Does Greater Long-Term IOP Variability Increase Probability of Primary Open Angle Glaucoma in the Ocular Hypertension Treatment Study (OHTS)? M.O. Gordon, J.A. Beiser, J. Miller, M. Kass, F. Gao The Ocular Hypertension Treatment Study Group (OHTS) May 3, 2011 ARVO Ft. Lauderdale, FL National Eye Institute, National Center for Minority Health and Health Disparities, NIH grants EY09307, EY09341, Unrestricted Grant from Research to Prevent Blindness, Merck Research Laboratories and Pfizer, Inc.

Sources of Support  NIH grants National Eye InstituteNational Eye Institute National Center for Minority Health and Health Disparities, EY09307, EY09341National Center for Minority Health and Health Disparities, EY09307, EY09341  Unrestricted Grants Research to Prevent BlindnessResearch to Prevent Blindness Merck Research LaboratoriesMerck Research Laboratories Pfizer, Inc.Pfizer, Inc.

Question? Is greater long-term IOP variability an independent factor for the risk of primary open angle glaucoma (POAG) in the OHTS? Is greater long-term IOP variability an independent factor for the risk of primary open angle glaucoma (POAG) in the OHTS?

Patient found eligible for OHTS Eligible untreated IOPs on 2 visits 2 sets of normal & reliable HVFs per VFRC Optic discs judged normal by ODRC Medication Topical treatment to lower IOP 20% and IOP < 24 mm Hg n=817Observation No topical treatment to lower IOP n=819 Randomization Monitoring Humphrey 30-2 q6 months Stereoscopic disc photos annually Adjust therapy if target not met

Analysis Dataset Baseline data: Age, CCT, PSD, VCD Baseline data: Age, CCT, PSD, VCD F/up of >4 visits ( 4 visits (< 3 unstable results) Eye to develop POAG or random eye Eye to develop POAG or random eye Censored data after: Censored data after: Initiation of topical hypotensive medication Eye surgery POAG

IOP Variables POAG eye or random eye Mean f/up IOP: Mean f/up IOP: Baseline & 6 mo. follow-up in Obs. group Standard deviation of f/up IOP Standard deviation of f/up IOP Maximum f/up IOP Maximum f/up IOP Range of f/up IOP Range of f/up IOP (maximum – minimum)

Definition of POAG 3 consecutive reliable, abnormal visual fields and/or 2 consecutive stereo-optic disc photographs showing progression per Reading Center 3 consecutive reliable, abnormal visual fields and/or 2 consecutive stereo-optic disc photographs showing progression per Reading Center Masked Endpoint Committee (DH, EH, RP) adjudicated if change could be attributable to POAG. Masked Endpoint Committee (DH, EH, RP) adjudicated if change could be attributable to POAG. Optic disc progression also had to be “clinically significant” (JND not adequate) Optic disc progression also had to be “clinically significant” (JND not adequate)

Result: Analysis Dataset Median 16 visits Median 16 visits 84% (687 of 819) Observation pts. 84% (687 of 819) Observation pts. (no CCT, censored< 4 visits, < 4 visits) 12% (84 of 687) pts developed POAG in 1 or both eyes 12% (84 of 687) pts developed POAG in 1 or both eyes

Description of F/up IOP Variables N=687 Obs. pts. Mean + SD Mean IOP Mean IOP SD IOP SD IOP Max IOP Max IOP Range IOP Range IOP

Correlation of IOP Variables Mean IOP SDMaxRange Mean IOP SD IOP Max IOP Range IOP-----

Analysis Plan Cox proportional hazards model for each measure of IOP variability Cox proportional hazards model for each measure of IOP variability Univariate Cox proportional hazards model Univariate Cox proportional hazards model Multivariable Cox proportional hazards models: Ag e, CCT, mean IOP, VCD, PSD + Var Multivariable Cox proportional hazards models: Ag e, CCT, mean IOP, VCD, PSD + Var C statistic for predictive accuracy C statistic for predictive accuracy Calibration chi-square for model fit Calibration chi-square for model fit

Quartiles Mean IOP (baseline & f/up) Obs. Group Cox Proportional Hazards Quartiles Mean IOP (baseline & f/up) Obs. Group Cox Proportional Hazards UnivariateMultivariable + Mean IOP Quartiles HR95% CIHR95% CI Q1: [15.5, 21.8)* Mean= N/A1.0N/A Q2: [21.9, 23.8) Mean= Q3: [23.9,26.0) Mean= Q4: [26.0,35.0] Mean= P-Value< C-statistic and Calibration χ and 4.48 *First quartile used as the reference level. +Multi-variable model includes baseline age, CCT, PSD, VCD and mean IOP.

Quartiles IOP SD Obs. group Cox Proportional Hazards UnivariateMultivariable IOP Standard Deviation HR95% CIHR95% CI Q1: [0.67, 1.88)* Mean=1.51.0N/A1.0N/A Q2: [1.89, 2.36) Mean= Q3: [2.37, 2.99) Mean= Q4: [3.0, 9.25] Mean= P-Value C-statistic and Calibration χ and 8.57 *First quartile used as the reference level. +Multi-variable model includes baseline age, CCT, PSD, VCD, mean follow up IOP, and IOP SD.

Quartiles IOP Max Obs. group Cox Proportional Hazards UnivariateMultivariable IOP Maximum HR95% CIHR95% CI Q1: [19.0, 25.6)* Mean= N/A1.0N/A Q2: [25.7, 27.6) Mean= Q3: [27.7, 30.2) Mean= Q4: [30.3, 51.0] Mean= P-Value< C-statistic and Calibration χ and 4.53 *First quartile was used as the reference level +Multi-variable model includes baseline age, CCT, PSD, VCD, mean follow up IOP, and IOP Max.

Quartiles IOP Range Obs. group Cox Proportional Hazards UnivariateMultivariable IOP Range HR95% CIHR95% CI Q1: [2.0, 6.2)* Mean=4.81.0N/A1.0N/A Q2: [6.2, 8.3) Mean= – – 3.8 Q3: [8.3, 10.2) Mean= – – 2.3 Q4: [10.2, 27.3] Mean= – – 2.1 P-Value C-statistic and Calibration χ and 4.74 *First quartile was used as the reference level +Multi-variable model includes baseline age, CCT, PSD, VCD, mean follow up IOP, and IOP range.

Summary (replicated w > 4,5,6 visits in Obs. group) Multivariable P-ValueC Statistic Calibration Chi-Square Mean IOP< SD IOP Max IOP Range IOP

Percent POAG (n=84/687) by Quartile of Mean IOP and SD IOP Interaction of mean IOP * SD IOP is not statistically significant < 21.9 >21.9 to <23.9 > 23.9 to <26.0 > 26.0 <1.9>1.9 to <2.4>2.4 to <3.0 > 3.0 Standard Deviation of f/up IOP 3%4%5% 2%8% 4% 6%19%6%15% 22%24%29%36% 0% Mean f/up IOP (mm Hg)

Conclusions Higher long-term IOP variability (SD) is an independent risk factor for the development of POAG in the Obs. Group in the OHTS Higher long-term IOP variability (SD) is an independent risk factor for the development of POAG in the Obs. Group in the OHTS SD IOP does not improve predictive accuracy of model with age, mean f/up IOP, CCT, VCD and PSD SD IOP does not improve predictive accuracy of model with age, mean f/up IOP, CCT, VCD and PSD SD IOP difficult to measure, uncertain value in clinical practice SD IOP difficult to measure, uncertain value in clinical practice

OHTS Clinical Centers  Bascom Palmer Eye Institute  Baylor Eye Clinic  Charles R. Drew University  Devers Eye Institute  Emory University Eye Center  Eye Associates of Washington, DC  Eye Consultants of Atlanta  Eye Doctors of Washington  Eye Physicians and Surgeons of Atlanta  Glaucoma Care Center  Great Lakes Ophthalmology  Henry Ford Hospitals  Johns Hopkins University  Jules Stein Eye Institute, UCLA  Kellogg Eye Center  Kresge Eye Institute  Krieger Eye Institute  Maryland Center for Eye Care  Mayo Clinic/Foundation  New York Eye & Ear Infirmary  Ohio State University  Salus University  Scheie Eye Institute  University of California, Davis  University of California, San Diego  University of California, San Francisco  University of Louisville  University Suburban Health Center  Washington Eye Physicians & Surgeons  Washington University, St. Louis

OHTS Resource Centers Study Chairman’s Office & Coordinating Center Washington University St. Louis, MO Visual Field Reading Center University of California, Davis Sacramento, CA Optic Disc Reading Center Bascom Palmer Eye Institute University of Miami Miami, FL

QUESTIONS? Thank-you!