Intraocular pressure correlation to progressive retinal nerve fiber layer loss in primary open angle glaucoma as measured by standard and modified goldmann applanation tonometers

Abstract

Purpose

Characterize the relationship between intraocular pressure (IOP) as measured by standard and modified Goldmann prisms and the progressive loss of retinal nerve fiber layer (RNFL) in a cohort of glaucoma patients.

Design

Retrospective cross-sectional cohort data analysis.

Participants

The study included all patients from a database of 1927 eyes, 966 patients with same visit sequential standard and modified Goldmann IOP measurements. From the database, 148 eyes, 75 patients met the inclusion criteria of a diagnosis of primary open angle glaucoma (POAG) with at least 5 sequential quality optical coherence tomographer (OCT) measurements.

Methods

Sequential OCT images were obtained with the spectral domain Zeiss OCT5000. Participants were all diagnosed with POAG by untreated IOP ≥ 22, disk changes, and visual field (HVF) loss consistent with glaucomatous optic neuropathy (GON). Included were 575 Goldmann IOP measurements with standard and modified prisms affixed to the Goldmann applanation tonometer (GAT) armature. A modified prism includes a corneal conforming applanation surface minimizing the cornea’s contribution to the IOP measurement. The study included a total of 940 OCT visits with an average of 6.3 visits per eye over an average of 4.9 years. Retinal nerve fiber layer (RNFL) loss rate was calculated by serial linear fit of average RNFL thickness measurements. Demographics as well as central corneal thickness (CCT) and corneal hysteresis (CH) data were also collected.

Outcome measures

Pearson correlation coefficients and random coefficient models were used to evaluate the relationship between mean standard and modified IOP measurements and RNFL thickness measurements over time in POAG subjects. Secondary outcomes of CCT and CH correlation to RNFL were similarly analyzed.

Results

For all 148 POAG eyes, the overall rate of RNFL loss for an average standard GAT IOP of 17.9 mmHg was 1.08 µm per year (p = 0.002). Each 1-mmHg increase in standard GAT IOP was associated with an additional RNFL loss of 0.047 µm per year (r = 0.153, p = 0.06). Each 1-mmHg increase in modified GAT IOP was associated with an additional RNFL loss of 0.084 µm per year (r = 0.289, p = 0.0005). A modified prism IOP measurement ≥ 22 mmHg indicates a 2.57 times greater probability of significant RNFL loss than a standard prism IOP measurement ≥ 22 mmHg, p < 0.0001.

Conclusions

Higher levels of GAT IOP during follow-up were related to higher rates of progressive RNFL loss detected by optic nerve OCT in treated POAG. A modified GAT prism surface demonstrates a significantly increased sensitivity, reliability and differentiation to progressive RNFL loss when compared to a standard GAT prism measured IOP.

Précis

A modified applanation surface prism with a corneal conforming shape used on a Goldmann tonometer appears to be a more sensitive and reliable indicator of progressive glaucomatous optic neuropathy as measured by retinal nerve fiber layer changes.

Background

The primary risk factor for progression of glaucomatous optic neuropathy (GON) has been identified as intraocular pressure (IOP) in several clinical trials [1,2,3,4, 6]. The Early Manifest Glaucoma Trial (EMGT) demonstrated each additional 1-mmHg of increased IOP was correlated with an average of 10% increased visual field loss [5]. At the opposite end of the of the POAG spectrum, the Advanced Glaucoma Intervention Study (AGIS) related less visual field loss with lower IOP in those with advanced POAG [4]. Even among those in a pre-glaucomatous state, the Ocular Hypertension Treatment Study (OHTS) showed that pharmacological lowering of intraocular pressure (IOP) was associated with a 54% reduced relative risk of developing POAG [7].

Although automated perimetry has been a baseline standard for the detection of GON progression, other imaging modalities measuring structural changes in the optic nerve have been shown to precede visual field loss [2, 3]. Changes detected by RNFL imaging are shown to often be the earliest indication of GON prior to associated visual field changes [8, 9]. RNFL imaging devices quantitatively measure the nerve fiber thickness and are a mainstay in diagnosis and treatment of glaucoma. Methods for evaluating the RNFL include spectral domain optical coherence tomography (SD-OCT) [10,11,12,13,14]. One clinical SD-OCT device commercially available is the Cirrus Zeiss OCT5000 (Carl Zeiss Meditec, Inc., Dublin, CA) which provides an objective and reproducible evaluation of the RNFL.

We know from, Medieros et al., that there is a positive relationship between Goldmann IOP and progressive RNFL loss in POAG patients by longitudinally measuring RNFL [15]. Eyes with POAG and glaucoma suspects progression were found to have an overall RNFL loss of 0.95 µm/year and 0.13 µm/year per increase of 1 mmHg. The Goldmann applanation tonometer (GAT) was used as the IOP reference measurement in all the aforementioned studies.

A modified Goldmann prism or correcting applanation tonometry surface (CATS) prism utilizes a centrally concave and peripherally convex surface, which partially matches corneal curvature. The altered prism is affixed to the Goldmann tonometer and uses the same measurement protocol without recalibration [16,17,18,19,20]. The prism was designed to minimize the IOP measurement errors found with the standard flat GAT prism [18]. Clinical studies using the modified prism have demonstrated that its IOP measurements are less influenced by variations in CCT, corneal hysteresis (CH), and tear film compared to the standard prism [16,17,18,19,20]. The purpose of this study is to examine the differential correlation in a cohort of glaucoma patients by measuring IOP with a standard flat surfaced Goldmann prism and a modified corneal conforming Goldmann prism to longitudinal RNFL changes measured by SD-OCT.

Methods

The study design was as a cross-sectional cohort study. Participants were included in a retrospective, longitudinal analysis to examine the correlation of optic nerve structural changes to IOP measured by Goldmann applanation tonometry using a standard and modified prism. Participants in this study were cross-sectionally provided without personal identifiable data according to a pre-established protocol between 1–1–20 and 2–15–23 at three participating centers within the Arizona Eye Consultants (Tucson, AZ) practice. These records included SD-OCT5000 measurement data with a maximum range of 1–1–14 to 2–15–23. All participants were undergoing routine follow-up visits for POAG diagnosed by standard diagnostic criteria (listed below) with required imaging and field tests. Deidentified participant data was entered into an electronic medical record database for collating and retrieval. All protocols and the methods described adhered to the tenets of the Declaration of Helsinki.

Subjects were undergoing comprehensive ophthalmologic examinations at each visit. Examination data included deidentified medical history and medication review, slit-lamp biomicroscopy, best-corrected visual acuity, IOP measurement using calibrated Goldmann applanation tonometry (GAT), dilated fundoscopic examination, gonioscopy, optic disc photography, and automated perimetry using 24–2 Full-threshold. Only subjects with gonioscopically demonstrated open angles were included. Subject exclusion criteria included less than 20/80 best-corrected visual acuity, more than ± 6.0 diopters refractive error, or cylinder correction greater than 3.0 diopters. Participant’s data was excluded if it was found to have any other systemic or ocular disease that could affect the optic nerve or visual field testing.

The study data included patients diagnosed with POAG and paired IOP measurements completed with the standard and modified Goldmann prisms. Ensuring the strictest definition, data was included if eyes were classified as having glaucoma having 2 repeatable abnormal visual field tests, indicated by a pattern standard deviation outside of the 95% normal confidence limits. Specifically, eyes were classified as POAG if they included a history of standard GAT measured IOP (> 21 mmHg) without signs of secondary open angle glaucoma. Bilateral eligible eyes from a given patient’s data were included in the analysis where applicable. Statistical procedures were used to take into account the correlation between measurements within the same patient (see below).

POAG subjects with a minimum 5 follow-up visits in which high quality SD-OCT measurements were obtained and at least 2 same visit sequential Goldmann IOP measurements using both the standard and modified prisms were required for inclusion in this study. Quality OCT measurement excluded those with decentration, artifacts or optical image drop-out near the 3.2 mm measurement circumference.

Intraocular pressure measurements were completed on a calibrated Goldmann applanation tonometer BM- 900 (Haag-Streit, Bern, Switzerland). Measurement techniques were in accordance with the tonometer instructions for use (IFU) and ANSI Z80.10–2014/18- Annex A- A.3 Protocol for using the reference tonometer [21].

Sequential OCT images were obtained with the spectral domain Zeiss OCT5000 (Carl Zeiss Meditec, Inc., Dublin, CA). Participants had been diagnosed with POAG by untreated IOP > 21, disk changes, and visual field (HVF) loss consistent with glaucomatous optic neuropathy (GON). Retinal nerve fiber layer (RNFL) loss rate was calculated by serial linear fit of global average RNFL thickness data measurements. The RNFL thickness measurements were obtained on a 3.2-mm diameter calculation circle around the optic nerve head. In addition to demographics data, central corneal thickness (CCT) was obtained using a Zeiss OCT5000 (Carl Zeiss Meditec, Inc., Dublin, CA) and corneal hysteresis (CH) was obtained using the Ocular Response Analyzer (ORA) (Reichert Technologies, Buffalo, NY).

Statistical Analysis using random coefficient models were used to evaluate the relationship between IOP and RNFL thickness measurements over time. Previous studies have used these general linear mixed effects (GLME) models with randomized slopes and intercepts to investigate the correlation of RNFL to multiple variables in glaucoma [22, 23]. The model was constructed evaluating relationships between IOP and time dependent OCT RNFL thickness measurements. Linear regression SD-OCT RNFL thickness measurements were designated as the dependent variable and the time varying predictor was IOP.

Almost all patients had a left and right eye included so the effects of patient laterality was included in the GLME. It was modeled as a random effect and intercept alongside fixed effect coefficients. Random intercept inclusion allowed for baseline RNFL variation and randomized slope coefficients allow for progressive RNFL loss variations among eyes and patients. Random effects models were estimated in the R programming language, R-version 4.3.3, using maximum likelihood and the function lme() from the nlme package (R Project for Statisical Computing).

Each model included patient as a random intercept effect. Fixed effect covariates included eye, sex, race, age, and average Mod-Std IOP difference (M-S.avg). Because approximately 20% of patients had missing values for corneal hysteresis (CH), estimation occurred with and without the CH variable. The inclusion of random intercepts allows for the variation in baseline RNFL, whereas the random slopes allow for the variation in the rate of progressive RNFL loss among eyes and patients.

The general basis form of the models for an IOP measurement was as follows:

$$IOP=\left({\beta \left(0\right)}^{GAT}+{I}^{CATS}*{\beta \left(0\right)}^{\Delta }+\beta {\left(0\right)}^{cct}*CCT+\beta {\left(0\right)}^{CH}*CH+\beta {\left(0\right)}^{age}*Age\right)+\left({\beta \left(1\right)}^{GAT}+{I}^{CATS}*{\beta \left(1\right)}^{\Delta }+\beta {\left(1\right)}^{cct}*CCT+\beta {\left(1\right)}^{CH}*CH+\beta {\left(1\right)}^{age}*Age\right)*RNFL Loss+(1+RNFL Loss | Eye, Race, Sex)$$

Subsequent models were built with and without corneal hysteresis (CH) and separately Mod and Std IOP as well as same day Mod-Std differential IOP.

Results

The cross-sectional study included deidentified data from all patients from the EMR database examined between the dates of 1–1–20 and 2–15–23. The collection data included 1927 eyes from 966 patients with same visit sequential standard and modified Goldmann IOP measurements. From the IOP database pool, the full analysis dataset contained 148 eyes from 75 patients met the inclusion criteria of a diagnosis of primary open angle glaucoma (POAG) with at least 5 sequential quality optical coherence tomographer (OCT) measurements. Annualized RNFL was estimated for each patient eye via linear regression on RNFL readings, with time as the predictor variable. RNFL regression ranged from − 6.27 to 2.32 microns, with mean − 1.08 and standard deviation 1.30.

The full analysis data set Included 575 Goldmann IOP measurements with standard and modified prisms affixed to the calibrated Goldmann applanation tonometer (GAT) armature. The study included a total of 940 OCT visits with an average of 6.3 ± 1.4 visits per eye over an average of 4.9 ± 1.5 years.

Average values of the RNFL thickness throughout the evaluation period was 74 ± 13 µm. Paired modified minus standard IOP measurements averaged 3.8 ± 1.6 mmHg per patient over the collection time interval.

There were 9 predictor variables, described below.

• 1. Mod IOP avg: Patient eye average intraocular pressure from the newer modified prism, in mm Hg. Ranges from 12.00 to 42.50 mmHg, with mean 19.97 ± 4.37 mmHg.

• 2. Std IOP avg: Patient eye average intraocular pressure from the standard prism, in mm Hg. Ranges from 9.00 to 39.50 mmHg, with mean 17.95 ± 4.23 mmHg.

• 3. Mod. IOP minus Std. IOP avg: Patient eye average difference (Mod. – Std.), only computed when both measurements are taken during the same visit. Ranges from − 2 to 6 mmHg, with mean 1.92 ± 1.73 mmHg standard deviation. If a patient had only one pressure measurement during a visit, that measurement is included in the Mod. or Std. IOP average, but was not used in this differential (Mod.-Std.) IOP variable. Therefore, this variable is not identical to (Mod. IOP avg – Std. IOP avg).

• 4. CH avg: Patient eye average corneal hysteresis. Ranges from 3.70 to 13.55 mmHg, with mean 9.03 ± 1.66 mmHg standard deviation. Of the 75 patients, 14 did not have measured values for this variable.

• 5. Eye: Indicator variable for OD (right) or OS (left) eye.

• 6. Age: Highest age in years recorded for each patient on any visit. Ranges from 13 to 92, with mean 73.16 ± 11.20 standard deviation.

• 7. Sex: Reported sex of the patient. 47 patients were female and 28 were male.

• 8. Race: Reported race/ethnicity of the patient. Caucasian was selected for 50 patients (67%), Non-white/Hispanic for 15 (20%), African-American for 8 (11%), and Asian for 2 (2.7%).

• 9. CCT: Measured central corneal thickness. Mean 542 ± 49 µm standard deviation.

Correlations and regressions without covariates

For all 148 POAG eyes, the overall rate of RNFL loss for an average standard GAT IOP of 17.9 mmHg (19.9 mmHg by modified prism) was 1.08 ± 1.30 µm per year (p = 0.0005). Demonstrated in Fig. 1, Each 1-mmHg increase in standard GAT IOP was associated with an additional RNFL loss of 0.047 µm per year (r = 0.153, p = 0.06). Also demonstrated in Fig. 1, Each 1-mmHg increase in modified GAT IOP was associated with an additional RNFL loss of 0.084 µm per year (r = 0.289, p = 0.0005).

Fig. 1

- RNFL loss in POAG (annualized) verses average IOP measured by a standard and modified Goldmann prism

The graph in Fig. 1 shows RNFL loss per year as the vertical variable, with Mod and Std average IOP readings on the horizontal axis. Orange triangles represent Std IOP readings, while blue circles represent Mod IOP readings. Univariate regression lines are included. Higher measurements of modified and standard IOP are associated with more negative slopes of RNFL loss per year. The modified prism IOP slope is statistically significant at p = 0.0005 and the standard prism IOP slope approaches significance at p = 0.06 (Table 1).

Table 1 Modified and standard prism IOP regression analysis output summary

To examine the differences between modified and standard slopes, an interaction term was created. The interaction term represents a difference in slope between the two lines. The interaction coefficient value was not statistically significantly different from zero (p = 0.299). Although the Modified prism slope is steeper at − 0.084 µm/yr/mmHg compared to the standard slope of − 0.047 µm/yr/mmHg, claiming that the slopes differ significantly in unpaired measurements is not supported, as the 95% confidence interval for the modified slope of (− 1.474, − 0.419) contains the standard slope.

There is a possibility of insufficient study power in unrelated standard and modified IOP measurement slope differences correlated to RNFL loss due to time and operator variations in IOP measurement. To account for this, a same-time IOP measurement (paired) differential between modified and standard prisms was examined to directly interpret the difference in average IOP as it correlates to RNFL loss in POAG. IOP which was measured by standard or modified prisms at separate times were excluded. Figure 2 illustrates each 1 mmHg difference between modified and standard prism IOP measurements was significantly correlated to RNFL loss at 0.66 µm per year (r = 0.289, p = 0.0004) (Table 2).

Fig. 2

- RNFL loss in POAG (annualized) verses average paired differential IOP modified minus standard Goldmann prism measurements

Table 2 Modified minus standard prism differential IOP regression analysis output summary

CCT did not correlate to RNFL loss in treated POAG (p = 0.51). Without including the known CH covariate of IOP, each 1-mmHg decrease in CH was associated with an additional RNFL loss of 0.244 µm per year (p = 0.0001).

In the table of Pearson correlations between continuous variables, Mod. IOP average and Std. IOP average have a strong correlation of + 0.915 between them. Mod average has a greater correlation strength with the response, magnitude 0.281 against 0.153 for Std (Table 3).

Table 3 Pearson correlation strength between continuous variables

Ocular hypertension can be defined as an IOP ≥ 22 mmHg measured by either a modified or standard Goldmann prism. Furthermore, significant RNFL loss progression can be defined as ≥ 1.0 µm/yr. The lower right quadrant (labeled IOP predicted progression region) in Fig. 1 inclusive of the lines delineated by x = 22 mmHg vertically and y = 1.0 µm/yr. horizontally can be used to can be used to calculate the relative risk of RNFL progression ≥ 1.0 µm/yr. This analysis indicates an IOP measurement ≥ 22 mmHg with the modified prism is 2.57 times more likely to demonstrate significant RNFL loss than a standard prism at the same IOP, p < 0.0001.

Multivariate models with covariates

Building the multivariate models almost all patients have a left and right eye, so the effects of the patient were included. Patient effects were modeled as a random effect alongside fixed effect coefficients. Random effects models were estimated in the R programming language, version 4.3.3, using maximum likelihood and the function lme() from the nlme package.

Each model included patient as a random intercept effect. Fixed effect covariates included eye, sex, race, CCT, age, and average Mod-Std difference (Mod-Std.avg). Because approximately 20% of patients had missing values for corneal hysteresis (CH), estimation occurred with and without the CH variable. There are a total of 6 models, 2 with CH and 4 without CH. Each set includes models with neither Mod IOP nor Std IOP, Mod IOP only, Std IOP only, both Mod IOP and Std IOP.

Model summary results from GLME

Summaries of fixed effect results from each of the 6 GLME models. The “ ~ ” symbol separates the response from predictor variables. Output includes each coefficient, standard error, t-value, and P-value. The Models included Standard IOP alone, Modified IOP alone, Standard and Modified IOP together, Differential Mod.- Std. IOP, and corneal hysteresis with either Standard or Modified IOP.

Modeled covariates tend to have consistency across models, as described below.

• Higher values of modified prism IOP are always associated with more negative regression slopes for RNFL loss. This aligns with the univariate result. In both models with and without CH, the coefficient is statistically significant at the 5% level. (Tables 5, 6 and 9)

• Higher values of standard prism IOP are always associated with more negative regression slopes for RNFL loss. This aligns with the univariate result. In models without CH, the coefficient approaches statistically significant at the 5% level.(p = 0.027,0.072) (Tables 4 and 6). In the model with CH, it does not reach significance (p = 0.113) (Table 8)

• The Mod. – Std. differential IOP (Mod-Std) is significantly associated with a more negative slope of RNFL loss, p = 0.0007 (Table 7). On the other hand, in the models including both Std IOP and Mod IOP, this differential IOP (Mod.-Std) variable has an insignificant impact on RNFL loss slope. This change is not surprising, as direct measurements are often more informative than a computed difference between direct measurements.(Table 6)

• Age, Race, CCT and Eye never have a significant impact on the slope of RNFL loss. (Tables 4, 5, 6, 7, 8 and 9)

• Being male has an impact of around − 0.40 microns per year, compared to being female. This coefficient is never statistically significant at the 5% level, but the P-values are between 0.073 and 0.085 in all six models. (Tables 4, 5, 6, 7, 8 and 9)

• Decreased corneal hysteresis (CH) always has a significant negative impact on slope of RNFL loss, with an increase between 0.14 and 0.17 microns of RNFL loss for each 1 mmHg decrease in CH level. P-values range from 0.028 to 0.035. (Tables 8 and 9)

Table 4 Standard Prism only: Fixed effects: regress.yr ~ Std.IOP.avg + age + sex + race + Eye + CCT

Table 5 Modified Prism only: Fixed effects: regress.yr ~ Mod.IOP.avg + age + sex + race + Eye + CCT

Table 6 Standard and modified prism together and paired differential IOP, Mod.IOP-Std.IOP: Fixed effects: regress.yr ~ Std.IOP.avg + Mod.IOP.avg + age + sex + race + Eye + CCT + Mod.IOP-Std.IOP

Table 7 Paired differential IOP, Mod.IOP-Std.IOP: Fixed effects: regress.yr ~ age + sex + race + Eye + CCT + Mod.IOP-Std.IOP

Table 8 Average Corneal Hysteresis with Avg. Standard Prism IOP: Fixed effects: regress.yr ~ Std.IOP.avg + CH.avg + age + sex + race + Eye + CCT

Table 9 Average Corneal Hysteresis with Avg. Modified Prism IOP: Fixed effects: regress.yr ~ Mod.IOP.avg + CH.avg + age + sex + race + Eye + CCT

Discussion

The Pearson correlation coefficient indicates an increased reliability and predictive value to progressive RNFL loss in treated POAG by the modified prism IOP measurement compared to the standard prism IOP measurement. The steeper slope of 0.084 µm/yr/mmHg per year with the modified prism compared to 0.047 µm/yr per year in standard prism provides a greater sensitivity and differentiation in measurement predicting RNFL loss. Furthermore, each 1 mmHg difference between modified and standard prism IOP measurements was associated with RNFL loss at 0.66 µm per year. The increased modified prism IOP reliability is supported by the multivariate analyses with standard prism correlations similar to prior studies [15].

In both the univariate and multivariate analysis, the modified prism IOP measurement demonstrated more usefulness than the standard prism IOP measurement when predicting the rate of RNFL loss. There are several indications: Modified prism IOP has a stronger correlation magnitude and statistical significance in average modified prism IOP and paired differential (modified-standard) IOP. In the multivariate random effects models, modified prism IOP coefficients show statistical significance more often both individually and in paired differential IOP measurements. When both standard prism IOP and modified prism IOP variables are entered, the modified prism IOP maintains the proper direction in the counter-balance collinear effect, indicating greater predictive strength.

Using standard benchmarks of ocular hypertension with an IOP ≥ 22 mmHg and a significant progressive RNFL loss (average) > 1.0 µm/yr., a prognostic relative risk can be calculated between the modified and standard prism IOP measurements. A modified prism IOP measurement ≥ 22 mmHg indicates a 2.57 times greater probability of significant RNFL loss compared to a standard prism IOP measurement ≥ 22 mmHg in treated POAG. Healthy subjects estimated an age-related loss of 0.08 µm/year in average RNFL thickness [24]. The average rate of RNFL loss in the present study of progressing POAG eyes (> 1.0 µm per year, or > 1.5% per year) was 2.14 µm per year compared with only 0.27 µm per year (< 0.5% per year) in non-progressing POAG (P = 0.01). This large difference in modified and standard prism measured IOP as an indicator of POAG progression (2.57 times) could support a hypothesis of decreased incidence in normal tension glaucoma (NTG) as measured by the modified prism for future examination.

There is a strong association in the multivariate analysis between decreased corneal hysteresis (CH) and progressive RNFL loss even when IOP is included as a covariate measured by standard or modified prisms. These findings indicate CH may have a separate effect on glaucoma progression than IOP as found in previous studies examining CH and RNFL loss [25, 26]. Zhang, et al. found that each 1 mm Hg decrease in CH was associated with a 0.13 μm/year increase in RNFL loss which is almost identical to our 0.14 to 0.15 μm/year increase in RNFL loss in the present study’s multivariate analysis [26].

It is possible that the increased RNFL loss with lower standard IOP measurements compared to modified prism IOP measurements is associated with a significant clinician reliance on IOP in assessing early signs of POAG progression initiating more aggressive treatment which is independent of RNFL or HVF changes. Multiple studies support a decreased IOP error with the modified corneal conforming surface [17,18,19,20]. No significant standard verses modified prism IOP bias has been demonstrated in healthy, disease-free patients [16, 17].

The multivariate analysis with the limited sample size indicates no effect on treated POAG progression indicated by RNFL loss associated with CCT. The study analysis found no association of RNFL loss with age, eye, and race in treated POAG. Although the association between male gender and increased RNFL loss was not statistically significant at p = 0.054 to p = 0.105, it may have been with a larger sample size, which has been noted in other studies [27]. Although the modified prism IOP has been shown to correct for CCT error, it also has been shown to simultaneously correct for corneal rigidity, curvature, and tear film possibly accounting for its more significant association with glaucoma progression than the standard prism IOP [16,17,18,19,20].

Quadrant specific TSNIT and bifurcated high and low progression rates analyses could be conducted with a larger study. Rates of OCT RNFL loss were higher for the inferior and superior sectors around the optic nerve in an associated study and in agreement with neuroretinal rim loss in glaucoma. [15, 28] Our primary goal in this study was the correlation between IOP and overall RNFL loss. Further analysis incorporating pattern standard deviation (PSD) or visual field index (VFI) could be collected on the study population to assess IOP correlation to visual field changes in treated POAG. This would be beneficial as historical glaucoma clinical trials have used visual fields as the sole end point to determine POAG progression. However, it has been shown that structural damage demonstrable by OCT may often precede detectable associated visual field changes in earlier disease [2, 3]. As we accrue additional longitudinal data, an analysis of visual field data should be possible from follow-up on the study subjects.

The modified Goldmann prism holds significant potential for re-establishing IOP and its historic benchmarks as a primary indication of POAG progression. Since both the modified and standard prisms are designed to measure the same IOP in normal healthy eyes with nominal corneas, no reinterpretation of IOP benchmarks is required. Furthermore, any modified prism difference in IOP from historic standard prism measurements would simply be corrected to account for the original error allowing the practitioner to act accordingly to a new and more accurate IOP. If we consider the diagnosis of NTG as being the false negative of IOP measurement diagnosing glaucoma, a potential demonstration of minimal NTG incidence using the modified prism would significantly improve the sensitivity of Goldmann IOP as a primary diagnostic test for glaucoma and a more useful screening tool.