IOP, this parameter was not significantly associated patients were men (53%) and white (88%). Based upon with visual field progression (P = 0.651) (Table 3). Similar their baseline visual field MD, 54 eyes (37.5%) had early results were found when testing the predictive value of glaucoma (MD better than 5 dB); 45 eyes (31.2%) had office-based peak IOP (P = 0.569) (Table 4). Goodness-of- moderate glaucoma (MD between 5 and 12 dB), and 45 fit statistics for each model are shown in Table 5. Overall, eyes (31.2%) had advanced glaucoma (MD worse than the model testing WDT peaks had a better performance 12 dB). Patients underwent a mean of 5 ± 2 (range, 4 to than those testing office-based IOP measurements. 16) visual field tests spanning 28 ± 15 (6 to 83) months. Summary statistics are shown in Table 1. Forty-seven eyes (32.6%) reached a visual field progression endpoint. As DISCUSSION expected, the rate of visual field MD change (ie, MD slopes) In this longitudinal study, we tested the hypothesis between progressing vs. stable eyes was statistically sig- that WDT peaks can predict the risk of glaucomatous vis- nificant [difference b = 0.10 dB/yr, 95% confidence ual field progression in a population with established and interval (CI) = 0.14 to 0.06; P < 0.001; mixed-effects treated open-angle glaucoma with office-based IOP meas- linear model]. urements r18 mm Hg. We also compared its predictive Eyes with worse visual field MD tended to be on more value with those from office-based IOP measurements, IOP-lowering medications (b = 0.86; 95% CI, 1.74 to namely mean and peak IOP. We found that the WDT peak 0.01; P = 0.053) and were more likely to be on prosta- was an independent predictor of progression whereas office- glandins (b = 0.75; 95% CI, 3.14 to 1.63; P = 0.535) based IOP measurements failed to show a significant although it did not reach statistical significance. There was association with visual field progression. Of note, each a significant correlation between baseline and peak IOP mm Hg higher WDT peak at baseline increased the risk of during the WDT (r = 0.68; P < 0.001). progression by 11% (Table 2). These findings support the When investigating the WDT peak as a predictor, utility of the WDT as a useful method to assess the risk of multivariable analysis revealed that there was a significant glaucomatous progression in patients with treated primary predictive value on future progression [hazard ratio open-angle glaucoma. (HR) = 1.11, 95% CI = 1.02 to 1.21, P = 0.013] (Table 2). The WDT was previously described as a test to diag- When comparing eyes with WDT peaks Z versus nose open-angle glaucoma, although it was later aban- <18 mm Hg, the group with higher IOP was more likely to doned because of high false-positive and false-negative

Copyright r 2017 Wolters Kluwer Health, Inc. All rights reserved. www.glaucomajournal.com | 769 Copyright r 2017 Wolters Kluwer Health, Inc. All rights reserved. De Moraes et al J Glaucoma Volume 26, Number 9, September 2017

TABLE 2. Univariable and Multivariable Cox-Proportional Hazards Models Testing the Association Between Water Drinking Test Parameters, Potential Confounders, and Visual Field Progression Univariable Multivariable HR 95% CI P HR 95% CI P Age (y) 0.99 0.96-1.03 0.804 CCT (mm) 0.99 0.98-1.00 0.700 No. medications 1.27 0.95-1.70 0.103 1.17 0.90-1.51 0.222 Use of prostaglandins 1.05 0.52-2.12 0.885 Baseline visual field MD (dB) 0.96 0.91-1.00 0.104 0.95 0.91-1.00 0.083 WDT peak IOP (mm Hg) 1.12 1.02-1.22 0.012 1.11 1.02-1.21 0.013 WDT IOP fluctuation (mm Hg)* 1.12 1.01-1.26 0.039

Note: IOP fluctuation was not entered in the multivariable model because of collinearity with Peak IOP. *Defined as (Peak IOP—Baseline IOP) during the water drinking test. CCT indicates central corneal thickness; CI, confidence intervals; HR, hazard ratio; IOP, intraocular pressure; MD, mean deviation; WDT, water drinking test.

rates.32,33 More recently, this test has been proposed as a glaucoma patients followed for a mean period of 2 years, method to: (i) evaluate the eye’s aqueous humor outflow the authors found that the mean peak IOP during the WDT facility, (ii) detect IOP variability, and (iii) estimate 24-hour was significantly higher in patients with prior visual field IOP peaks. Previous cross-sectional and retrospective progression compared with patients who had not pro- studies suggested that IOP peaks during the WDT could be gressed.34 Note that although the present study confirms an indicator for the likelihood of progression.29,34 In 1993, the findings of this previous report, which included a dif- Yoshikawa et al35 investigated the predictive value of three ferent sample with a retrospective study design,34 in the different parameters (outflow facility, WDT response, and present study we helped improve the level of evidence of diurnal changes in IOP) in 54 normal-tension glaucoma those findings in a prospective design. This was suggested patients followed for 3 to 7 years. The authors found that by the World Glaucoma Association10 as a necessary step the single most indicative predictor of visual field pro- to further validate the use of the WDT in clinical practice. gression was the maximum IOP levels after the WDT. Over Finally, to date, there have been no cross-sectional or a decade later, Susanna et al29 investigated the relationship longitudinal published articles contesting the results of any between asymmetric glaucomatous visual field damage and of the aforementioned studies. WDT response in patients with bilateral damage and found In addition to a significant predictive value, a clinically that despite similar baseline IOP, fellow eyes with worse useful stress test should demonstrate good repeatability. In visual field MD had higher IOP peaks after the WDT, other words, if the test is done on different days and granted which may reflect the status of the drainage system of the no changes in treatment or adherence occurred, it is eye. In a retrospective analysis of 76 eyes of open-angle important that the test results are repeatable. Babic et al22

FIGURE 1. Kaplan-Meier survival estimates comparing patients with water drinking test peak IOP Z18 mm Hg versus <18 mm Hg. Log-rank test revealed a statistically significant difference (P = 0.014).

770 | www.glaucomajournal.com Copyright r 2017 Wolters Kluwer Health, Inc. All rights reserved. Copyright r 2017 Wolters Kluwer Health, Inc. All rights reserved. J Glaucoma Volume 26, Number 9, September 2017 Water Drinking Test and Visual Field Progression

TABLE 3. Multivariable Cox-Proportional Hazards Model Testing TABLE 5. Goodness-of-fit Statistics of the Cox-Proportional the Association between Office-based Mean IOP, Confounders, Hazards Models in Tables 2–4 and Visual Field Progression BIC AIC Multivariable WDT peak IOP 388.86 379.95 HR 95% CI P Mean office IOP 395.20 386.29 Highest office IOP 395.01 386.10 Number of medications 1.22 0.92-1.63 0.165 Baseline visual field MD (dB) 0.96 0.91-1.00 0.115 AIC indicates Akaike information criteria; BIC, Bayesian information Office mean IOP (mm Hg) 1.02 0.90-1.16 0.651 criteria; IOP, intraocular pressure.

CI indicates confidence intervals; HR, hazard ratio; IOP, intraocular pressure; MD, mean deviation. of medications and worse baseline MD being associated with increased risk of progression (Table 2). Studies have shown showed this to be true regarding the WDT is a population that multiple dosing and complicated dosing regimens are with treated open-angle glaucoma. All patients had the associated with poorer adherence to glaucoma therapy,46,47 WDT test performed twice, 3 to 6 months apart, during which could result in increased rates of progression.48 In which no changes in therapy occurred. They found that addition, more complex medical regimens are more common IOP peaks during the test had excellent repeatability in patients with more severe damage, which are at greater risk (intraclass correlation coefficient, 0.85). Another group of progression.2 recently confirmed these results in patients with exfoliation 36 Our study has limitations, namely the fact that this was syndrome and exfoliative glaucoma, a subtype of glau- not a randomized clinical trial in which patients would coma known to have higher IOP variability than primary 37 (ideally) be assigned to either being monitored with WDT open-angle glaucoma. Given the repeatability of the measurements or just office-hour IOP measurements. We WDT peaks, numerous investigators have employed the minimized selection bias by selecting consecutive patients test to evaluate treatment efficacy with medical,38–40 41,42 43,44 from a cohort in which all patients have undergone the WDT laser, and surgical interventions. routinely at all office visits, that is, there was no a priori It is noteworthy that in the present study, there was a determination of who would/would not undergo the test. To significant correlation between baseline and peak IOP underscore the usefulness of the WDT in treated patients, we measurements during the WDT, that is, eyes with higher included a cohort of primary open-angle glaucoma patients baseline pressure were more likely to have higher peaks. who were deemed well-controlled (all office-based IOP This seems like an apparent contradiction with the studies measurements r18 mm Hg) between the time of the baseline above reporting an excellent repeatability of the test, WDT and the last visual field analyzed and who underwent regardless of baseline IOP. One should notice, however, no treatment changes in this period. The decision to include that these are 2 different concepts: (1) in the repeatability such population hence limits the generalizability of our studies, patients were tested on different days and the out- findings. However, the study sample, in fact, depicts a group come measure was the repeatability of the WDT peaks of patients that could benefit the most by a test that helps within-patients between-days. As described above, despite predict glaucoma progression (ie, patients with reasonably undergoing the WDT on different days (with different controlled IOP during office hours). In addition, we did not baseline IOP each time), the peak IOP was very repeatable. assess adherence in this period, one of the reasons being that, In the present study (2), we assessed the relationship to date, there is no reference standard to identify nonadherent between baseline and peak IOP between-patients on a single patients accurately. Finally, unlike major glaucoma clinical baseline visit. Different patients (or eyes) with different trials that found an association between mean follow-up IOP levels of damage, on different medications, with variable and progression in patients with established glaucoma [ie, the adherence are expected to have different IOP measurements Early manifest Glaucoma Trial (EMGT)4 and the United under steady-state conditions. Therefore, those with worse Kingdom Treatment Study (UKGTS)49] our study did not disease and poorer IOP control who started with a higher find a significant effect of office-based IOP on the risk of baseline measurement were expected to have higher peak 30,34,45 progression. This could be in part because of our inclusion/ IOP during a stress test. exclusion criteria which had a ceiling IOP of 18 mm Hg Another interesting finding was that the univariable and during office hours. As reported in the Advanced Glaucoma multivariable models showed a trend for an increased number Intervention Study (AGIS), eyes with 100% of visits with IOP <18 mm Hg over 6 years had mean changes from baseline in visual field defect score close to 0 during follow- TABLE 4. Multivariable Cox-Proportional Hazards Model Testing up.50 This lack of association between follow-up IOP and the Association Between Office-Based Highest IOP, Confounders, progression was also reported in the Canadian Glaucoma and Visual Field Progression Study,51 The Low-pressure Glaucoma Treatment Study Multivariable (LoGTS),52 and the Collaborative Initial Glaucoma Treat- ment Study (CIGTS).53 By performing a stress test, however, HR 95% CI P a subclinical impairment of the aqueous outflow becomes No. medications 1.22 0.92-1.61 0.155 more evident and the relationship between IOP and pro- Baseline visual field MD (dB) 0.96 0.91-1.00 0.110 gression (otherwise concealed with office-hour measurements) Office peak IOP (mm Hg) 1.03 0.92-1.14 0.569 may reach statistical significance. A similar explanation has CI indicates confidence intervals; HR, hazard ratio; IOP, intraocular been attributed to the role of stress tests as predictors of pressure; MD, mean deviation. myocardial infarction in treated, asymptomatic patients, for example.54

Copyright r 2017 Wolters Kluwer Health, Inc. All rights reserved. www.glaucomajournal.com | 771 Copyright r 2017 Wolters Kluwer Health, Inc. All rights reserved. De Moraes et al J Glaucoma Volume 26, Number 9, September 2017

In summary, IOP peaks detected during the WDT are 18. De Moraes CG, Jasien JV, Simon-Zoula S, et al. Visual field predictive of future visual field progression in patients with change and 24-hour IOP-related profile with a contact lens treated open-angle glaucoma with office-based IOP meas- sensor in treated glaucoma patients. Ophthalmology. 2016;123: urements equal or <18 mm Hg, whereas the average and 744–753. 19. Bengtsson B, Heijl A. Diurnal IOP fluctuation: not an peak IOP measured over multiple office visits were not independent risk factor for glaucomatous visual field loss in significant. Our study findings warrant further validation in high-risk . Graefes Arch Clin Exp Oph- clinical trials in which the outcomes of patients without thalmol. 2005;243:513–518. treatment or with other levels of IOP at baseline followed 20. Choi J, Kim KH, Jeong J, et al. Circadian fluctuation of mean with the WDT are compared with those of patients fol- ocular perfusion pressure is a consistent risk factor for normal- lowed with office-hour measurements only. Nonetheless, tension glaucoma. Invest Ophthalmol Vis Sci. 2007;48:104–111. the present study helps increase the level of evidence of the 21. Hatanaka M, Sakata LM, Susanna R Jr, et al. Comparison of clinical utility of the WDT as a stress test in the manage- the intraocular pressure variation provoked by postural change ment of primary open-angle glaucoma. and by the water drinking test in primary open-angle glaucoma and normal patients. J Glaucoma. 2016;25:914–918. 22. Babic M, De Moraes CG, Hatanaka M, et al. Reproducibility REFERENCES of the water drinking test in treated glaucomatous patients. 1. Weinreb RN, Khaw PT. Primary open-angle glaucoma. Lancet Clin Experiment Ophthalmol. 2015;43:228–233. (London, England). 2004;363:1711–1720. 23. Morrow DA. Cardiovascular risk prediction in patients with 2. Susanna R Jr, De Moraes CG, Cioffi GA, et al. Why do people stable and unstable coronary heart disease. Circulation. 2010; (Still) go blind from glaucoma? Transl Vis Sci Technol. 121:2681–2691. 2015;4:1–12. 24. Waugh NR, Shyangdan D, Taylor-Phillips S, et al. Screening 3. Gordon MO, Beiser JA, Brandt JD, et al. The Ocular for type 2 diabetes: a short report for the National Screening Hypertension Treatment Study: baseline factors that predict Committee. Health Technol Assess. 2013;17:1–90. the onset of primary open-angle glaucoma. Arch Ophthalmol. 25. Anderson SD. Challenge tests to assess airway hyperrespon- 2002;120:714–720. siveness and efficacy of drugs used in the treatment of asthma. 4. Leske MC, Heijl A, Hyman L, et al. Predictors of long-term J Aerosol Med. 1996;9:95–109. progression in the early manifest glaucoma trial. Ophthalmol- 26. De Moraes CG, Furlanetto RL, Reis AS, et al. Agreement ogy. 2007;114:1965–1972. between stress intraocular pressure and long-term intraocular 5. Nouri-Mahdavi K, Hoffman D, Coleman AL, et al. Predictive pressure measurements in primary open angle glaucoma. Clin factors for glaucomatous visual field progression in the Experiment Ophthalmol. 2009;37:270–274. Advanced Glaucoma Intervention Study. Ophthalmology. 27. Vasconcelos-Moraes CG, Susanna R Jr. Correlation between 2004;111:1627–1635. the water drinking test and modified diurnal tension curve in 6. Drance S, Anderson DR, Schulzer M. Risk factors for untreated glaucomatous eyes. Clinics (Sao Paulo). 2008;63: progression of visual field abnormalities in normal-tension 433–436. glaucoma. Am J Ophthalmol. 2001;131:699–708. 28. Kumar RS, de Guzman MH, Ong PY, et al. Does peak 7. Boland MV, Quigley HA. Risk factors and open-angle intraocular pressure measured by water drinking test reflect glaucoma: classification and application. J Glaucoma. 2007;16: peak circadian levels? A pilot study. Clin Experiment Oph- 406–418. thalmol. 2008;36:312–315. 8. Coleman AL, Miglior S. Risk factors for glaucoma onset and 29. Susanna R Jr, Hatanaka M, Vessani RM, et al. Correlation of progression. Surv Ophthalmol. 2008;53(suppl 1):S3–S10. asymmetric glaucomatous visual field damage and water- 9. Rivera JL, Bell NP, Feldman RM. Risk factors for primary drinking test response. Invest Ophthalmol Vis Sci. 2006;47: open angle glaucoma progression: what we know and what we 641–644. need to know. Curr Opin Ophthalmol. 2008;19:102–106. 30. Goldberg I, Clement CI. The water drinking test. Am J 10. Association WG. Consensus 10: Diagnosis of Primary Open Ophthalmol. 2010;150:447–449. Angle Glaucoma; 2016. Available at: http://www.worldglau 31. Anderson DR. Automated Static Perimetry. St. Louis: Mosby; coma.org/consensus-10-preliminary-report/. Accessed August 1992. 27, 2016. 32. Roth JA. Inadequate diagnostic value of the water-drinking 11. Weinreb RN, Garway-Heath DF, Leung C, et al. Progression test. Br J Ophthalmol. 1974;58:55–61. of Glaucoma: The 8th Consensus Report of the World Glaucoma 33. Miller D. The relationship between diurnal tension variation Association. Amsterdam, The Netherlands: Kugler Publica- and the water-drinking test. Am J Ophthalmol. 1964;58: tions; 2011. 243–246. 12. Blumberg D, Skaat A, Liebmann JM. Emerging risk factors 34. Susanna R Jr, Vessani RM, Sakata L, et al. The relation for glaucoma onset and progression. Prog Brain Res. 2015; between intraocular pressure peak in the water drinking test 221:81–101. and visual field progression in glaucoma. Br J Ophthalmol. 13. De Moraes CG, Juthani VJ, Liebmann JM, et al. Risk factors 2005;89:1298–1301. for visual field progression in treated glaucoma. Arch 35. Yoshikawa K, Inoue T, Inoue Y. Normal tension glaucoma: Ophthalmol. 2011;129:562–568. the value of predictive tests. Acta Ophthalmol (Copenh). 14. Musch DC, Gillespie BW, Lichter PR, et al. Visual field 1993;71:463–470. progression in the collaborative initial Glaucoma treatment 36. Ozyol E, Ozyol P, Karalezli A. Reproducibility of the water- study the impact of treatment and other baseline factors. drinking test in patients with exfoliation syndrome and Ophthalmology. 2009;116:200–207. exfoliative glaucoma. Acta Ophthalmol. 2016;94:e795–e798. 15. De Moraes CG, Liebmann JM, Greenfield DS, et al. Risk 37. Ritch R. The management of exfoliative glaucoma. Prog Brain factors for visual field progression in the low-pressure Res. 2008;173:211–224. glaucoma treatment study. Am J Ophthalmol. 2012;154: 38. Ozyol P, Ozyol E, Baldemir E. Intraocular pressure dynamics 702–711. with prostaglandin analogs: a clinical application of the water- 16. Chauhan BC, Mikelberg FS, Balaszi AG, et al. Canadian drinking test. Clin Ophthalmol. 2016;10:1351–1356. glaucoma study: 2. risk factors for the progression of open- 39. Susanna R Jr, Sheu WP. Comparison of latanoprost with angle glaucoma. Arch Ophthalmol. 2008;126:1030–1036. fixed-combination dorzolamide and timolol in adult patients 17. Asrani S, Zeimer R, Wilensky J, et al. Large diurnal with elevated intraocular pressure: an eight-week, randomized, fluctuations in intraocular pressure are an independent risk open-label, parallel-group, multicenter study in Latin America. factor in patients with glaucoma. J Glaucoma. 2000;9:134–142. Clin Ther. 2004;26:755–768.

772 | www.glaucomajournal.com Copyright r 2017 Wolters Kluwer Health, Inc. All rights reserved. Copyright r 2017 Wolters Kluwer Health, Inc. All rights reserved. J Glaucoma Volume 26, Number 9, September 2017 Water Drinking Test and Visual Field Progression

40. Hatanaka M, Grigera DE, Barbosa WL, et al. An eight-week, 47. Schwartz GF. Compliance and persistency in glaucoma follow- multicentric, randomized, interventional, open-label, phase 4, up treatment. Curr Opin Ophthalmol. 2005;16:114–121. parallel comparison of the efficacy and tolerability of the fixed 48. Rossi GC, Pasinetti GM, Scudeller L, et al. Do adherence rates combination of timolol maleate 0.5%/brimonidine tartrate and glaucomatous visual field progression correlate? Eur J 0.2% versus fixed combination of timolol maleate 0.5%/ Ophthalmol. 2011;21:410–414. dorzolamide 2% in patients with elevated intraocular pressure. 49. Garway-Heath DF, Crabb DP, Bunce C, et al. Latanoprost for J Glaucoma. 2008;17:674–679. open-angle glaucoma (UKGTS): a randomised, multicentre, 41. Waisbourd M, Savant SV, Sun Y, et al. Water-drinking test in placebo-controlled trial. Lancet. 2015;385:1295–1304. primary angle-closure suspect before and after laser peripheral 50. Investigators A. The Advanced Glaucoma Intervention Study (AGIS): 7. The relationship between control of intraocular iridotomy. Clin Experiment Ophthalmol. 2016;44:89–94. pressure and visual field deterioration.The AGIS Investigators. 42. Kerr NM, Lew HR, Skalicky SE. Selective laser trabeculo- Am J Ophthalmol. 2000;130:429–440. plasty reduces intraocular pressure peak in response to the 51. Chauhan BC, Mikelberg FS, Artes PH, et al. Canadian water drinking test. J Glaucoma. 2016;25:727–731. Glaucoma Study: 3. Impact of risk factors and intraocular 43. Medeiros FA, Pinheiro A, Moura FC, et al. Intraocular pressure pressure reduction on the rates of visual field change. Arch fluctuations in medical versus surgically treated glaucomatous Ophthalmol. 2010;128:1249–1255. patients. J Ocul Pharmacol Ther. 2002;18:489–498. 52. Musch DC, Gillespie BW, Niziol LM, et al. Intraocular 44. Mansouri K, Orguel S, Mermoud A, et al. Quality of diurnal pressure control and long-term visual field loss in the intraocular pressure control in primary open-angle patients treated Collaborative Initial Glaucoma Treatment Study. Ophthalmol- with latanoprost compared with surgically treated glaucoma ogy. 2011;118:1766–1773. patients: a prospective trial. Br J Ophthalmol. 2008;92:332–336. 53. Blumenthal DS, Weiss JL, Mellits ED, et al. The predictive 45. Mocan MC, Kasim B, Muz E, et al. Intraocular pressure value of a strongly positive stress test in patients with minimal characteristics of exfoliative glaucoma and exfoliation syn- symptoms. Am J Med. 1981;70:1005–1010. drome as determined with the water drinking test. J Glaucoma. 54. Currie PJ, Kelly MJ, Harper RW, et al. Incremental value of 2016;25:301–305. clinical assessment, supine exercise electrocardiography, and 46. Tsai JC. A comprehensive perspective on patient adherence to biplane exercise radionuclide ventriculography in the predic- topical glaucoma therapy. Ophthalmology. 2009;116(suppl 11): tion of coronary artery disease in men with chest pain. Am J S30–S36. Cardiol. 1983;52:927–935.

Copyright r 2017 Wolters Kluwer Health, Inc. All rights reserved. www.glaucomajournal.com | 773 Copyright r 2017 Wolters Kluwer Health, Inc. All rights reserved.