K Sugiyama, K Ishida, H Uchida, T Yamamoto, Y Kitazawa
Department of Ophthalmology, Gifu University School of Medicine, Gifu, Japan

Optic disc haemorrhages in glaucoma usually appear as splinter or flame-shaped haemorrhages in the superficial nerve fibre layer on the optic nerve surface (figure 1). Disc haemorrhage was first described by Bjerrum in 1889.¹ In a later article, Bjerrum used the term 'glaucoma haemorrhagicum' to describe the coexistence of glaucoma with disc haemorrhage.² However, their possible association with progression of glaucoma had not been appreciated. The first report relating disc haemorrhage to progression of glaucoma was made in 1970 by Drance and Begg.³ These authors reported a small, flame-shaped haemorrhage on the optic disc of a 56-year-old woman with a glaucomatous optic disc and visual field changes that progressed after the haemorrhage had disappeared.

Figure 1. Disc haemorrhage located on the border between the retinal nerve fibre layer defect and the apparently healthy-looking retinal nerve fibre layer.¹³

Since Drance's report, both retrospective and prospective studies have shown the prevalence of disc haemorrhages. A higher prevalence in normal tension glaucoma (NTG) and a lower prevalence in a healthy population have been described by many investigators including our group.^4,5 Kitazawa et al. reported a 5-fold higher prevalence of disc haemorrhagesin normal tension glaucoma than in primary open angle glaucoma in a Japanese population (table 1).⁴

^{Table 1. Prevalance of disc haemorrhages in glaucoma (%)4}

Airaksinen et al. demonstrated that disc haemorrhages were often found in the inferotemporal sector of the optic disc, where they were twice as frequent as those located supero-temporally in 112 patients with open angle glaucoma (figure 2).⁶ In our clinic, 49.3% of disc haemorrhages occurred inferotemporally and 32.9% occurred superotemporally in patients with NTG.

Figure 2. Percentage distribution of 112 disc haemorrhages in 14 sectors around the optic disc. Half the haemorrhages were situated in the inferotemporal quadrant.⁶

However, there are conflicting reports regarding the prognostic significance of disc haemorrhage, and some researchers believe that disc haemorrhage is unrelated to visual field progression. The influence of disc haemorrhage on the glaucomatous process of optic nerve damage is still unclear. Moreover, neither the topographic relation of disc haemorrhages to retinal nerve fibre layer defects nor the association of disc haemorrhages with peripapillary atrophy have been precisely determined. Peripapillary atrophy is thought to be another risk factor in glaucomatous optic disc damage, and more prevalent in patients with NTG.⁷ In this article, we try to resolve the above-mentioned issues using studies performed in our clinic.

All patients with NTG in these studies were diagnosed in our clinic based on the following criteria:

• glaucomatous optic disc abnormalities and corresponding glaucomatous visual field defects
• normal open angle (greater than or equal to Shaffer's grade 2)
• intraocular pressure (IOP) less than or equal to 21 mm Hg without medication by Goldmann applanation tonometry, including during the 24-hour diurnal curve
• no pathology responsible for optic nerve damage by neuroradiological, rhinological, and general medical examinations including computed tomo-graphy (CT) and magnetic resonance imaging (MRI).

Disc Haemorrhage and Visual Field
Loss Progression

From a retrospective chart review of 465 consecutive patients with NTG at the Glaucoma Clinic of the Department of Ophthalmology, Gifu University Hospital from January 1985 to October 1998, we followed 70 patients with untreated NTG during a mean (± SD) follow-up period of 5.6 ± 2.3 years (range, 2 - 11.6 years). 32 patients had disc haemorrhage and 38 patients had no haemorrhage. We demonstrated that several clinical factors other than IOP were significantly associated with the progression of visual field loss during the natural course of NTG. By means of the Cox proportional hazards model, disc haemorrhage was identified to be the most risky factor associated with the progression of visual field loss in NTG (table 2).⁸

Table 2. Results of regression analysis of survival data based on the Cox proportional hazards model. Factors identified to be associated with visual field loss progression.¹

Abbreviations: DH=disc haemorrhage; CPSD=corrected pattern standard deviation, dB=decibel; bpm=beats per minute.

Our data revealed that a history of disc haemorrhage significantly increases the risk for progression of glaucomatous visual field defects measured by an automated field analyser (Humphrey Field Analyzer 630®, Zeiss Humphrey Systems, Dublin, California, USA, program 30-2). An eye with a history of disc haemorrhage can be expected to have a 20.3-fold greater chance of progression of visual field loss when mean deviation (MD) deterioration greater than 3.0 decibels (dB) demonstrated twice was defined as aggravation. An eye has a 3.3-fold greater chance of progression of visual field loss according to the following criterion: reproducible reduction in sensitivities of 10 dB in a cluster of 2 contiguous locations and/or deterioration of 5 dB in a cluster of 3 contiguous points, at least 1 of which was deteriorated by 10 dB compared with their baseline values.

Disc Haemorrhage and Localised
Wedge-Shaped Defect of Retinal
Nerve Fibre Layer

The essential pathological process in glaucoma is the loss of ganglion cell axons which results in localised or diffuse defects of the retinal nerve fibre layer. Jonas et al. reported that localised defects of the retinal nerve fibre layer were more frequently detected in NTG than in other types of glaucoma.⁹

Optic disc haemorrhages are often associated with rim notching or localised retinal nerve fibre layer defect at the site of bleeding. However, it remains unknown how frequently the localised damage to the retinal nerve fibre layer is associated with disc haemorrhage in patients with NTG. We prospectively investigated the frequency of localised wedge-shaped defects of the retinal nerve fibre layer in eyes with disc haemorrhage, using eyes without haemorrhage as controls.¹⁰

We studied 83 eyes of 83 patients with NTG, all of whom had developed new disc haemorrhages at the time of enrolment. We randomly selected 45 eyes of 45 patients with NTG with no history of disc haemorrhage during the follow-up period of > 2 years. There were no significant differences between the 2 groups in distributions of sex, age, refractive error, global indices of visual field (MD and CPSD), and the maximum, minimum, and variation of IOP in a 24-hour profile without medication (table 3).

Table 3. Patients' characteristics in normal tension glaucoma¹⁰

Abbreviations: IOP = intraocular pressure; ns = not significant; D = diopter; dB = decibel. Intraocular pressure was measured every 2 hours for 24 hours without medication. *Fisher's exact probability test, = Mann-Whitney U-test (mean ± standard deviation).

Figure 3. The image of the scanning laser ophthalmoscope showing splinter haemorrhage of the optic disc (large arrow) and a typical localised wedge-shaped defect of the retinal nerve fibre layer (arrows).¹⁰

Localised wedge-shaped defects of the retinal nerve fibre layer were identified by scanning laser ophthalmoscopy (SLO; Scanning Laser Ophthalmoscope®, Rodenstock GmBH, Munich, Germany) using an argon-blue laser (figure 3). One of the study group determined localised wedge-shaped defects of the retinal nerve fibre layer and disc haemorrhages from the SLO images. A localised wedge-shaped defect of the retinal nerve fibre layer was found in 72 of 83 eyes (86.7%) in the group with disc haemorrhage and 22 of 45 eyes (48.9%) in the group without haemorrhage. The difference was statistically significant (p < 0.0001, Fisher's exact probability test). Thus, more than 80% of eyes in the haemorrhage group had localised wedge-shaped defects of the retinal nerve fibre layer significantly more than in the non-haemorrhage group in patients with NTG. Moreover, within the same stage and pattern of visual field defect categories, the frequency of localised defects of the retinal nerve fibre layer was significantly greater in the disc haemorrhage group than in the group with no haemorrhage (table 4).

Table 4. Frequency of localised retinal nerve fibre layer defect by visual field stages and patterns of visual field defects in normal tension glaucoma.¹⁰

These results suggest that disc haemorrhage is closely associated with localised damage of the optic disc in patients with NTG. We demonstrated that disc haemorrhage is significantly associated with progression of visual field loss in patients with NTG.⁸ The higher frequency of localised defects of the retinal nerve fibre layer in eyes with disc haemorrhage may be responsible for progressive damage of the visual field over time in these eyes.