Spatial Pattern of Glaucomatous Visual Field Loss Obtained with Regionally Condensed Stimulus Arrangements

Rectangular (6° × 6°) test point arrangements in perimetry (e.g., grid 24-2 or grid 30-2) omit the immediate paracentral visual field region. Perimetry with regionally condensed stimulus arrangements reveals that glaucomatous visual field loss affects the superior paracentral area within an eccentricity of 3° in more than 50% of the eyes with predominantly mild to moderate glaucomatous visual field loss.

Abstract

Purpose.

To assess the spatial distribution of glaucomatous visual field defects (VFDs) obtained with regionally condensed stimulus arrangements.

Methods.

Sixty-three eyes of 63 glaucoma subjects were examined with threshold-estimating automated static perimetry (full threshold 4-2-1 dB strategy with at least three reversals) on an automatic campimeter or a full-field perimeter. Stimuli were added by the examiner to regionally enhance spatial resolution in regions that were suspicious for a glaucomatous VFD. These regions were characterized by contiguous local VFDs, attributable to the retinal nerve fiber bundle course according to the impression of the examiner. The added stimulus locations were subsets of a predefined, dense perimetric grid. All VFD locations with P < 0.05 (total deviation plots) were assessed by superimposing the visual field records of all participants.

Results.

Glaucomatous VFD loss occurred more frequently in the upper than in the lower hemifield, with a typical retinal nerve fiber-related pattern and a preference of the nasal step region. More than 50% of the eyes with predominantly mild to moderate glaucomatous field loss showed defective locations in the immediate superior paracentral region within an eccentricity of 3°.

Conclusions.

Conventional thresholding white-on-white perimetry with regionally enhanced spatial resolution reveals that glaucomatous visual field loss affects the immediate paracentral area, especially the upper hemifield, in many eyes with only mild to moderate glaucomatous visual field loss. Detailed knowledge about the spatial pattern and the local frequency distribution of glaucomatous VFDs is an essential prerequisite for creating regionally condensed stimulus arrangements for adequate detection and follow-up of functional glaucomatous damage.

Typical glaucomatous visual field loss is characterized by arcuate defects, nasal steps, and other patterns corresponding to the course of retinal nerve fibers that respect the nasal horizontal meridian and usually spare the visual field center.^1–5 Damage to the immediate paracentral visual field, leading to a so-called split fixation, is usually understood as a sign of advanced loss in cases of end-stage glaucomatous damage.⁶

In conventional perimetry, rectangular test point arrangements with a spacing of 6° × 6° (e.g., grid 24-2 or 30-2) that omit the 6° central visual field regions are most frequently used and, therefore, lack detailed spatial information.

The aim of this study was to obtain a more detailed pattern and a spatial frequency distribution of glaucomatous visual field loss by using locally condensed stimulus arrangements in regions of manifest or suspected visual field loss.

Subjects and Methods

Sixty-three eyes of 63 glaucoma subjects (34 women, 29 men; age range, 33–79 years; mean defect (MD), −11.4 to +0.5 dB; Aulhorn stage 1, 24 eyes; Aulhorn stage 2, 18 eyes; Aulhorn stage 3, 19 eyes; Aulhorn stage 4, two eyes) were examined with automated static perimetry (full threshold 4-2-1 dB strategy with at least three reversals) on an automatic campimeter (Tuebingen Computer Campimeter [TCC]) or on a full-field perimeter (Octopus 101-Perimeter; Haag-Streit Inc., Koeniz, Switzerland).

Patients met the inclusion criteria for high-pressure open-angle glaucoma (n = 39, aged 35 years or older; typical signs of glaucomatous optic neuropathy or retinal nerve fiber loss and typical visual field loss with IOP ≥22 mm Hg) or low-tension glaucoma (n = 24; typical signs of glaucomatous optic neuropathy or retinal nerve fiber loss and typical visual field loss with IOP values <22 mm Hg at any time; best-corrected visual acuity had to be equal to or better than 10/20). Visual field defect (VFD) was defined as at least three non-edge test locations^7,8 that had to be located within the superior/inferior hemifield^9–11 and had to be depressed to the 5% probability level (P < 0.05), with at least one non-edge point depressed to the 1% probability level (P < 0.01), according to the pattern deviation plot.⁷

Exclusion criteria were relevant opacities of the central refractive media and ophthalmologic diseases other than glaucoma that might interfere with the visual field.

MD values, as determined with the initial test point arrangement, ranged from −11.4 to +0.5 dB. The frequency distribution of MD and of the visual field stages according to the classification system of Aulhorn and Karmeyer¹² are shown in Figure 1. Sixty-two subjects had a best-corrected distant visual acuity above 10/20, and one subject had a visual acuity of 10/20. The study was approved by all local institutional review boards and adhered to the tenets of the Declaration of Helsinki.

Figure 1.

(A) Frequency distribution of the perimetric glaucoma stages of the 63 eyes of 63 patients according to the classification system of Aulhorn and Karmeyer.¹² (B) Frequency distribution of the MD.

Visual fields were initially assessed with automated static perimetry (full-threshold 4-2-1 dB strategy with at least 3 reversals) on the TCC with a 6° × 6° rectangular grid (77 test locations; Fig. 2A) or on the full-field perimeter (Octopus 101-Perimeter; Haag-Streit Inc.) on the 30-A grid (comprising 83 test locations in a polar arrangement within the 30° visual field that respected the horizontal and vertical meridians and that were more condensed toward the visual field center; Fig. 3). In subsequent sessions, the operator added stimuli in suspicious regions around the VFD to enhance the spatial resolution within these areas. Suspicious regions were defined as a cluster of at least three non-edge test locations that were suspicious for a glaucomatous VFD.^7,13 They had to be located within the superior/inferior hemifield^10,11 and had to be depressed to the 5% probability level (P < 0.05), with at least one non-edge point depressed to the 1% probability level (P < 0.01), according to the pattern deviation plot.⁷ One physician (US) demarcated the region of test point condensation with the help of a lasso tool. The line was placed around the cluster of pathologic visual field locations. This procedure is now automated in the most recent version of the condensation procedure (see Discussion).

Figure 2.

(A) Threshold-estimating static perimetry with regional stimulus condensation in the superior paracentral visual field clearly demarcates a circumscribed paracentral small retinal nerve fiber–related scotoma corresponding to a previous splinter hemorrhage shown in the (inset) optic disc photograph (the optic disc is turned upside down). Circles: rectangular 6° × 6° grid. (B) In the corresponding Humphrey 30-2 visual field, only one pathologic location was detected within the paracentral nasal superior quadrant.

Figure 3.

Threshold-estimating static perimetry with regional stimulus condensation in the superior paracentral visual field clearly demarcates a circumscribed paracentral small retinal nerve fiber-related scotoma. Circles: 30-A grid with polar test point arrangement. Only three pathologic locations were detected without condensed stimulus arrangement (red circles).

The added stimulus locations were subsets of a predefined, dense, perimetric grid (total of 191 locations within the 30° field) in a polar arrangement (Fig. 4). The stimulus locations respected the horizontal and vertical meridians and were more condensed toward the visual field center.

Figure 4.

Spatial frequency distributions of all glaucomatous visual field defects with P < 0.05, according to the individual total deviation plots. This result was assessed by electronically superimposing the visual field records of all participants. Local frequency values in which more than 50% of the eyes showed P < 0.05 are shaded in light gray to highlight the pattern distribution of glaucomatous visual field loss in the upper and lower hemifields.

The visual field results from the left eye format were converted into right eye format. Spatial frequency distributions of all visual field defect locations with P < 0.05, according to the total deviation plots, were summed across all participants to determine the percentage of eyes with a VFD at each location.

Results

As expected, maximum spatial frequency counts (above 90%) occurred in the blind spot area and in the upper rim region of the visual field, exceeding 25° of eccentricity (Fig. 4). Areas with local frequency values that exceeded 50% were shaded in gray for better visibility. In general, the area and extent of areas shaded in gray were greater in the upper than in the lower hemifield. Locations exceeding a local frequency of 50% occurred in the upper and lower hemifield, following an arcuate retinal nerve fiber–related pattern, with a regional preference of the (upper) nasal quadrant/nasal step. Areas with a local frequency exceeding 50% spared the lower paracentral hemifield and clearly involved its counterpart in the upper paracentral region; more than 50% of the eyes showed defective locations in the immediate superior paracentral region within an eccentricity of 3°. Locations within the temporal quadrant were rarely involved.

Two typical clinical findings with regionally enhanced test point condensation are demonstrated in Figures 2 and 3. The field record with the regionally condensed grid shows a deep circumscribed arcuate defect with connection to the blind spot area that affects the immediate superior paracentral region. The defect in Figure 2 was classified as Aulhorn stage 3¹² with the condensed test grid. Results of conventional thresholding perimetry, based on a rectangular 6° × 6° test point arrangement, would have been classified as Aulhorn stage 1 and classified as “early” (MD −0.37 dB) according to the classification system of Hodapp et al.¹⁴ or its more recent modification according to the classification system of Mills et al.¹⁵ Similarly, the example in Figure 3 shows a glaucomatous scotoma staged as Aulhorn 3 with the condensed grid and a scotoma staged as Aulhorn 2 with the original grid 30-A (polar test point arrangement).

Discussion

Previous pattern analyses based on rectangular grids primarily described the arcuate pattern of the field loss and the loss in the nasal horizontal meridian (nasal step).^2,3,14–24 Thus far, only a few studies have analyzed glaucomatous scotoma patterns with spatially high-resolution stimulus arrangements.^25–30 Previous pattern analyses based on static perimetry with enhanced test point density referred to the results of supraliminal static perimetry and, therefore, might have overlooked shallow visual field defects.³¹ Other results with a polar test point arrangement, similar to the one applied in this study, have already demonstrated that glaucomatous visual field loss occurs preferentially in the upper hemifield and affects the paracentral regions in a considerable number of cases.^{1,2,19,32–37}

We chose to grade the severity of glaucomatous field loss according to the classification system of Aulhorn and Karmeyer¹² because it primarily considered scotoma shape and extent. Scoring systems such as the Hodapp et al.¹⁴ algorithm and its modification by Mills et al.¹⁵ could not be applied in this case because these are based on rectangular test point arrangements. Glaucomatous visual field loss approaching the visual field center in only one hemifield is also called split fixation⁶ and has to be rated as a serious event or an impairment. Paracentral regions of the retina are used for reading^38–42 and numerous activities of daily living, such as driving and other steering, monitoring, or surveillance tasks.^43–49 Involvement of the paracentral region, especially in case of local overlap of binocular visual field defects, is often accompanied by serious impairment of quality of life.^{35,42,45,46,48,50–53} Deva et al.⁵⁴ found a paracentral visual field defect in 56% of their 107 glaucoma patients, however, applying a 24-2 grid.

Threshold assessment in perimetry is characterized by an unstable outcome with regard to defect depth. Even in normal visual field areas, variations approaching 3 dB (factor of 0.5–2.0 with regard to the local luminance value) are rated as stable.^55–57 This critically interferes with detection of change in follow-up analyses and can be only partially fixed by repeated perimetric sessions at baseline and follow-up. In contrast, we are not aware of any study demonstrating fluctuations of that magnitude with regard to defect size in glaucomatous visual field loss.

Local enhancement of test point condensation is not only an option for more exact delineation of the present scotoma pattern, it may also provide the information needed to detect progression earlier than can be detected with conventional perimetric grids.^58,59

A method for a computer-based, automated condensation of test locations (autoSCOPE [automated SCotoma-Oriented Perimetry]) has recently been developed and was presented at the 2010 meeting of the Imaging and Perimetric Society (Dietzsch J, et al. autoSCOPE: an algorithm for automated regional condensation of stimulus density for polar and rectangular perimetric grids).

In conclusion, conventional thresholding perimetry with enhanced spatial resolution reveals that glaucomatous visual field loss affects the immediate pericentral area, especially the upper hemifield, in many eyes with only mild to moderate glaucomatous visual field loss. Detailed knowledge about the spatial pattern and the local frequency distribution of glaucomatous visual field defects is an essential prerequisite for creating regionally condensed stimulus arrangements for adequate detection and follow-up of functional glaucomatous damage.