Comparison of Structural and Functional Features in Primary Angle-Closure and Open-Angle Glaucomas

STRUCTURED ABSTRACT

Purpose:

To identify structural and functional differences in primary angle-closure glaucoma (PACG) and primary open-angle glaucoma (POAG).

Patients and Methods:

In this large cross-sectional study, differences in structural and functional damage were assessed among POAG and PACG patients with optical coherence tomography and reliable visual field testing.

Results:

283 PACG and 4,110 POAG patients were included. Despite similar mean deviation on visual fields (mean [standard deviation] −7.73 [7.92] vs −7.53 [6.90] dB, p = 0.72), PACG patients had thicker global retinal nerve fiber layer (RNFL), smaller cup volume, smaller cup-to-disc ratio, and larger rim area than POAG (77 [20] vs 71 [14] μm, 0.32 [0.28] vs 0.40 [0.29] mm³, 0.6 [0.2] vs 0.7 [0.1], 1.07 [0.40] vs 0.89 [0.30] mm², p <0.001 for all), while POAG patients had more pronounced inferior RNFL thinning (82 [24] vs 95 [35] μm, p <0.001). In a multivariable analysis, hyperopia (odds ratio (OR): 1.24, confidence interval (CI): 1.13-1.37), smaller cup-to-disc ratio (OR: 0.69, CI: 0.61-0.78), thicker inferior RNFL (OR: 1.15, CI: 1.06-1.26) and worse mean deviation (OR: 0.95, CI: 0.92-0.98) were associated with PACG. Functionally, POAG was associated with superior paracentral loss and PACG with inferior field loss. After adjusting for average RNFL thickness, PACG was associated with more diffuse loss than POAG (TD differences 1.26-3.2 dB).

Conclusions:

PACG patients had less structural damage than POAG patients despite similar degrees of functional loss. Regional differences in patterns of functional and structural loss between POAG and PACG may improve disease monitoring for these glaucoma subtypes.

PRECIS

Using a large dataset, we showed structural and functional differences between primary angle closure glaucoma and primary open angle glaucoma. Primary angle closure glaucoma has relative structural preservation and worse functional loss inferiorly.

INTRODUCTION

Glaucoma is the leading cause of irreversible blindness worldwide, with an estimated 76 million people affected in 2020¹. The two main subtypes of glaucoma are primary open-angle glaucoma (POAG) and primary angle-closure glaucoma (PACG). Although POAG is more common^1,2, PACG carries a higher degree of visual disability, with a population-based prevalence of monocular blindness of 27% versus 9% in POAG³.

POAG and PACG have distinct underlying mechanisms by which aqueous outflow is obstructed, with differences in risk factors and clinical presentation^4,5. Prior comparative studies with sample sizes ranging from 64 to 146 have reported inconsistent findings regarding structural differences between the two glaucoma subtypes^6–9. For example, Boland et al. and Sihota et al. have reported that the retinal nerve fiber layer (RNFL) is on average thicker in PACG; on the other hand, Mohammadi et al. found no significant difference among 82 PACG and POAG patients. A few studies evaluated functional differences between the two subtypes, and they have been limited by sample sizes between 25 and 234 and primarily East Asian patient selection^7,10,11. In general, POAG demonstrated more pronounced visual field (VF) damage in the superior hemifield, whereas PACG was associated with more generalized VF loss^11,12.

We sought to provide detailed characterization of structural and functional differences between POAG and PACG in a large sample of patients from a United States-based tertiary care center.

MATERIALS AND METHODS

This study was approved by the Mass General Brigham institutional review board and adhered to the tenets set forth by the Declaration of Helsinki. Given the retrospective nature of this study using existing clinical data, the need for informed consent was waived.

Subject Identification and Data Collection

International Classification of Diseases, Tenth Revision (ICD-10) diagnosis codes were used to identify PACG and POAG patients from the Mass Eye and Ear glaucoma service (see Supplemental Table 1, Supplemental Digital Content, which lists inclusion diagnosis codes). Optical coherence tomography (OCT) scans of the peripapillary retinal nerve fiber layer (RNFL) (Cirrus, Carl Zeiss Meditec, Inc. Dublin, CA, USA) and SITA standard 24-2 Humphrey visual fields (Humphrey Field Analyzer, Carl Zeiss Meditec, Dublin, CA) were obtained during routine clinical care between 2015 and 2022. One eye per patient was used; the eye with worse VF mean deviation (MD) was selected if both eyes were eligible for inclusion. The most recent OCT scan with signal strength ≥ 6 and reliable VF (fixation losses ≤ 33%, false-positive and false-negative rate ≤ 20%) within 30 days of OCT was included^13–16. Intraocular pressure (IOP) and spherical equivalence (SE) measurements closest to the OCT imaging date were collected from medical records.

Two reviewers (GEJ and JAS) performed a manual chart review on a random sample of 40 subjects from each group to confirm the diagnosis based on ICD-10 codes. PACG diagnosis was confirmed by presence of laser peripheral iridotomy (LPI) and/or documentation of a narrow angle on gonioscopy, while POAG diagnosis was confirmed by documented open angle on gonioscopy and no evidence of secondary glaucoma. Two clinicians (MY and LQS) reviewed longitudinal VF testing to confirm the reproducibility of VF defects in both groups.

Primary Outcome Measures

We assessed differences in demographics and structural damage patterns measured by OCT between PACG and POAG. We also evaluated functional differences measured by the mean deviation (MD), pattern standard deviation (PSD), and total deviation (TD) values at each test location based on Glaucoma Hemifield Test regions¹⁷.

Statistical Analysis

All statistical analyses were conducted using R programming language (version 4.1.2, R Foundation, Vienna, Austria). In the univariable analyses, t-tests and chi-squared tests were used to compare continuous and categorical variables between POAG and PACG, respectively. P-values in all univariate analyses were adjusted for multiple comparisons. In the multivariable analyses, we applied logistic regression to predict PACG vs. POAG using (1) demographic and global ophthalmic parameters, and (2) quadrant RNFLT parameters with SE and disc area data. Only subjects with SE and IOP measurements within 30 days of OCT were included in the analyses. Demographic and global ophthalmic parameters included age, sex, race, ethnicity, time from first ICD-10 code to OCT imaging date (as an estimate of disease duration), SE, IOP, global RNFL thickness, vertical cup-to-disc ratio (CDR), rim area, disc area, cup volume, VF MD, PSD, and visual field index (VFI). The Bayesian information criterion was used to remove redundant parameters in model selection. Lastly, we compared TD values at each VF test location between groups with t-test and additionally with linear regression with an adjustment for average RNFLT. More specifically, the linear regression used average RNFLT and the binary outcomes of PACG versus POAG as inputs to predict the 52 TD values individually.

RESULTS

Patient Characteristics

283 PACG and 4,110 POAG patients were included. Manual chart review of 40 patients from each group confirmed the diagnosis of PACG or POAG based on the ICD-10 code for each patient: one POAG patient from the samples had undergone LPI elsewhere, but all other POAG patients had open angles on gonioscopy and no history of LPI.

POAG patients were older than PACG patients at the time of OCT imaging (mean [standard deviation] 69 [13] vs 67 [13] years, p = 0.04). Groups did not differ in racial distribution (60% and 61% White in PACG and POAG respectively, p = 0.83) or ethnicity (12% vs 9% Hispanic, p = 0.21, Table 1). A greater proportion of PACG patients were female compared to POAG patients (62% vs 52%, p = 0.003). PACG and POAG patients had similar treated IOP (16 [5] vs 16 [5] mmHg, p = 0.62; Table 1). PACG patients were more hyperopic, with SE of +0.29 [2.05] diopters (D) (median: +0.13 D; interquartile range [IQR]: −0.75, +1.34 D) compared to −0.96 [3.28] D (median −0.38 D; IQR: −1.63, +0.50 D) in the POAG group (p <0.001; Table 1).

Table 1.

Baseline characteristics of PACG and POAG patients.

Patient Characteristic	PACG (n = 283)	POAG (n = 4110)	p-value
Age (years)	67 [13]	69 [13]	0.04
Race (%) a			0.83
White	59.7 (169 / 283)	60.6 (2491 / 4110)
Black	15.9 (45 / 283)	17.6 (723 / 4110)
Asian	8.1 (23 / 283)	6.9 (284 / 4110)
Other or Unknown	16.3 (46 / 283)	14.9 (612 / 4110)
Ethnicity (% Hispanic)	11.7 (33 / 283)	9.0 (370 / 4110)	0.21
Gender (% Female)	62.2 (176 / 283)	52.4 (2154 / 4110)	0.003
IOP (mmHg)	16 [5]	16 [5]	0.62
SE (Diopters)	+0.29 [2.05]	−0.96 [3.28]	<0.001
Median [IQR]	+0.13 [−0.75, +1.34]	−0.38 [−1.63, +0.50]

Significant p-values at p < 0.05 are presented in bold. P-values were adjusted for multiple comparisons.

^aCalculated with chi-square test.

Abbreviations: PACG, primary angle-closure glaucoma; POAG, primary open-angle glaucoma; IOP, intraocular pressure; SE, spherical equivalence; IQR, interquartile range.

Structural Differences

Comparison of optic nerve head (ONH) measurements from OCT revealed smaller CDR, smaller cup volume, larger rim area, and larger disc area in PACG compared to POAG (0.6 [0.2] vs 0.7 [0.1], p <0.001; 0.32 [0.28] vs 0.40 [0.29] mm³, p <0.001; 1.07 [0.40] vs 0.89 [0.30] mm², p <0.001; 1.96 [0.39] vs 1.90 [0.46] mm², p = 0.03, respectively; Table 2).

Table 2.

Structural and functional features of PACG and POAG patients.

Parameter	PACG (n = 283)	POAG (n = 4110)	p-value
RNFLT (μm)
Average	77 [20]	71 [14]	<0.001
Temporal	56 [15]	54 [15]	0.06
Superior	89 [29]	83 [22]	0.001
Nasal	68 [18]	63 [14]	<0.001
Inferior	95 [35]	82 [24]	<0.001
Rim Area (mm2)	1.07 [0.40]	0.89 [0.30]	<0.001
Disc Area (mm2)	1.96 [0.39]	1.90 [0.46]	0.03
CDR	0.6 [0.2]	0.7 [0.1]	<0.001
Cup volume (mm3)	0.32 [0.28]	0.40 [0.29]	<0.001
HVF MD (dB)	−7.73 [7.92]	−7.53 [6.90]	0.72
Median [IQR]	−5.11 [−10.89, −1.61]	−5.27 [−11.26, −2.37]
HVF PSD (dB)	5.05 [3.65]	5.65 [3.87]	0.01
HVF VFI (%)	81 [24]	81 [21]	0.84

All data are presented as mean [standard deviation], unless otherwise specified. All structural measurements were obtained with optical coherence tomography. One eye per patient was used in these analyses. When both eyes were eligible we used the eye with worse MD.

Significant p-values at p < 0.05 are presented in bold. P-values were adjusted for multiple comparisons.

Abbreviations: PACG, primary angle-closure glaucoma; POAG, primary open-angle glaucoma; CDR, vertical cup-to-disc ratio; RNFLT, retinal nerve fiber layer thickness; HVF MD, Humphrey visual field mean deviation; IQR, interquartile range; PSD, pattern standard deviation; VFI, visual field index.

PACG patients had thicker global RNFL than POAG patients (77 [20] vs 71 [14] μm, p < 0.001) despite similar MD (−7.73 [7.92] vs −7.53 [6.90] dB, p = 0.72; Table 2). PACG patients had thicker RNFL than POAG in the superior and nasal quadrants (89 [29] vs 83 [22] μm, p = 0.001; 68 [18] vs 63 [14] μm, p <0.001; Table 2). POAG patients had more pronounced RNFL thinning compared to PACG patients in the inferior quadrant (82 [24] μm in POAG vs 95 [35] μm in PACG, p <0.001; Table 2), as well as in clock hours 5, 6, and 7 (74 [22] vs 86 [34] μm, 88 [31] vs 103 [43] μm, 85 [31] vs 96 [42] μm; p <0.001 for all, see Supplemental Table 2, Supplemental Digital Content, which shows RNFL clock hour comparison).

Functional Differences

Similar proportions of subjects in both groups had the first VF selected (37.5% in the PACG group and 36.1% in the POAG group, p = 0.69). POAG patients and PACG patients had similar MD values (for POAG, mean [SD]: −7.73 [7.92] dB; median: −5.11 dB; IQR: −10.89, −1.61 dB; for PACG, mean [SD] −7.53 [6.90] dB; median −5.27 dB; IQR −11.26, −2.37 dB; p = 0.72, Table 2). The two groups also had statistically similar values of VFI (81 [24] % in PACG and 81 [21] % in POAG, p = 0.84). On the other hand, POAG patients had higher PSD than PACG patients (5.65 [3.87] vs 5.05 [3.65] dB, p = 0.01; Table 2). When comparing TD values, POAG patients had significantly greater loss in the superior paracentral region with TD differences between 1.50 dB and 2.30 dB (p-values 0.005 and 0.04, Figure 1b and c). Conversely, PACG patients had significantly lower TD values in the inferotemporal region of the VF compared to POAG patients, with differences between 1.70 dB and 2.08 dB (p-values 0.03–0.04, Figure 1b and c). Manual chart review of a subset of 40 PACG patients showed that inferior VF loss was present in 28 patients whereas 12 PACG patients did not have visual loss in this region. Among the 28 patients with inferior field loss, all field defects were reproducible in follow-up VF tests and not rim artifact, defined as a ring-shaped defect in the periphery with very low sensitivity and a difference of −10dB or more compared to the neighboring points closer to fixation^18,19. Similarly, a chart review of 40 POAG patients confirmed 17 instances of superior paracentral loss, which were reproducible and not artifact, while the other 23 did not have visual loss in this region.

Figure 1.

Significant Functional Differences between POAG and PACG.

(a) TD location key. Comparisons of pointwise locations of (b) raw TD differences between PACG and POAG using t-test to compare the 52 individual TD values between the PACG and POAG groups with (c) corresponding significant p-values of the t-tests in the visual field. (d) Adjusted TD differences and (e) significant p-values between PACG and POAG. Adjusted TD difference is calculated by using the binary outcome PACG vs POAG to predict each of the 52 TD values while adjusting for average RNFLT with linear regression.

Color scale and number values refer to the difference between PACG relative to POAG. Only significant pointwise differences at p < 0.05 are included and color-coded. Points denoted with 0 represent either no difference between subtypes or differences that were not statistically significant at p < 0.05. For the p-value graphs, color scale and numbers refer to the p-value of the difference of PACG relative to POAG at the visual field test location.

Abbreviations: TD, total deviation; RNFLT, retinal nerve fiber layer thickness; PACG, primary angle-closure glaucoma; POAG, primary open-angle glaucoma.

Comparison of TD values after adjusting for structural damage revealed more pronounced differences between POAG and PACG (see Supplemental Table 3, Supplemental Digital Content, which shows raw and adjusted TD differences at each VF location)²⁰. PACG patients had worse TD values across all VF locations except for the paracentral region, with differences between 1.20 dB and 3.20 dB (p-values <0.001 to 0.05, Figure 1d and e). TD differences were more pronounced in the inferior hemifield than in the superior hemifield.

Multivariable Analysis

90 PACG and 1,679 POAG patients with SE and IOP measurements within 30 days of OCT were included in the multivariable regression analyses to identify factors associated with PACG diagnosis. Baseline characteristics of this subset of patients are shown in Supplemental Table 4 (see Supplemental Digital Content). In the model assessing demographic and global ophthalmic parameters, Bayesian information criterion identified race, gender, ethnicity, age, IOP, rim area, disc area, cup volume, time from first ICD-10 code to OCT imaging, PSD and VFI as weak predictors and removed them from model selection. Hyperopia (odds ratio [OR]: 1.24; 95% confidence interval [CI]: 1.13-1.37; p <0.001), smaller CDR measured on OCT (OR: 0.69 for 0.1 increase in CDR; 95% CI: 0.61-0.78; p <0.001), and worse MD (OR: 0.95; 95% CI: 0.92-0.98; p = 0.001) were independently associated with PACG (Table 3). In a separate model assessing for quadrant RNFLT associated with PACG using SE and disc area as covariates, preserved inferior RNFL (OR: 1.15 for 10 μm increase in RNFLT; 95% CI: 1.06-1.26; p = 0.001) and SE (OR: 1.23; 95% CI 1.11-1.36; p <0.001) were independently associated with PACG (Table 3). Bayesian information criterion identified the superior RNFL, nasal RNFL, temporal RNFL and disc area as redundant parameters and removed them from model selection.

Table 3.

Multivariable logistic regression models of parameters associated with PACG.

Parameter	OR	95% CI	p-value
Global parameters a
SE (Diopters)	1.24	[1.13, 1.37]	<0.001
CDR b	0.69	[0.61, 0.78]	<0.001
HVF MD (dB)	0.95	[0.92, 0.98]	0.001
RNFLT by quadrant (μm) c
Inferior d	1.15	[1.06, 1.26]	0.001

90 PACG patients and 1,679 POAG patients with available SE and IOP measurements within 30 days of OCT imaging were included. Two models are shown. The first model assessed global parameters associated with PACG and the second model assessed quadrant RNFLT measurements associated with PACG using SE and disc area as covariates. In both models, Bayesian information criterion was used to identify redundant parameters and remove from model selection. Significant p-values at p < 0.05 are presented in bold.

^aParameters associated with PACG from logistic regression of global parameters. Parameters included race, gender, ethnicity, age at the time of OCT imaging, time from first ICD-10 code to OCT imaging, SE, IOP, average RNFLT, rim area, disc area, vertical CDR, cup volume, HVF MD, PSD, and VFI. Bayesian information criterion removed race, gender, ethnicity, age, time from ICD-10 code to OCT, IOP, rim area, disc area, cup volume, PSD and VFI as redundant parameters from model selection.

^bRelative PACG risk for 0.1 increase in CDR.

^cParameters associated with PACG from logistic regression of 4 quadrant RNFLT measurements, SE, and disc area after Bayesian information criteria were used to remove redundant parameters. SE remained significantly associated with PACG (OR: 1.23; 95% CI 1.11-1.36; p <0.001).

^dOR is reported as a relative PACG risk for 10 μm increase in RNFLT.

Abbreviations: OR, odds ratio; CI, confidence interval; SE, spherical equivalence; IOP, intraocular pressure; OCT, optical coherence tomography; ICD-10, International Classification of Diseases, Tenth Revision; RNFLT, retinal nerve fiber layer thickness; CDR, vertical cup-to-disc ratio; HVF MD, Humphrey visual field mean deviation; PSD, pattern standard deviation; VFI, visual field index.

Patient Examples

OCT and VF testing from example POAG and PACG patients are in Figure 2. On OCT, the PACG patient had a smaller CDR (0.71 vs 0.87), greater rim area (0.98 vs 0.77 mm²), and greater disc area (2.34 vs 1.75 mm²) than the POAG patient. Average RNFLT was similar between patients (75 vs 74 μm), but the POAG patient had thinner RNFL in the inferior quadrant and in clock hour 5 (Figure 2d and i). On VF testing, the POAG patient demonstrated MD of −4.41 dB with a superior paracentral defect whereas the PACG patient had worse MD of −6.02 dB with superior and inferior field loss (Figure 2e and j). The inferior field loss pattern was reproducible on subsequent VF’s and was distinguishable from rim artifact upon closer inspection of TD values. The PACG patient had also undergone enhanced depth imaging (EDI) OCT imaging (Spectralis, Heidelberg Engineering GmbH, Heidelberg, Germany) as part of a previous study from our group²¹. Evaluation of the volumetric B-scans revealed prelaminar holes located in the superior optic nerve head (see Supplemental Figure 1, Supplemental Digital Content, showing the prelaminar holes on EDI-OCT imaging).

Figure 2.

Representative examples of POAG and PACG disease patterns.

Testing from the left eye of a POAG patient is shown in the top two rows. The POAG eye had a SE of −0.75 D. Fundus photo showed inferior thinning of the neuroretinal rim (a). The OCT report of the left eye measured a vertical CDR of 0.87, rim area of 0.77 mm², and disc area of 1.75 mm² (b). The RNFL thickness curve (dashed line) measured an average thickness of 74 μm and showed inferior thinning (c). The RNFL quadrant and clock hour maps also showed inferior RNFL thinning (d). Corresponding 24-2 HVF MD was −4.41 dB and the pattern deviation plot demonstrated superior paracentral field loss in 2021 which was reproducible in subsequent HVF in 2023 (e).

Testing from the left eye of a PACG patient is shown in the bottom two rows. The PACG eye had a SE of +2.50 D and had undergone prior laser peripheral iridotomy. Fundus photo did not show obvious thinning of the neuroretinal rim (f). The OCT report of the left eye measured a vertical CDR of 0.71, rim area of 0.98 mm², and disc area of 2.34 mm² (g). The RNFL thickness curve (dashed line) measured an average thickness of 75 μm and shows inferior thinning (h). The RNFL quadrants and clock hours showed borderline superior and more inferior thinning (i). Corresponding 24-2 Humphrey visual field MD was −6.02 dB and the pattern standard deviation plot demonstrated superior defects and inferior field loss (red outline) (j). Further review of subsequent HVF’s and the TD values confirmed that the inferior loss was reproducible and not lens rim artifact.

Abbreviations: POAG, primary open-angle glaucoma; PACG, primary angle-closure glaucoma; SE, spherical equivalence; OCT, optical coherence tomography; CDR, cup-to-disc ratio; RNFL, retinal nerve fiber layer; HVF, Humphrey visual field; MD, mean deviation.

DISCUSSION

Despite the common endpoint of optic neuropathy, PACG and POAG differ in pathophysiological mechanisms and clinical presentations. In POAG, aqueous outflow is diminished through a normal-appearing trabecular meshwork, leading to a slowly progressive optic neuropathy with thinning of the neuroretinal rim and gradual VF loss^4,22,23. In contrast, in PACG, physical occlusion of the trabecular meshwork by peripheral iris tissue leads to significant IOP elevation and glaucomatous optic neuropathy^4,24. However, the distinctions of optic nerve damage and VF loss have not been comprehensively described for PACG and POAG and as such, this is the largest study to date evaluating disease patterns in these two subtypes. In our dataset of 4,393 patients, POAG had more pronounced RNFL thinning in the inferior quadrant, while PACG had a thicker average RNFL at similar VF MD. Relative sparing of the inferior RNFL was significantly associated with PACG in our multivariable analysis. Functionally, POAG was associated with superior paracentral loss with TD differences of 1.5 to 2.3 dB, while PACG was associated with inferior field with TD differences of 1.7 to 2.1 dB and diffuse loss with TD differences of 1.26 to 3.2 dB at a similar degree of structural damage. We believe that our findings contribute to subtype-specific information that enhances the diagnosis and monitoring of these two glaucoma subtypes based on structural and functional differences.

Our study confirms demographic and ophthalmic differences between PACG and POAG. Gender differences between PACG and POAG have been reported in many epidemiologic studies, with PACG being more common among females^25–27. PACG patients in our dataset were on average younger than POAG patients at the time of OCT imaging. While there is age-related RNFL thinning of 0.16 μm/year, the difference in age between our groups does not account for the magnitude of the RNFL differences noted and is separate from age of diagnosis²⁸. PACG patients were also more hyperopic than POAG by 1.25 D, which is consistent with differences of 1.00–3.95 D in prior comparative studies^6,7,29,30. In terms of ONH summary parameters, like prior studies, our PACG patients demonstrated smaller CDR, smaller cup volume, shallower cups, and larger rim area than POAG^31,32. While the mean difference of CDR between PACG and POAG was 0.1, the relative structural preservation in PACG was further demonstrated by smaller cup volume, larger rim area and thicker average RNFL compared to POAG. These differences in ONH morphology are thought to reflect differences in disease mechanisms, where elevated IOP plays a major role in PACG, whereas optic nerve vulnerability along with pressure-dependent and pressure-independent damage occurs in POAG^7,31–34.

In this study, eyes with PACG had thicker average RNFL than those with POAG by 6 μm at the same average MD of −8 dB. This finding is consistent with prior reports on structural damage patterns in PACG and POAG^7,9. In a cohort of 146 patients Boland et al. also reported preserved RNFLT by 6 μm in PACG compared to POAG at MD of −9.8 and −10.0 dB respectively⁷. Conversely, Mohammadi et al. reported no difference in RNFL between POAG and PACG, but the study enrolled only 82 patients with more severe disease (MD of −11.0 dB)⁸. When we examine RNFL by sector, POAG had more inferior RNFL thinning than PACG by an average of 13 μm. With adjustment for SE and disc area which can impact RNFLT ^35,36, each 10 μm of preserved inferior RNFL was associated with 15% increased odds of PACG. Manassakorn et al. also described similar focal RNFL thinning in clock hour 6 in POAG compared to PACG at average MD of −4.60 dB and −4.30 dB respectively⁹. However, unlike our study, they did not detect differences in RNFL thinning at MD < −8 dB, likely due to having only 55 patients in this more advanced disease subgroup⁹. Overall, this study provides further clarification to previously reported differences in structural damage between PACG and POAG. We have been able to provide support for subtype-specific RNFL thinning while controlling for other differences that may impact measured RNFL thickness. At similar disease severity as measured by MD and VFI, PACG eyes show relative preservation of the RNFL compared to POAG, particularly in the inferior quadrant. These differences in structural damage patterns likely reflect the distinct disease mechanisms of these two glaucoma subtypes.

Our functional analysis provides additional information to previous reports on visual loss patterns in PACG and POAG. Yousefi et al. and Gazzard et al. observed that POAG tends to have greater damage in the superior than inferior hemifield^11,12. Nouri-Mahdavi et al. reported that VF loss in POAG is more likely to involve the paracentral TD points than in PACG⁶. Generalized depression was observed in smaller PACG patient cohorts^10,11. Leveraging a larger dataset, we have been able to uncover subtle differences at specific test locations. We showed worse VF loss with TD difference of 1.5 to 2.3 dB in the superior paracentral region along with higher PSD in POAG; and at the same degree of structural damage, PACG had more diffuse loss illustrated by the adjusted TD comparison and further supported by worse MD and better PSD. Furthermore, we have also identified involvement of the inferior VF in PACG which was confirmed through a review of a random sample of patients to be reproducible and not rim artifact.

Altogether, our findings support likely differences in the underlying pathophysiology of POAG and PACG. PACG is mechanistically a pressure-dependent disease, where acute, intermittent, or chronic angle closure secondary to anterior chamber crowding contributes to optic nerve ischemia and damage^37,38. This pattern of pressure elevation leads to diffuse retinal ganglion cell apoptosis with corresponding generalized visual field loss^11,39. On the other hand, in POAG, optic nerve susceptibility predisposes the neuroretinal rim to localized damage at more physiologic pressures^33,34. The inferotemporal region of the ONH has been previously identified as a site of vulnerability and inferotemporal RNFL thinning is often an early distinguishing sign of POAG^34,40,41. Interestingly, unlike previous studies, we have further uncovered inferior field loss in PACG, with TD difference up to 3.2 dB worse than POAG at the same degree of RNFL loss. While this retrospective study was not designed to uncover mechanisms of damage, our prior work describes prelaminar holes in the ONH of PACG eyes, as illustrated by an example of a PACG patient with inferior field loss and prelaminar holes in the superior ONH (see Supplemental Figure 1, Supplemental Digital Content)²¹. Surprisingly, the profound structural damage within the ONH of this patient was associated with only borderline superior RNFL thinning. This suggests that structure-function correlation may manifest differently in PACG compared to POAG. While additional research is needed to confirm the mechanism of inferior field loss, clinicians should not dismiss inferior field loss as an artifact in the presence of preserved RNFL on OCT in a patient with clinical evidence of angle closure. PACG may manifest as functional loss in the inferior region or diffuse loss before significant optic nerve cupping or RNFL thinning.

This study has several limitations. First, we identified subjects using ICD-10 diagnosis codes which was important to identify a large dataset of patients, but also is subject to misclassification and may compromise the diagnostic accuracy of our data. Any misclassification in this study is likely non-differential, and such errors would tend to reduce our ability to detect differences. To identify systematic errors in classification, we reviewed a subset of charts and found only one possible error in 80 charts. Second, this dataset lacked lens status and axial length data, thus the refractive error difference we reported may slightly differ from the true difference. Functional comparisons may also have been influenced slightly by lens status. Third, hyperopia is likely associated with rim artifact⁴². While we confirmed that the inferior loss in 40 randomly selected PACG patients were not from rim artifact, we could not rule out all rim artifacts in our VF dataset. Fourth, while we selected for the most recent OCT-VF pairs and included only reliable VF’s, we did not assess for VF learning effect, although the two groups included similar proportions of first VF test. Fifth, our study identified fewer PACG patients than POAG patients, which parallels the prevalence of PACG and POAG in individuals aged 40-80 years in North America (0.26% and 3.29% respectively)¹. Despite this, our PACG group was still larger than most prior comparative studies^6–8,31. Additionally, most subjects included were White, which limited our power to detect racial and ethnic differences. Furthermore, the design of this study at a single care center may limit generalizability to the broader population of glaucoma patients. Lastly, given the cross-sectional nature of this study, we were unable to assess differences in disease progression, surgical intervention, and clinical outcomes between the PACG and POAG patients in our study. Time from first ICD-10 code to OCT imaging did not differentiate PACG from POAG in our multivariable analysis, but this parameter is only an estimate for disease duration given the lack of longitudinal data in this dataset. Future investigations with longitudinal data, factoring in history of glaucoma surgeries and true disease duration, would further clarify patterns of disease evolution of these two glaucoma subtypes.

In conclusion, this large cross-sectional study identified important differences in structural and functional damage between POAG and PACG, in addition to verifying previously published findings regarding demographic and ocular differences. Structurally, POAG patients had greater thinning of the inferior RNFL while PACG patients had relatively preserved RNFL. At the same degree of structural damage, PACG patients had worse MD and more generalized field loss. Functionally, POAG was associated with superior paracentral loss and PACG with inferior field loss. Clinicians may better diagnose and monitor patients with POAG or PACG by confirming structural or functional damage in these disease-specific locations.

Supplementary Material

Supplemental Content (tif, pdf, etc., PDF Preferred)

Supplemental Table 1. PACG and POAG ICD-10 diagnoses codes.

Supplemental Table 2. Comparison of RNFLT by clock hour sectors between PACG and POAG patients.

Supplemental Table 3. TD Differences without and with adjustment for average RNFLT between PACG and POAG patients.

Supplemental Table 4. Baseline characteristics of PACG and POAG patients included in logistic regression analysis of factors associated with PACG.

Supplemental Figure 1. PACG patient with inferior visual field defect and prelaminar holes in the superior ONH.