Clinical Significance of the Location of Recurrent Optic Disc Hemorrhage in Glaucoma

Abstract

Purpose: To compare clinical characteristics and visual field (VF) progression between glaucomatous eyes with disc hemorrhage (DH) recurring at the same or different locations in relation to the initial DH location.

Methods: One hundred forty-seven open-angle glaucoma patients with DH who were observed for more than 4 years and had more than six VF tests were included. Disc hemorrhage that recurred at the same location as the initial DH was defined as a DH at the same location, and DH that recurred at a different location was defined as a DH at a different location. Overall rate of VF change using the mean deviation (MD) was compared between groups using a linear mixed model, and Kaplan-Meier survival analysis was performed.

Results: Seventy-seven (52.4%) eyes had nonrecurrent DH, and 70 (47.6%) eyes had recurrent DH detected during the follow-up period. Recurrent DH at the same location (−0.32 dB/y) was not significantly different from the nonrecurrent DH group in terms of the rate of VF change (P = 0.116). However, the recurrent DH group at different locations (−1.07 dB/y) had a significantly different rate of VF change compared to the recurrent DH group at the same location (P = 0.014). By Kaplan-Meier analysis, eyes with recurrent DH at different locations showed the fastest time to VF progression compared to other groups (log rank test, P < 0.001).

Conclusions: Eyes with recurrent DH recurring at different locations from initial DH sites had more pronounced VF progression.

Disc hemorrhage (DH) is a prominent feature of glaucoma.^1,2 The Ocular Hypertension Treatment Study demonstrated that it is also a significant risk factor for the development of glaucoma.³ In the Early Manifest Glaucoma Trial and Collaborative Normal-Tension Glaucoma Study (CNTGS), it was significantly associated with glaucoma progression.^4,5 Typically, DH develops in association with notching, progressive changes in the optic disc rim, localized retinal nerve fiber layer (RNFL) defects, and enlargement of localized RNFL defects.^6–9 Many studies have shown that rim notching or loss, focal RNFL defects, and corresponding localized and global visual field (VF) deterioration occur after DH.^10–12 Some studies have shown more pronounced progressive changes in the VF in eyes with recurrent DH compared to eyes with a nonrecurrent DH.¹³ However, other studies have shown that eyes with recurrent DH do not have greater glaucoma progression than those with a nonrecurrent DH.^14,15 The clinical significance of recurrent DH remains controversial. 

In this study, we investigated open-angle glaucoma (OAG) patients with DH who were followed up for at least 4 years. The clinical characteristics and VF progression of OAG eyes with recurrent DH were compared according to the recurring clock hour sites and the location in relation to the initial DH. 

Methods

Subjects

We retrospectively reviewed the medical records of OAG patients with DH who were seen consecutively by a glaucoma specialist (CKP) from January 2005 to August 2010 at the glaucoma clinic of Seoul St. Mary's Hospital. Patients who were followed up for at least 4 years with visits at 2- to 6-month intervals after the initial visit were selected for the study. For the initial workup, each patient received a complete ophthalmic examination, including a review of the medical history, measurement of best-corrected visual acuity, refraction, slit-lamp biomicroscopy, gonioscopy, Goldmann applanation tonometry, central corneal thickness using ultrasound pachymetry (Tomey Corporation, Nagoya, Japan), axial length using ocular biometry (IOL Master; Carl Zeiss Meditec, Dublin, CA, USA), dilated stereoscopic examination of the optic disc, color and red-free disc/fundus photography (Canon, Tokyo, Japan), Heidelberg Retina Tomography (HRT; Heidelberg Engineering GmbH, Heidelberg, Germany), Stratus optical coherence tomography (OCT; Carl Zeiss Meditec), and Humphrey VF examination using the Standard 24-2 test (Carl Zeiss Meditec). During the follow-up period, patients had serial disc and red-free fundus photography taken regularly at intervals of 6 ± 2 months, and HRT or OCT, as well as VF examinations, regularly at intervals of 6 to 12 ± 2 months. 

For glaucoma diagnosis, patients had to fulfill the following criteria: glaucomatous optic disc appearance (such as diffuse or localized rim thinning, a notch in the rim, or a vertical cup-to-disc ratio higher than that of the other eye by more than 0.2) and glaucomatous VF loss (defined as a pattern standard deviation [P < 0.05] or glaucoma hemifield test results [P < 0.01] outside the normal limits in a consistent pattern on two qualifying VFs), both confirmed by two glaucoma specialists (H-YLP, CKP); and an open angle on gonioscopic examination. 

All patients had to meet the following additional inclusion criteria to be entered into the study: a best-corrected visual acuity ≥ 20/40, a spherical refraction within ± 6.0 diopters, a cylinder correction within ± 3.0 diopters, consistently more than six reliable VFs (defined as false negative < 15%, false positive < 15%, and fixation losses < 20%), and mean deviation (MD) better than −15.00 dB. Patients were excluded on the basis of any of the following criteria: a history of any retinal disease, including diabetic or hypertensive retinopathy; a history of eye trauma or surgery with the exception of uncomplicated cataract surgery; other optic nerve disease besides glaucoma; and a history of systemic or neurologic diseases that might affect the VF. If glaucoma incisional or laser treatment was performed during the follow-up, only the data obtained before the treatment were analyzed. If both eyes were eligible for the study, one eye was randomly chosen for the study. 

The Institutional Review Board from Seoul St. Mary's Hospital approved this study, which adhered to the principles of the Declaration of Helsinki. 

Disc Stereophotography

Serial disc photographs were evaluated thoroughly during the total follow-up period. Two glaucoma specialists (EKK and H-YLP) evaluated the stereoscopic optic disc photographs, and each observer was blinded to the patient's clinical information and test results. Discrepancies between the two observers were resolved by consensus. A DH was defined as an isolated flame-shaped or splinter-like hemorrhage on the optic disc or peripapillary area extending to the optic disc border. Alternative causes of hemorrhage were excluded by diagnostic testing for ischemic optic neuropathy, papillitis, retinal vein occlusion, diabetic retinopathy, and posterior vitreous detachment. 

All DHs that occurred during follow-up were evaluated and recorded based on the disc photographs. Nonrecurrent DH was defined as eyes with DH only once during the total follow-up. Recurrent DH was defined as eyes with more than one DH during follow-up. Eyes with recurrent DH were classified into four groups according to the recurring location of the DH as shown in Figure 1. Eyes were initially classified into “recurrent DH at one clock hour” if the DH recurred at one clock hour site (Figs. 2A, 2B) or “recurrent DH at more than one clock hour” if the DH recurred at multiple clock hour sites (Figs. 2C, 2D). For each DH, recurrent DH was classified as “DH at same location” or “DH at different locations.” Recurrent DH that occurred at the same location compared to the initial DH was defined as DH at the same location (Figs. 2A, 2C).^14–16 Disc hemorrhage that occurred at different locations compared to the initial DH, but within the same clock hour, was defined as DH at different locations (Figs. 2B, 2D). Eyes that could not be classified into these four categories (e.g., multiple DH at presentation without recurrence or single DH at one site and another single DH at another site without recurrence) were excluded. 

Figure 1

 

Recurrent disc hemorrhage (DH) was defined as the occurrence of more than one DH during the total follow-up period. There were eyes with recurrent DH at one clock hour site or at more than one clock hour site. For each clock hour site, recurrent DH was classified as “DH at same location” or “DH at different locations.” Recurrent DH that occurred at the same location compared to the initial DH was defined as a DH at the same location. Disc hemorrhage that occurred at a different location compared to the initial DH, but within the same clock hour of the initial DH, was defined as a DH at different locations. According to this classification, we can divide recurrent DH into four groups: at one clock hour/same location group, at one clock hour/different location group, at more than one clock hour/same location group, and at more than one clock hour/different location group.

Figure 2

 

The initial disc hemorrhage (DH) is indicated by a white outline, and recurrent DH is indicated by a green outline. (A) Recurrent DH at one clock hour/same location; (B) recurrent DH at one clock hour/different locations; (C) recurrent DH at more than one clock hour/same location; and (D) recurrent DH at more than one clock hour/different locations. In recurrent DH eyes at more than one clock hour site, if one or more DH were at a different location, it was classified into recurrent DH at different locations. As shown in (D), superotemporal DH recurred at the same location, but the inferotemporal DH recurred at a different location from the initial DH location.

Disc hemorrhage occurring at the borders of a localized RNFL defect was classified as “DH accompanying a RNFL defect.” Localized RNFL defects were identified using red-free RNFL photography. A localized defect was defined as a wedge-shaped defect reaching the optic disc and covering less than 60° of the optic disc circumference. 

Definition of VF Progression

Visual field progression was determined by event-based analysis using the Guided Progression Analysis (GPA) software from the Humphrey Field Analyzer. Visual field progression was defined as a significant decrease from baseline (two VFs) pattern deviation at three or more of the same test points on two or three consecutive VF tests. The software classifies VF progression as “possible progression” or “likely progression,” respectively. Only the “likely progression” condition was defined as VF progression. 

Statistical Analysis

Baseline characteristics were compared between the single and recurrent DH groups. Normally distributed data were compared using independent Student's t-test. The χ² test was used to compare categorical data. 

The overall progression rate of MD was determined from the serial VF measurements using a linear mixed model. Models were fitted with fixed coefficients (fixed effects) of time (months), age (years), baseline MD, and baseline pattern standard deviation of the VF, accepting random intercepts and coefficients (thus, random effects) for all eyes when analyzing the effect of time. The progression rates of MD were compared between groups by means of testing the interaction term in linear mixed models. 

Kaplan-Meier survival analysis and the log rank test were used to compare the cumulative risk ratio of progression between groups stratified by DH. The first time VF progression was found was regarded as the endpoint in survival analysis. The end of follow-up was when patients without progression were censored. Statistical analysis was performed using the SPSS statistical package (SPSS 12.0; Chicago, IL, USA). A P value less than 0.05 was considered statistically significant. 

Results

A total of 147 eyes of 147 glaucoma patients with DH who met the inclusion and exclusion criteria were included. Five eyes were excluded because the features of the recurrent DH could not fit the classification system. The mean follow-up period was 68.54 ± 9.42 months (range, 48–80 months). Of 147 eyes, 77 (52.4%) had a nonrecurrent DH, and the remaining 70 (47.6%) had recurrent DH detected during the follow-up period. 

Eyes with nonrecurrent DH and recurrent DH had similar baseline and follow-up characteristics, including the baseline MD of the VF (Table 1). Disc hemorrhages accompanying a localized RNFL defect were significantly more frequent in the recurrent DH group (74.3%) than in the nonrecurrent DH group (51.9%, P = 0.004). Systemic factors such as diabetes mellitus and systemic hypertension were not different between the two groups (P = 0.503 and P = 0.298, respectively). However, significantly more patients in the recurrent DH group (27.1%) were taking aspirin compared to the nonrecurrent DH group (14.3%, P = 0.042). 

Table 1

 

Characteristics of Open-Angle Glaucoma Patients With Nonrecurrent or Recurrent Disc Hemorrhage During Follow-Up Periods

 

Subgroups of eyes with recurrent DH were analyzed based on the location of DH recurrence (Table 2). Among 70 eyes, 56 (80.0%) had recurrent DH at the same location and 14 (20.0%) had DH at different locations compared to the initial DH. The two subgroups had similar baseline and follow-up characteristics. Among 56 eyes with recurrent DH at the same location, 44 (78.6%) recurred within one clock hour and 12 (21.4%) recurred at more than one clock hour. This showed a significant difference between the two groups (P = 0.001). Disc hemorrhages accompanying a localized RNFL defect were significantly more frequent in the recurrent DH group at the same location (80.4%) than in the group with recurrent DH at different locations (50.0%, P = 0.027). Aspirin medication was significantly more frequent in the recurrent DH group at different locations (50.0%) compared to the recurrent DH group at the same location (21.4%, P = 0.038). 

Table 2

 

Characteristics of Open-Angle Glaucoma Patients With Recurrent Disc Hemorrhage at the Same Location or at Different Locations

 

Baseline RNFL thicknesses measured using OCT and baseline optic disc parameters measured using HRT were not significantly different between the recurrent DH groups at the same location and at different locations (Table 3). 

Table 3

 

Baseline Retinal Nerve Fiber Layer and Optic Nerve Head Parameters of Open-Angle Glaucoma Patients With Recurrent Disc Hemorrhage at the Same Location or at Different Locations

 

Comparisons between the groups of the slope of the MD of the VF by linear mixed model are shown in Table 4. There were no differences in terms of VF progression between the nonrecurrent DH group (−0.31 dB/y) and the recurrent DH group (−0.39 dB/y; P = 0.668). However, recurrent DH at one clock hour (−0.22 dB/y) was significantly different from recurrent DH at more than one clock hour (−0.67 dB/y; P = 0.028). Recurrent DH at the same location (−0.32 dB/y) was not significantly different in terms of VF progression from the nonrecurrent DH group (P = 0.116). However, the recurrent DH group at different locations (−1.07 dB/y) had a significantly different VF MD slope compared to the recurrent DH group at the same location (P = 0.014). By subgroup analysis, recurrent DHs at different locations showed significant difference of VF MD slope compared to recurrent DHs at the same location in eyes with recurrent DH both at one clock hour and at more than one clock hour (Fig. 3). Recurrent DH at one clock hour/different locations and recurrent DH at more than one clock hour/different location showed significant difference in VF MD progression. In eyes with recurrent DH, there was a trend in increasing VF MD slope as the number of DHs during follow-up increased. However, it was not statistically significant (Table 5). 

Table 4

 

Progression Rate of Mean Deviation in Glaucoma Patients With Disc Hemorrhage

 

Figure 3

 

The rate of visual field (VF) mean deviation (MD) change during the total follow-up period in each group is shown. There were no differences in terms of VF progression between the nonrecurrent disc hemorrhage (DH) group (−0.31 dB/y) and the recurrent DH at one clock hour/same location group (−0.32 dB/y). However, recurrent DH at the same location and different locations showed a significant difference in both the one clock hour group (−0.29 dB/y for recurrent DH at same location, −0.77 dB/y for recurrent DH at different locations) and the more than one clock hour group (−0.42 dB/y for recurrent DH at same location, −0.99 dB/y for recurrent DH at different locations). Recurrent DH at different locations showed significant difference between eyes with recurrent DH at one clock hour (−0.77 dB/y) compared to recurrent DH at more than one clock hour (−0.99 dB/y).

Table 5

 

Comparison of the Visual Field Progression Rate of Mean Deviation Between Glaucoma Patients According to the Number of Disc Hemorrhages

 

Of 77 eyes, 8 (10.4%) in the nonrecurrent DH group showed VF progression. Of 70 eyes, 15 (21.4%) showed VF progression in the recurrent DH group. Among the 15 progressed eyes with recurrent DH, 6 (40.0%) were from the recurrent DH group at the same location and 9 (60.0%) were from the recurrent DH group at different locations (P < 0.001). The nonrecurrent DH group and recurrent DH group did not show differences in cumulative probability of VF progression on Kaplan-Meier analysis (log rank test, P = 0.164; Fig. 4A). The recurrent DH group at different locations showed a greater cumulative probability of VF progression than the recurrent DH group at the same location when assessed on Kaplan-Meier analysis (log rank test, P = 0.009; Fig. 4B). There were statistically significant differences between four groups subdivided by the site and location of DH as assessed by Kaplan-Meier analysis (log rank test, P < 0.001; Fig. 4C). 

Figure 4

 

Kaplan-Meier analysis of the probability of remaining without deterioration of the visual field in patients with disc hemorrhage (DH). Log rank tests comparing eyes with nonrecurrent DH and recurrent DH did not reveal any statistically significant differences ([A] P = 0.164). However, the same test revealed significant differences in VF progression between recurrent DH eyes at the same location and different locations ([B] P = 0.009), as well as significant differences between recurrent DH by number of clock hour sites and by location according to the initial DH location ([C] P < 0.001).

Discussion

We observed that eyes with recurrent DH did not show more rapid VF progression compared to eyes with nonrecurrent DH. However, clinical significance in terms of VF progression was evident when eyes with recurrent DH were stratified into groups based on the DH location. Eyes with DH recurring at different locations from the initial DH showed significantly more rapid VF progression than eyes that had a nonrecurrent DH. Disc hemorrhage recurring at different locations from the initial DH was frequently found in eyes with recurrent DH at more than one clock hour site, had fewer accompanying localized RNFL defects, and was associated with history of more frequent aspirin medication. 

Recurrent DH is reported to occur in approximately 12% to 73% of all DH cases.^14,17 Variances in follow-up periods and study population may contribute to this wide-ranging incidence of recurrent DH. In studies with longitudinal observation periods, the incidence of DH is reported to plateau, with very few initial DHs found after approximately 5 years of follow-up.¹⁸ In our study, we included patients who were followed up at least 4 years to ensure that the incidence of DH would plateau in these eyes. Also, patients who were followed up with visits at 2- to 6-month intervals with photos taken every 6 months were included to minimize missed DH during intervals.¹⁸ In our study, 47.6% of all DH cases were recurrent, which is similar to results from previous studies that had a follow-up period more than 4 to 5 years.^14,15 

Disc hemorrhage is a significant risk factor for the development and progression of glaucoma.^5,19 However, some controversy surrounds the clinical significance of recurrent DH. Siegner and Netland¹¹ reported no difference in the rate of progression of optic disc shape or VF defects in eyes with recurrent DH and nonrecurrent DH. Prata et al.²⁰ reported that DH recurrence was not a significant factor affecting the rate of VF progression in glaucoma patients with DH. Rasler et al.²¹ demonstrated no difference in VF progression between eyes with nonrecurrent and recurrent DH. The Ocular Hypertension Treatment Study found that eyes with recurrent DH showed no difference in global VF progression, defined by MD rates of change, compared to eyes with a nonrecurrent DH.¹⁹ On the other hand, Kim and Park¹⁴ and Laemmer et al.²² found significantly greater RNFL changes after DH, but not VF deterioration, in recurrent DH eyes. Ishida et al.¹³ reported that their recurrent DH group exhibited more pronounced progressive changes in the VF than the nonrecurrent DH group. However, all of these studies compared recurrent DH eyes with nonrecurrent DH eyes only regarding aspects of structural and functional deterioration. Thus, these contradictory results may be due to the varied characteristics of recurrent DH. 

More specifically, we found that recurrent DH can be classified into subgroups. First, cases of recurrent DH can be classified based on the site of recurrence. Recurrent DH could occur at one clock hour site or more than one clock hour site. Second, the location of recurrence can be the same as the initial DH location or different. Eyes with recurrent DH at different locations had features of “migrating,” implying that the DH locations move with the direction of RNFL defect progression, or recurring at totally different clock hour sites. By further classifying recurrent DH into subgroups based on these features, we found that the clinical significance varies based on the site of recurrence and based on the location. Eyes with recurrent DH at one clock hour or at more than one clock hour but recurring at the same location showed similar VF progression compared to eyes with a nonrecurrent DH. However, eyes with recurrent DH at one clock hour or at more than one clock hour but recurring at different locations from the initial DH location showed faster VF progression compared to nonrecurrent DH eyes or recurrent DH eyes at the same location. Approximately 80.4% of eyes with recurrent DH at the same location were frequently associated with a localized RNFL defect (representative case in Fig. 5). Eyes with a “migrating” DH showed enlargement of the RNFL defect with DH recurring at the margin of the RNFL defect moving in the direction of RNFL defect enlargement (representative case in Fig. 6). Eyes with DH recurring at different locations presented more frequently at more than one clock hour site (71.4%). This suggests that clinicians must pay attention to eyes with a DH at multiple clock hour sites, because DH of these eyes can recur at different locations from the initial DH and can progress more rapidly (representative case in Fig. 7). However, these two findings, that is, different locations and multiple sites, may not represent different clinical implications. The two findings may be linked together and their clinical implications may need further investigation. 

Figure 5

 

A representative case of a normal-tension glaucoma eye with disc hemorrhage (DH) recurring at more than one clock hour/same location. Recurrent DH occurred at the temporal and nasal borders of a localized retinal nerve fiber layer (RNFL) defect. Recurrent DH location was the same as the initial DH at each site. There was no enlargement or deepening of the localized RNFL defect or visual field progression after 54 months (mean deviation slope, −0.24 dB/y).

Figure 6

 

A representative case of a normal-tension glaucoma eye with recurrent disc hemorrhage (DH) at one clock hour/different locations. Recurrent DH occurred at the temporal border of a localized retinal nerve fiber layer (RNFL) defect, which moved in the temporal direction with the enlargement of the localized RNFL defect. Initial DH and the temporal border of the localized RNFL defect are shown in red. The next recurrent DH and the temporal border of the localized RNFL defect are color coded as orange → yellow → green → blue, and the final DH and the temporal border of the localized RNFL defect are shown in purple. The initial DH (red) and the final DH (purple) are more than one clock hour apart. A pattern of “migrating” recurrent DH is seen. Topographic Change Analysis of the Heidelberg Retinal Tomograph shows optic disc progression in the inferotemporal region (red area). There was significant visual field progression in this case after 52 months (mean deviation slope, −0.98 dB/y).

Figure 7

 

A representative case of a normal-tension glaucoma eye with recurrent disc hemorrhage (DH) at more than one clock hour/different locations. Recurrent DH occurred in the temporal border of the localized retinal nerve fiber layer (RNFL) defect in the inferotemporal region and the superotemporal region. Those in the inferotemporal region recurred at the same location as initial DH without progression of the localized RNFL defect. However, DH in the superotemporal region recurred at different locations from the initial DH location, showing enlargement of the RNFL defect toward the temporal direction as well as superior visual field (VF) defect progression. There was significant VF progression in this case after 55 months (mean deviation slope, −1.16 dB/y).

Subjects with recurrent DH at different locations were more likely to be taking aspirin. Systemic factors such as diabetes mellitus and systemic hypertension were also more frequent in such subjects, but the difference was not statistically significant. One study showed that use of aspirin and systemic hypertension were associated with DH in eyes with normal-tension glaucoma.²³ In our study, the use of aspirin was significantly different between the nonrecurrent and recurrent DH groups. This difference was more pronounced between the recurrent DH group at the same location (21.4%) and different locations (50.0%), occurring at a higher rate in the latter. 

The exact pathogenesis of DH is not well understood. One proposed mechanism for DH is mechanical rupture of the small blood vessels at the level of the lamina cribrosa.²⁴ One recent report showed that DH was more frequent in eyes with RNFL defect enlargement.⁹ The authors of that study speculated that enlargement of RNFL defect causes damage or loss of capillaries at the border of a RNFL defect. This may be explained by “collapse theory,” which refers to structural alteration resulting in mechanical rupture of the vessels leading to DH.²⁵ Eyes with recurrent DH at one clock hour or the same location may not progress faster because those hemorrhages are likely to occur due to ongoing collapse of neuroretinal tissue at the edge of the localized RNFL defects, which may be explained by collapse theory. In this case, DH may be a result of structural progression. However, in our study, recurrent DH at different locations was less associated with a localized RNFL defect compared to eyes with recurrent DH at the same location. Because these subjects had more frequent use of aspirin, there are possibilities that vascular effects may have a role in the pathogenesis of DH. Recurrent DH at more than one clock hour site or different locations may occur where neuroretinal rim is still present and adds another site of progressive glaucomatous damage. In this situation, DH may serve as a cause of progression and its pathogenesis may result from vascular insult to the optic disc, which implies “vascular theory.” There are hypotheses that DH may develop due to vascular effects including microinfarctions within the optic nerve head (ONH) and localized vascular insufficiency at the ONH.⁸ Vascular contributions, such as vasospasm, dysfunctional autoregulation of the blood flow to the ONH, or disrupted blood–retinal barrier found in patients with primary vascular dysregulation, may account for the pathogenesis of DH.^26–28 Further studies are needed to confirm our hypothesis and to elucidate its clinical significance in glaucoma progression. 

Our study has several limitations that must be acknowledged. This study was a retrospective design, and this serves as a major limitation. The frequency of follow-up would impact the detection of DH and glaucoma progression. Patients who are at greater risk could have been followed up more frequently. Eyes with DH may have received advance treatment after the detection of DH. Also, eyes with recurrent DH may have been followed up more closely. There is a report showing that rate of VF progression was different before and after DH, and this was related to the reduction of IOP in the posthemorrhage period due to enhanced treatment.²⁹ This may in part explain why the global rates of VF progression were not significantly different between eyes with nonrecurrent and recurrent DH. The number of glaucoma medications at the last follow-up visit was not different between nonrecurrent and recurrent DH groups in our study; however, this issue may have affected the results of this study. We included patients with minimum of six VFs, and these VF examinations were usually performed at 6- to 12-month intervals. More frequent VF testing using linear regression to detect VF progression is suggested. Chauhan et al.³⁰ have addressed this issue with mathematical modeling, and concluded that it is important to obtain a sufficient number of reliable VFs in the early follow-up.³⁰ However, even with six VF fields during the first 2 years, the VF slope estimates could be quite variable with regard to detecting progression.³¹ Gardiner and Crabbe³² showed that VFs more frequently performed than every 4 months would not lead to early detection of progression with linear regression. Considering these issues, we used global MD rates of change to compare the rate of progression between groups and the GPA program for survival analysis. Timely use of event analysis has been shown to decrease the number of VFs required to detect progression while at the same time preserving good specificity.^33,34 The minimum number of VFs required has been suggested to be six to eight using linear regression, while five VFs using GPA analysis was adopted in our study (minimum of six VFs).³⁵ More importantly, the number and sites of DHs may have been underestimated in some patients; we included only DHs that were detected. Usually DHs last for only approximately 6 to 12 weeks after the first presentation.² Intervals of photographs taken every 6 to 12 months could have misclassified eyes with recurrent DH as nonrecurrent DH, or our classification may be different from the real condition. However, we attempted to mitigate this by including only patients who had been followed up for at least 4 years. Also, if eyes with recurrent DH were misclassified into nonrecurrent DH due to underestimation, this may act to add faster-progressing eyes to the nonrecurrent DH group that may account unfavorable to our analysis. Many reports have shown that most DHs recur within two clock hours of the initial DH.^9,14–16 However, two clock hours from the initial DH resulted in a smaller number of DHs that recurred at different locations, so we modified this to a one clock hour difference so as to increase our sample size. 

In conclusion, we found that eyes with recurrent DH at more than one clock hour site and/or recurrent DH at different locations from the initial DH had more pronounced VF progression. This study classified recurrent DH eyes into subgroups, which explains the conflicting results in past studies on the clinical significance of recurrent DH. When eyes present with recurrent DH at more than one clock hour site and at different locations from the initial DH, more comprehensive diagnostic protocols and aggressive treatments may be needed to prevent the progression of glaucoma. 

Acknowledgments

Supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, Information, Communication and Technology (ICT), and Future Planning (MIST) (No. NRF-2014R1A1A3049403). The authors alone are responsible for the content and writing of the paper. 

Disclosure: H.-Y.L. Park, None; E.K. Kim, None; C.K. Park, None