Fluorescein Filling Defects and Quantitative Morphologic Analysis ofthe Optic Nerve Head in Glaucoma

Objectives  To evaluate absolute filling defects of the optic nerve head in normaltension glaucoma (NTG) and primary open-angle glaucoma (POAG) and to comparethe filling defects with topographic analysis of the optic disc.

Methods  Twenty-five patients with NTG, 25 patients with POAG, and 25 age-matchedcontrols were included. Fluorescein angiograms were performed by means ofa scanning laser ophthalmoscope. The extent of absolute filling defects ofthe optic nerve head was assessed using digital image analysis of early-phaseangiograms. Topographic measurements of the optic disc were acquired usingthe Heidelberg Retina Tomograph II.

Results  Absolute filling defects were significantly larger (P = .001) and were seen more often (P<.001)in patients with NTG (n = 18) and POAG (n = 19) compared with controls (n= 3). Rim area (P = .006), rim volume (P = .007), cup-disc area ratio (P = .008),linear cup-disc ratio (P = .005), maximum cup depth(P = .002), cup shape measure (P = .03), and nerve fiber layer thickness (P =.008) and cross-sectional area (P = .006) were significantlydifferent between patients with glaucoma and controls. Absolute filling defectswere significantly correlated with cup area (r =0.31; P = .007), rim area (r =−0.38; P<.001), rim volume (r = −0.35; P = .002), cup-disc arearatio (r = 0.49; P<.001),linear cup-disc ratio (r = 0.48; P<.001), cup shape measure (r = 0.27; P = .02), and nerve fiber layer thickness (r = −0.33; P = .004) and cross-sectionalarea (r = −0.30; P =.009).

Conclusions  Fluorescein filling defects of the optic disc are present in NTG andPOAG. The extent of these filling defects is correlated with the morphologicdisc damage.

The pathogenetic concepts of glaucoma, defined as a progressive opticneuropathy characterized by optic nerve head excavation and glaucomatous visualfield loss, include mechanical and vasogenic mechanisms.¹ Avascular failure leading to perfusion deficits of the optic nerve head, retina,choroid, or retrobulbar vessels, by means of vasosclerosis, small vessel disease,vasospasms, or autoregulatory dysfunction, may contribute to the nerve fiberloss in glaucomatous optic neuropathy.^2-5 Themechanical damage is regarded as intraocular pressure (IOP)–dependentaxonal dysfunction and loss. The lamina cribrosa and changes in extracellularmatrix seem to have a substantial effect on the mechanical damage.^6,7

Fluorescein angiographic studies may describe perfusion alterationsof the optic nerve head, retina, and choroid. In different studies,^8-16 morphologicand dynamic perfusion variables demonstrated impaired ocular blood flow inglaucoma.

Fluorescein filling defects of the optic nerve head are areas of hypoperfusion,and they have been described in glaucomatous optic neuropathy since the 1970s.^8,12,17-29 Absolutefilling defects are persistent hypofluorescent areas, and they seem to correspondto capillary dropout in the surface nerve fiber layer of the optic disc.^{8,12,13,20,23,29} Incontrast, relative defects are areas of delayed fluorescence, and they showa slower filling pattern with fluorescein.^8,12,20 Thefilling defects are interpreted as areas of hypovascularity, as they are reproducible,with no consistent correlation with IOP and systemic blood pressure.^8,10,12,13

The number, extent, and topography of fluorescein filling defects correspondto visual field loss, nerve fiber layer defects, and cupping in glaucoma.^17,20,26-32 Absolutefilling defects are larger and of greater number in patients with glaucomacompared with those with ocular hypertension or controls.^20,21,29,32 Severalinvestigators^8,12,33 havereported high specificity of filling defects for glaucoma and anterior ischemicoptic neuropathy. Furthermore, the regions of pallor of the disc in opticatrophy seem to result from alterations in the tissue reflectance after axonalloss and from alterations in extracellular matrix and glial tissue ratherthan from a decrease in microvascular structures.^34-37 Incontrast, O'Day et al²⁴ also reported decreasedfluorescence in different types of optic atrophy. In glaucoma, fluoresceinfilling defects of the optic disc are preferentially located at the marginof the optic disc excavation, mainly inferotemporally and superotemporally,and are more often found at the wall than at the floor of the cup.^{8,12,13,30,38} Severalresearchers^8,12,30 emphasizethe relevance of fluorescein filling defects in glaucomatous optic neuropathyand postulate that filling defects may be the initial damage in glaucoma.

The Heidelberg Retina Tomograph II (HRT II) (Heidelberg Engineering,Heidelberg, Germany) is a confocal scanning laser ophthalmoscope for quantitativestereometric analysis of the optic nerve head.^41-43 Thescanning laser technique allows for 3-dimensional assessment of the opticdisc based on a digital image of its surface. Differentiation of the neuroretinalrim and the optic nerve head cup requires an operator-dependent contour line–basedstandard reference plane.^41-43 Moststereometric variables depend on this reference plane.^41-46 TheHRT II aims to detect glaucomatous optic disc appearances and structural changesin the retinal nerve fiber layer in glaucoma. Morphologic assessment of theoptic disc in detecting early glaucoma may improve diagnostic reliabilityif structural damage precedes functional damage, as measured by conventionalwhite-on-white perimetry.

The purpose of this study is to investigate the correlation betweenhypofluorescent areas of the optic disc and morphologic damage in glaucomatousoptic neuropathy. Absolute fluorescein filling defects of the optic disc arecompared with stereometric variables of the optic nerve head, as measuredby confocal scanning laser tomography (HRT II). The filling defects of theoptic disc and its stereometric variables are evaluated in patients with normaltension glaucoma (NTG), patients with primary open-angle glaucoma (POAG),and controls. The filling defects and morphologic variables of the optic nervehead are compared among groups and correlated with each other.

Twenty-five patients with NTG, 25 patients with POAG, and 25 age-matchedcontrols are included in this prospective clinical study. For statisticalanalysis, 1 eye of each participant was randomly chosen. All individuals,including control subjects, provided informed consent. Adherence to the Declarationof Helsinki for research involving human subjects is confirmed.

Patients with NTG and POAG had glaucomatous optic nerve head cuppingand glaucomatous visual field defects as defined by the European GlaucomaSociety in the absence of retinal or neurologic disease affecting the visualfield. The diagnostic criteria for glaucomatous visual field loss are as follows.Field loss was considered significant when (1) glaucoma hemifield test resultswere abnormal, (2) 3 points were confirmed with P<.05probability of being normal (one of which should have P<.01), not contiguous with the blind spot, or (3) the correctedpattern SD was abnormal with P<.05.⁴⁷ Allvariables were confirmed on 2 consecutive visual field examinations performedusing the Humphrey visual field analyzer (model 750; Humphrey-Zeiss, San Leandro,Calif) (full-threshold program 24-2).

All patients with glaucomatous visual field loss underwent diurnal curvesof IOP measurements (Goldmann applanation tonometry) at 8 AM,noon, 4 PM, 8 PM, and midnight without any topicalor systemic IOP-lowering medication. In patients with NTG, IOP never measuredgreater than 21 mm Hg.

Visual acuity was 20/40 or better, and no previous laser or surgicaltreatment had been performed. Patients with refractive aberrations of morethan ±4 diopters, diabetes mellitus, and hypersensitivity to sodiumfluorescein were excluded from this study.

Control subjects had no history of ophthalmologic disease. Automaticstatic white-on-white and short-wavelength automated perimetry did not revealsubstantial visual field loss. Nerve fiber layer imaging using a scanninglaser ophthalmoscope (SLO; Rodenstock, Ottobrunn, Germany) with blue light(argon-blue 488 nm) indicated a regular nerve fiber layer structure withoutany nerve fiber bundle defects.

No statistically significant differences between patients with NTG andPOAG were found for age, refraction, systolic and diastolic blood pressure,and heart rate. Patients with POAG had a significantly higher IOP comparedwith patients with NTG (P = .02). Seventeen patientswith POAG, 16 patients with NTG, and 16 controls had a history of systemiccardiovascular disease, including arterial hypertension, treated with systemicmedications. The mean number of local IOP-lowering medications was 1.48 forPOAG, 0.64 for NTG, and 0.24 for controls. Six controls were initially treatedas patients with NTG but were later reevaluated as controls with physiologicexcavation of the optic nerve head. The clinical and demographic characteristicsof all individuals included in the study are given in Table 1.

Patients with POAG, patients with NTG, and control subjects underwenta detailed ophthalmologic examination, videofluorescein angiography usingthe SLO, and a scanning laser tomographic examination using the HRT II.

Fluorescein angiography of the optic nerve head was performed usingthe SLO. The confocal video scanning laser ophthalmoscope, with a resolutionof 512 × 512 pixels, detects temporal high-resolution images with highfrequencies (25 Hz). To visualize the capillary network of the optic nervehead, the 20° field of observation of the SLO was used. Videofluoresceinangiograms permit the selection of images with the best possible visualizationof the superficial capillaries.^15,16,29 Tostart the angiography, 10% sodium fluorescein dye (2.5 mL) was injected intoan antecubital vein. The videofluorescein angiograms were performed with theoptic nerve head centered. Images of the early phase (<3 minutes) weredigitized visualizing the superficial capillaries of the optic nerve head.The angiograms were analyzed offline using digital image analysis (MatroxInspector; Matrox Electronic Systems Ltd, Dorval, Quebec). The extent of absolutefluorescein filling defects was measured in relation to the area of the opticnerve head (percentage of the optic disc). Absolute filling defects of theoptic nerve head are defined as areas of persistent hypofluorescence duringthe whole angiogram. During the angiogram, the focus was changed from theneuroretinal rim to the bottom of the cup to avoid artefacts. For evaluationof the hypofluorescent areas of the optic nerve head, the digitized singleimages were analyzed in a masked manner. Three observers (N.P., A.R., andO.A.) measured the extent of the absolute filling defects in agreement. Asa reference for the disc area, digitized red-free images (argon laser 488-nmSLO) of the optic nerve head were used.

Systolic and diastolic blood pressure and heart rate were measured aftera 5-minute rest in the sitting position before fluorescein angiography. Nervefiber layer imaging with a blue laser (argon laser 488 nm) was performed.Nerve fiber layer defects confirmed the diagnosis of POAG or NTG.

Visual field examinations were performed using the Humphrey visual fieldanalyzer and the white-on-white 24-2 full-threshold program. The standardvisual field variables of mean deviation, pattern standard deviation, short-termfluctuation, and corrected pattern standard deviation were used for diagnosisand statistical analysis.

All patients and control subjects underwent confocal scanning lasertomography of the optic nerve head using the HRT II (software 2.01). The HRTsoftware analyzes the mean topography of 3 consecutively performed confocalscanning laser images of the optic disc. The variability of the scanning imagesis expressed by the standard deviation of the topography. The border of theoptic nerve head at the level of the Elschnig scleral ring was outlined manuallyby an experienced examiner (N.P.). Depending on this operator-based contourline, the HRT II software 2.01 calculates a reference plane delineating theneuroretinal layer from the optic cup. Most of the stereometric variablesdescribing the optic nerve head depend on this reference plane. The contourline–based reference plane is located perpendicular to the z-axis, 50µm below the contour line at 354° to 360° of the optic nervehead circumference. Magnification error was corrected using keratometry valuesfor each individual. For statistical analysis, the following variables weredetermined: disc area, cup area, rim area, cup volume, rim volume (area aboveand volume below the reference plane), cup-disc area ratio, linear cup-discratio, mean cup depth, maximum cup depth, cup shape measure (the third momentof the frequency distribution of depth values relative to the contour line),height variation contour (maximum minus minimum of the relative height valuesof the contour line), nerve fiber layer thickness, and nerve fiber layer cross-sectionalarea (the calculated distance and area between the reference plane and thecontour line). Disc area, mean and maximum cup depth, height variation contour,and cup shape measure are independent of the selection of the reference plane.

The fluorescein filling defects and the stereometric variables of theoptic disc were compared among groups using analysis of variance. Correlationswere tested using the Fisher r to z test. In all analyses, P<.05 was regardedas statistically significant.

Patients with POAG and NTG more often had absolute filling defects ofthe optic nerve head compared with controls. Absolute filling defects werepresent in 18 of the 25 patients with NTG, 19 of the 25 patients with POAG,and only 3 of the 25 controls (P<.001). The diagnosticvalidity to differentiate patients with glaucoma from controls was expressedas a specificity of 88% and a sensitivity of 74%. The absolute filling defectsof the optic nerve head were significantly larger in patients with NTG andPOAG compared with controls (P<.01). The extentof the filling defects was not significantly different in POAG and NTG (P = .91) (Table 2).

The following stereometric variables of the optic nerve head measuredby confocal scanning laser ophthalmoscopy (HRT II) differed significantlybetween patients with glaucoma (POAG and NTG) and controls. Patients withglaucoma had smaller neuroretinal rim areas and rim volumes and larger cup-discarea ratios and linear cup-disc ratios. The maximum cup depth was smallerin patients with glaucoma. The cup shape measure showed significantly lessnegative values in the glaucoma groups, and the nerve fiber layer thicknessand cross-sectional area were smaller. No significant difference between patientswith glaucoma and controls was found for disc area, cup area, cup volume,mean cup depth, and the standard deviation of the calculated mean topographyof the optic disc. The only variable found to differ significantly betweenpatients with POAG and NTG was the height variation contour (P<.05). The results are given in Table 2.

Further analysis was performed to investigate correlations between fluoresceinfilling defects and stereometric variables of the optic disc. For all participantsincluded in this study, the absolute filling defects were significantly correlatedwith cup area (r = 0.31), rim area (r = −0.38), rim volume (r = −0.35),cup-disc area ratio (r = 0.49), linear cup-disc ratio(r = 0.48), cup shape measure (r = 0.27), and retinal nerve fiber layer thickness (r = −0.33) and cross-sectional area (r =−0.30) (P<.05 for all) (Table 3). No correlations were found for disc area, cup volume,mean and maximum excavation depth, and height variation contour. For patientswith POAG, the filling defects were significantly correlated with cup-discarea ratio (r = 0.43) and linear cup-disc ratio (r = 0.43) (Table 4).In NTG, the filling defects were significantly correlated with rim area (r = −0.51), rim volume (r =−0.51), cup-disc area ratio (r = 0.55), linearcup-disc ratio (r = 0.55), maximum cup depth (r = −0.40), and nerve fiber layer thickness (r = −0.49) and cross-sectional area (r = −0.48) (Table 4).Controls exhibited a significant correlation of the filling defects with themaximum excavation depth only (r = 0.58; P<.01). None of the other stereometric variables of the controlswere statistically significantly correlated with the extent of the fillingdefects.

Absolute fluorescein filling defects of the optic nerve head were statisticallysignificantly larger and were seen more often in patients with POAG and NTGcompared with controls. These results confirm findings of various previousstudies.^{8,12,20,21,29,32} Inthe present study, filling defects of the optic disc as a tool of differentiationbetween patients with glaucoma and controls had a specificity of 88% and asensitivity of 74%. The filling defects observed on fluorescein angiogramsreflect an area of hypoperfusion of the superficial nerve fiber layer of theoptic nerve head in glaucomatous optic neuropathy and seem to correspond tocapillary dropout.^8,12,29

Several histologic studies examined changes in the vascular structureof the optic nerve head in glaucoma to study interference of capillary lossand morphologic change of the optic disc. In experimental nonglaucomatousoptic atrophy, the number of capillaries remained stable and was expressedas a ratio to the optic nerve tissue. The size and relative volume of thecapillaries diminished, whereas fluorescein angiography did not alter.^36,37,48 The studies of experimentaloptic disc pallor showed a rearrangement of astrocytes beside ganglion nervefiber loss. Quigley and Anderson³⁷ and Radiusand Maumenee⁴⁸ interpreted these findings ascausative for optic disc pallor rather than the vascular alterations becausein complete ganglion cell loss, capillaries are still present in a pale opticdisc, and fluorescein angiography of the optic disc was not altered. Furthermore,optic atrophy of various causes, including ischemia, is rarely combined withan increasing cup-disc ratio, as in glaucomatous optic nerve degeneration.⁴⁸ Sebag et al^49-51 foundreduced blood volume (approximately 50%) and oxygen delivery (approximately40%) using vessel oxymetry, laser Doppler technique, and disc reflectometryin experimental optic atrophy. The Doppler measurements were substantiatedby histologic studies of microsphere distribution (decrease of 80% in flowin anterior optic atrophy). Again, no abnormalities were detected by fluoresceinangiography.^49-51 Incontrast to glaucomatous optic atrophy, optic pallor in optic atrophy seemsto result from ganglion cell loss, astrocyte rearrangement, or reduced bloodvolume or oxygen content, alterations not seen with fluorescein angiography.

In glaucomatous optic neuropathy, Elschnig,⁵² Cristini,⁵³ and François and Neetens⁵⁴ founda reduced capillary network in the optic nerve head and choriocapillaries.These qualitative studies^52-54 emphasizedcapillary rarefaction in glaucomatous optic neuropathy. Kornzweig et al⁵⁵ and Alterman and Henkind⁵⁶ describedselective atrophy in radial peripapillary capillaries in postmortem eyes withchronic glaucoma and in experimental glaucoma. Quigley et al,^36,37,57 however,stated that in their quantitative histologic studies of experimental glaucoma,capillary atrophy paralleled the nerve fiber loss. They found a stable capillary-tissueratio in glaucomatous optic neuropathy and could not detect early damage angiographically.Consequently, these alterations in the capillary network of the optic nervehead were assumed to be secondary to nerve fiber tissue loss.^36,37,57

In contrast, clinical studies^8,12,29 dealingwith fluorescein angiographic filling defects of the optic nerve head statedthat the filling defects, at least in some cases, precede morphologic damage,and vertical studies^25,29,32 revealeda strong interrelationship to functional defects. The few longitudinal follow-upstudies available concluded that filling defects emphasized progressive opticnerve damage,³⁹ and new field defects wererelated to new filling defects in glaucomatous optic neuropathy.⁴⁰

In the present study, the filling defects were correlated with the stereometricvariables of the optic nerve head measured by scanning laser ophthalmoscopy.The statistically significant correlation with cup-disc area ratio and linearcup-disc ratio confirmed previous findings, even more so since in controlsno such correlation was found. The filling defects were positively correlatedwith cup area and cup shape measure and negatively correlated with rim volumeand nerve fiber layer thickness and cross-sectional area. In NTG, we founda statistically significant correlation with various variables, although inPOAG the filling defects were only statistically significantly correlatedwith cup-disc area ratio and linear cup-disc ratio. As patients with POAGand NTG did not differ in the extent of the filling defects and all stereometricvariables except height variation contour, this may reflect a stronger relationof vascular and morphologic damage of the optic nerve head in NTG. The questionof whether NTG refers to a single disease entity or to a subgroup of open-angleglaucoma with lower tolerance to IOP must be mentioned again. The extent andincidence of fluorescein filling defects of the optic nerve head did not differin POAG and NTG in the presented study. Therefore, the concept of ischemicdamage of the optic nerve head (ie, capillary dropout) in glaucomatous opticneuropathy seems to be applicable to both types of glaucoma. Whether capillarydropout of the optic nerve head precedes or follows neuronal loss needs tobe clarified in longitudinal follow-up studies.

Few studies investigated blood flow variables compared with stereometricvariables of the optic nerve head or functional data. Ciancaglini et al⁵⁸ found a significant correlation between blood flowvariables measured using laser Doppler flowmetry and nerve fiber layer variablesof scanning laser ophthalmoscopy. Kuba et al⁵⁹ couldnot find such correlation comparing laser Doppler flowmetry and scanning laserpolarimetry. However, these studies have methodological limitations, as thedependence of blood flow measurement by scanning laser Doppler flowmetry onnerve fiber layer structures and the tissue volume included in the analysisremains unclear. Arend et al¹⁴ investigatedfluorescein angiograms of patients with NTG and asymmetrical visual fieldloss. The altitudinal visual field defects were associated with prolongedarteriovenous passage time.¹⁴ The fluoresceinfilling defects were highly correlated with the visual field testing in variousstudies.^{8,12,17,26,29} Ina blue field entoptic phenomenon approach, Sponsel et al⁶⁰ measuredhigher leukocyte velocities in eyes with better visual function, as expressedby the global index mean deviation. In a study by Fontana et al,⁶¹ pulsatileocular blood flow was lower in NTG eyes with field loss compared with thecontralateral normal visual fields. Ciancaglini et al⁶² founda correlation between laser Doppler flowmetry variables of the lamina cribrosaregion and visual field defects, although no correlation was found for theneuroretinal rim. In an indocyanine green fluorescence angiography study bySato et al,⁶³ the watershed zones includingthe optic nerve head were associated with larger field defects in NTG.

In summary, this study implies a relationship between morphologic damageand superficial capillary loss observed in fluorescein angiography of theoptic nerve head in glaucoma. Longitudinal studies are needed to clarify whetherthis capillary dropout is primary or secondary to nerve fiber tissue lossor functional defects, as clinical and histologic studies revealed differentresults in the past.

Corresponding author and reprints: Niklas Plange, MD, Augenklinikdes Universitätsklinikum Aachen, Pauwelsstr. 30, 52057 Aachen, Germany(e-mail: nplange@ukaachen.de).

Submitted for publication April 10, 2003; final revision received August27, 2003; accepted September 10, 2003.

This study was presented in part at the Deutsche Ophthalmologische GesellschaftCongress 2002; September 27, 2002; Berlin, Germany.