This was a prospective study that included patients with primary open-angle glaucoma enrolled in the Vascular Imaging in Glaucoma Study between January 1, 2020, and December 31, 2021. This study was approved by the institutional review board of the University of Miami Miller School of Medicine and adhered to the tenets of the Declaration of Helsinki. All patients signed an informed consent form that was approved by the institutional review board.

All patients underwent 2 baseline visits, 1 month apart, and follow-up visits occurred at 4-month intervals. At each visit, the patients’ medical and ophthalmic history was recorded, including past and current systemic and ocular medications. Subjects underwent a comprehensive ophthalmological exam at each baseline and follow-up visit, including best corrected visual acuity assessment, IOP measurement by Goldman applanation tonometry, slit lamp biomicroscopy, and optic disc exam. RNFL thickness analysis was performed at the baseline and follow-up visits using spectral-domain optical coherence tomography (SDOCT). A 24-hour ABPM was obtained after the first baseline visit, and office-based seated BP measurements were evaluated at each follow-up study visit. Subjects were treated at the discretion of the attending physician.

Glaucomatous eyes showed evidence of glaucomatous optic neuropathy features, including presence of neuroretinal rim narrowing, notching, excavation, disc hemorrhage, or RNFL defects, and repeatable glaucomatous SAP defects (24-2C Swedish Interactive Thresholding Algorithm Faster, Humphrey Field Analyzer 3, Carl Zeiss Meditec) defined as a pattern SD with P value below 5% or glaucoma hemifield test outside 95% normal limits. All patients had at least one confirmatory visual field (VF) examination. Exclusion criteria consisted of corneal or retinal pathology, prior intraocular surgery except for uncomplicated cataract extraction or glaucoma surgery, age <40 or >89 years, best corrected visual acuity worse than 20/40, and less than 1 year of follow-up.

All subjects underwent SDOCT imaging at baseline and 4-month follow-up intervals. Imaging of the RNFL and ONH were captured using Cirrus HD-OCT 6000 (software version 11.5.2.54532, Carl Zeiss Meditec, Dublin, CA). This spectral-domain platform enables assessment of peripapillary RNFL thickness and topographic analysis of the ONH as previously described. ^, A peripapillary scan centered on the optic disc (200 × 200 pixel optic disc cube) was acquired from both eyes of each subject by trained technicians prior to administration of mydriatic agents. All SDOCT images were reviewed to assess quality, and images with signal strength ≤6, poor centration, motion or shadowing artifact, or segmentation errors were excluded.

All included subjects underwent 24-hour ABPM in the habitual posture (sitting during the day and supine at night) after the first baseline visit. The Welch Allyn monitor (ABPM 7100 Ambulatory Blood Pressure Monitor, I.E.M. GmbH Cockerillstr) was used to perform the 24-hour BP assessment. A physician discussed the ABPM instructions with the patient, explained the possible hazards in detail, and ensured that the patient understood all functions and observable points before a 24-hour measurement. An initial measurement demonstration was done at the clinic office to check the values of the first measurement for plausibility and to ensure correct cuff position, so that subsequent automatic measurements could be processed correctly. Ambulatory 24-hour systolic, diastolic, and heart rate were recorded every 30 minutes during the day and every 60 minutes at night, and the average, maximum, minimum, range, and SD of daytime and nighttime systolic and diastolic BPs for each patient were calculated. Diurnal BP measurements were defined as recordings from 8 am to 7:5 9 pm (hours 08:00-19:59); nocturnal measurements from 8 pm to 7:5 9 am (hours 20:00-07:59).

At each follow-up visit thereafter, BP measurements were obtained from the brachial artery in a sitting upright position by trained personnel. Subjects were allowed to rest for at least 5 minutes before each measurement. The systolic BP (SBP), diastolic BP (DBP), and MAP were recorded, and the mean, peak, and lowest values over the study period for each parameter were calculated for each patient and eye and used in the analysis.

Rates of glaucomatous RNFL progression were evaluated by measuring the change in global RNFL thickness using linear mixed models (LMM), a previously described technique used to evaluate longitudinal disease metrics. LMM analyses account for nonindependence between longitudinal data points and permit the inclusion of both eyes from the same subject. Random slopes and intercepts are introduced to account for subgroup correlations between data points. Best linear unbiased predictions are subsequently used to estimate the slopes of this rate of progression. The rates of RFNL change were calculated for each eye and reported as µm/y. Power and sample analysis using reference data obtained from previously published large longitudinal studies evaluating the rates of progression in patients with glaucoma indicated that a sample size of 50 eyes would achieve 80% power to identify differences between groups of up to 0.30 µm/y using slopes derived from LMMs.

Statistical analyses were performed using commercial software Stata version 18 (StataCorp LP) on subjects that underwent baseline 24‑hour ABPM and had at least 5 OCT examinations over ≥2 years of follow‑up with high-quality SDOCT images with signal strength >6. Categorical variables were reported as absolute numbers and percentages. Mean ± standard deviation (SD) was reported for normally distributed variables and median ± range for metric that are not normally distributed.

Univariable LMM regressions were used to evaluate the effect of each BP parameter on the rates of RNFL loss over time. Multivariable models were subsequently utilized to evaluate the same relationships between BP and RNFL loss, while adjusting for potential confounding factors. Due to collinearity among the BP metrics, separate multivariable models were derived, each adjusted for age, sex (assigned at birth), self-reported race, baseline RNFL, central corneal thickness, and follow-up IOP. The estimates from the multivariable LMM were also used to predict RNFL trajectories across a range of systemic BP levels. Mean values were assumed for the remaining clinical characteristics when calculating the trajectories, which are presented with 95% confidence intervals (CI).