GLAUCOMA

Relationship between disc margin to fovea distance and centralvisual field defect in normal tension glaucoma

Marvin Lee & Hosung Jin & Jaehong Ahn

Received: 24 May 2013 /Revised: 30 September 2013 /Accepted: 24 October 2013 /Published online: 22 November 2013# Springer-Verlag Berlin Heidelberg 2013

AbstractBackground To investigate the relationship between oculargeometric factors, including temporal disc margin to foveadistance (DFD) measured by optic disc stereophotography(ODP) and central visual field (VF) defect, in normal-tension glaucoma (NTG) patients.Methods This retrospective, single-center, cross-sectionalstudy included 88 eyes of 88 NTG patients with mild VFdefects (MD>−6.0 dB). NTG patients were divided into twogroups according to VF tests: central VF-invading and centralVF-sparing groups. Optic nerve head (ONH) parameters in-cluding disc dimensions, peripapillary atrophy (PPA), andDFD were obtained by ODP, and retinal nerve fiber layer(RNFL) thickness was measured by Stratus optical coherencetomography (OCT).Results In the invading group, DFD was shorter (3.642±0.401 mm) than in the sparing group (3.877±0.278 mm; p =0.002). The sparing group had more vertically oval ONH (p=0.023) and wider temporal PPAwidth (p =0.031). The RNFLthickness in the invading group was thinner in the temporaland inferior quadrants, but thicker in the superior quadrantthan that of the sparing group. In a multiple linear regressionanalysis, DFD was the only geometric factor associated withdegree of central VF involvement (p =0.002). DFD was pos-itively correlated with temporal RNFL thickness in the sparinggroup (r =0.484, p<0.001) but not in the invading group (r =−0.080, p =0.631).Conclusions Eyes with a shorter DFD should be monitoredcarefully because central VF involvement appears to be relat-ed to shorter DFD in NTG patients with mild VF defects.

Keywords Normal tension glaucoma . Discmargin to foveadistance . Paracentral scotma . Optic disc stereophotography

Introduction

Visual field (VF) defects in glaucoma usually begin in thearcuate area, corresponding to retinal nerve fiber layer(RNFL), merged into superior and inferior quadrant of opticnerve head (ONH), most probably due to structural vulnera-bility of the lamina cribrosa in these sectors [1, 2]. If the site ofdamage is near the area corresponding to the insertion ofpapillomacular bundle (PMB) into the ONH, paracentral sco-toma (PCS) close to the fixation can be observed in the VFtest. A central VF defect can profoundly influence dailyactivities in glaucoma patients [3–5]. If initial PCS progressesfurther, to decreased foveal function, it can be a significantrisk factor for visual loss after glaucoma surgery [6]. Althoughthe central area of the VF is thought to be invaded in theadvanced stages of glaucoma, PCS can occur in some patientswith even mild VF defects. It seems that glaucomatous dam-age progresses from the central to the peripheral VF area insome patients, but occurs in the opposite way in others [7]. Toanswer the question as to which factors could affect thesedifferences, risk factors for initial PCS have been investigated[8–13]. Highmyopia [8], autonomic dysfunction [9], systemichemodynamic factors [10], and disc hemorrhage [10] havebeen reported to be risk factors for paracentral involvement inpatients with mild VF defects. These risk factors seem to berelated to vascular dysregulation, which is thought to have animportant role in the pathogenesis of normal tension glaucoma(NTG) [14]. It has also been reported that NTG patients showmore localized and central VF defects than primary open-angle glaucoma patients [15]. These findings suggest thatdevelopment of initial PCS may be related to ocular andsystemic hemodynamic disturbances.

M. Lee :H. Jin : J. Ahn (*)Department of Ophthalmology, Ajou University School ofMedicine,San 5, Wonchon-dong, Yeongtong-gu, Suwon 443-729, Koreae-mail: chrisahn@ajou.ac.kr

Graefes Arch Clin Exp Ophthalmol (2014) 252:307–314DOI 10.1007/s00417-013-2513-2

In addition to hemodynamic disturbances, structural andfunctional characteristics of initial PCS have also been inves-tigated to identify associated structural factors [8, 9]. Whencompared with glaucoma patients with initial peripheral nasalstep (PNS), those with initial PCSwere reported to showmoreinferotemporal damage to the ONH and peripapillaryRNFL[9] and higher superior hemifield involvement in theVF test [8]. Considering these reports, it seems thatglaucomatous damage to the inferotemporal ONH has a ten-dency to show VF defects closer to fixation. In addition to thesite of damage, it is also possible that other anatomical factors,including spatial distribution and length of retinal ganglioncell (RGC) axons originating from receptive fields of theparafoveal area, may be related to the early presentation ofPCS.

Therefore, we investigated whether geometric factors—including disc margin to fovea distance (DFD), ONH param-eters, and peripapillary RNFL thickness—could be related toinitial central VF involvement in normal-tension glaucoma(NTG) patients with mild VF defects.

Materials and methods

This study was cross-sectional and retrospective. Medicalrecords of NTG patients and normal subjects who visited theglaucoma clinic at the Department of Ophthalmology, AjouUniversity Hospital, from January 2009 to August 2012 werereviewed. The present study conformed to the Declaration ofHelsinki. It was approved by the institutional review board ofAjou University Hospital.

One hundred and eighty-nine NTG patients with meandeviation (MD) values better than −6.0 dB were screenedfrom the medical records in our study. All subjects underwentcomprehensive ophthalmological examinations, including areview of their medical history, best-corrected visual acuity(BCVA), spherical equivalent of refractive error (SE), slit-lamp biomicroscopy, central corneal thickness (CCT; DGH-500, DGH technology, Exton, PA, USA), IOP, and ocularpulse amplitude (OPA), measured by Goldman applanationtonometry (GAT) and dynamic contour tonometry (DCT;Pascal, Swiss Microtechnology, Port, Switzerland),gonioscopy, dilated fundoscopic examination and automatedperimetry (HFA; Carl ZeissMeditec, Dublin, CA, USA) usingSwedish interactive thresholding algorithm (SITA) standard24–2 strategies. Stereo optic-disc photography (ODP; AFC-210; NIDEK, Aichi, Japan) and optical coherence tomogra-phy (OCT; Stratus OCT; Carl Zeiss Meditec) were performedon each eye on the same day. VF tests, ODP, and OCTexaminations were performed within 3 months of other ocularexaminations, including IOP measurements.

NTG patients had a gonioscopically open angle andshowed at least two reproducible VF test results, compatible

with glaucoma, according to Anderson’s criteria for minimalabnormality in glaucoma [16]. RNFL defect was noted in atleast one of red-free RNFL photography and OCT, which mustcorrespond to the VF defects. We enrolled only NTG patientswho had mild VF defects with MD values better than −6.0 dB,according to the Hodapp–Parrish–Anderson criteria.[17]

Our subjects were divided into two groups according to thepresence of significant PCS: the central VF-invading andcentral VF-sparing groups. NTG patients who had significantPCS were classified in the invading group. NTG patients whohad at least one point with a p -value below 0.5 % in the fourcentral points (within 5° of fixation) or at least one point with ap -value below 1 % within 5° of fixation where neighboringpoints also had a p -value less than 0.5 % on a total deviationprobability map were considered to have significant PCS. Thedegree of central VF involvement was quantified by the sumof the Collaborative Initial Glaucoma Treatment Study(CIGTS) score of the central four points of the total deviationmap, rated according to the CIGTS scoring system [18]. Thequantified degree of central VF involvement was defined asthe central 4 CIGTS score.

ODPwere taken using anAFC-210 (NIDEK, Aichi, Japan)and geometric parameters, such as disc area (DA), disc diam-eter (DD), rim area (RA), vertical cup to disc (VCD) ratio, areaand width of peripapillary atrophy (PPA), and DFD, wereanalyzed using the intrinsic semi-automated disc analysisalgorithms. The reader pointed to at least 15 points at theoutline of the optic disc and the cup. Then a complete circle,which represents the disc and cup margin, was drawn auto-matically. Finally, DA, DD, RA, and the VCD ratio werecalculated using the intrinsic software as described previously[19]. Area and width of PPA were also measured by theintrinsic software after free-form drawing of the outlinearound the PPA.

DFD was defined as the distance from the temporal discmargin to the fovea, and was calculated using the ruler func-tion of the ODP intrinsic software. As shown in Figure 1, DFDwas calculated from the imaginary line connecting the centerof the optic disc and the fovea. The center of the optic disc wasdefined as the point at which the longest and the shortest axisof optic disc intersected. The longest axis of the optic disc wasdefined as a line connecting the superior and inferior poles ofthe ONH. The shortest axis was perpendicular to the center ofthe longest axis. The midpoint of the longest axis was equi-distant from the superior and inferior poles of the optic disc,and was thought to be closest to the horizontal raphe of theRNFL distribution of individual patients. The foveal reflexwas defined as the center of the foveola. To minimizeestimation error, we measured DFD three times separatelyand used mean values. If PPA was present, the PPA widthwas defined as the distance between the outer border ofthe PPA and the temporal disc margin on an imaginaryline of the DFD.

308 Graefes Arch Clin Exp Ophthalmol (2014) 252:307–314

RNFL thickness was measured by OCT, and the pupilswere not routinely dilated during OCT imaging. The I/S ratiowas calculated as the RNFL thickness of the inferior quadrantdivided by that of the superior quadrant.

To verify the relationship between RNFL thickness of OCTand DFD in healthy eyes, 42 eyes from 42 consecutive normalsubjects were also analyzed. Normal subjects had normal VFtest results, and showed no RNFL defect in OCT or red-freeRNFL photographs.

All recorded IOPs were<21 mmHg before treatment andduring follow-up in all subjects, and no subject used systemicmedications that could affect IOP or BP. None of the patientshad undergone ocular surgery, including cataract or refractivesurgery and laser treatment to control intraocular pressure.

Only reliable VF tests (<25 % fixation loss, false negatives,and false positives) were included in our analysis. OCT im-ages showing good centration and signal strength≥7 wereincluded in the analysis. Any subject with a tilted disc or ill-defined border of the foveola was excluded. Subjects who hada BCVA less than 20/30, or ocular or systemic diseases otherthan glaucoma that could affect the optic nerve or visual field,were excluded. Patients who showed unreliable DCT mea-surements, for which the quality scores (Q) were>, were alsoexcluded.

The main outcome measures were structural parameters ofONH, including DFD and peripapillary RNFL thickness re-lated with initial PCS in VF tests. DFD was the primaryoutcome measure, and other ONH parameters (disc

Fig. 1 Examples of patients fromcentral visual field invading (left)and sparing (right) group. Totaldeviation map of Humphreyvisual field test showedparafoveal scotoma in centralvisual field invading group (a),but not in sparing group (b). Theretinal nerve fiber layer thicknessof optical coherence tomographydecreased in the inferior quadrantin the central visual field invadinggroup (c), but did not in bothsuperior and inferior quadrants insparing group (d). DFD (distancebetween disc margin and fovea)was measured on optic discphotographs using intrinsic rulerfunction of the device on animaginary line connectingbetween the center of the opticnerve head and the fovea incentral visual field invading (e)and sparing (f) group

Graefes Arch Clin Exp Ophthalmol (2014) 252:307–314 309

dimension, VHD ratio, and PPA) and RNFL parameters takenby OCTwere secondary outcome measures.

Comparison between the invading and sparing groups ofNTG patients was conducted using the independent sample t -test and χ2 test. In addition, comparison between the normaland NTG groups was conducted by ANOVAwith Dunnett‘spost hoc comparison test. Correlations between parameterswere assessed by Pearson correlation and multiple linearregression analyses. Descriptive results are shown as meansand standard deviations. P -values<0.05 were considered toindicate statistical significance. Data were analyzed using theSPSS software (ver. 20.0; SPSS Inc., Chicago, IL, USA).

Results

In total, 88 eyes of 88 NTG patients from 189 screenedpatients were enrolled according to the inclusion and exclu-sion criteria. NTG patients were divided into the central VF-sparing (50 eyes) and -invading (38 eyes) groups. There wasno significant difference in age, male/female ratio, SE, IOPmeasured by GAT and DCT, or CCT (Table 1). The SE of theinvading group (range, −7.625 to +0.875 D) was not signifi-cantly different from that of the sparing group (range, −8.250to +1.750 D) and the proportions of myopia, hyperopia, andemmetropia were not different between groups (p =0.574, χ2

test). Four of 50 eyes in the sparing group and three of 38 eyesin the invading group had high myopia (SE<−6.00 D; p =0.986, χ2 test). NTG patients were treated with topical IOP-lowering medications, such as beta blockers, alpha-2 agonists,carbonic anhydrase inhibitors, and/or prostaglandin analogs.The class (p >0.1, χ2 test) and number (1.5 ± 1.0 vs 1.3 ± 1.0,p =0.362) of medications did not differ between the invadingand sparing groups.

As shown in Table 1, OPA in the invading group washigher than in the sparing group. DFD of the invading groupwas significantly shorter than that of the sparing group. ThePPA width in the invading group was narrower than in thesparing group. The vertical disc diameter to horizontal discdiameter (VHD) ratio was lower in the invading group, indi-cating that the optic disc in the invading group showed a lessvertically oval shape. RNFL thicknesses of the temporal andinferior quadrant were thinner in the invading group, and thatof the superior quadrant was thinner in the sparing group. As aresult, the I/S ratio of RNFL thickness was significantly lowerin the invading group (p <0.001). Of 50 eyes, 27 (54 %) in thesparing group and 32 of 38 eyes (84 %) in the invading groupshowed superior hemifield involvement in the visual fieldtests (p =0.003, χ2 test).

In a univariate analysis, the central 4 CIGTS score corre-lated positively with OPA and RNFL thickness of the superiorquadrant, and negatively with DFD, VHD ratio, and I/S ratioof OCT RNFL thickness (Table 2). In a multiple regression

analysis, only DFD correlated significantly with the central 4CIGTS score (Table 2).

Structural parameters related to DFD were also investigat-ed. In all NTG patients, DFD was positively correlated withVDD but negatively correlated with HDD. As a result, theVHD ratio correlated positively with DFD. PPA width andtemporal RNFL thickness also correlated positively withDFD. Temporal RNFL thickness correlated positively withDFD in the sparing group, but not in the invading group.Nasal RNFL thickness correlated negatively with DFD inthe sparing group, but not in the invading group (Table 3).

The relationship between RNFL thickness and DFD wasalso investigated in normal patients to assess whether thetemporal and nasal RNFL thickness could be related to DFDin healthy eyes. Although disc size in the normal groupdiffered from that of the glaucoma patients, age was notsignificantly different between the normal and NTG groups.The DFD of normal subjects was not significantly differentfrom that of the invading group, but was significantly shorterthan that of the sparing group (Table 4). In these normalsubjects, DFD correlated positively with temporal RNFLthickness (r =0.516, p <0.001) and negatively with nasalRNFL thickness (r =−0.322, p =0.038).

Discussion

Consistent with previous reports [7–11], our study showedthat parafoveal involvement in NTG patients with mild VFdefects was predominantly in the superior hemifield. Recently,clinical characteristics of initial paracentral versus peripheralVF defects have been examined. In one study, disc hemor-rhage and systemic risk factors (hypotension, migraine,Raynaud’s phenomenon, sleep apnea) were more commonin the initial PCS group than in the initial PNS group [10]. Itwas also reported that untreated IOP of the initial PCS groupwas significantly lower than that of the initial PNS group.These findings suggest that initial PCSmay be associated withsystemic or local vascular dysregulation, and glaucoma pa-tients with initial PCS may be more influenced by IOP-independent factors than those with initial peripheral VFdefects. Although untreated IOP was not significantly differ-ent between the initial PCS and PNS NTG subgroups in theirstudy, this may be supported by another report in which NTGpatients with low-teen IOP showed more central VF defectsthan those with high-teen IOP [13]. As reported previously[12], OPA in the central VF-invading group was higher than inthe central VF-sparing group (Table 1). If OPA indicatesocular blood flow, then ocular hemodynamic parametersmay differ between groups. Although it is unclear whetherelevated OPA in the eyes with initial PCS is a causative factorof PCS, or a compensatory reaction of the eyes to PCS, theassociation between initial PCS and elevated OPA suggests

310 Graefes Arch Clin Exp Ophthalmol (2014) 252:307–314

that vascular dysregulation mechanisms generate initial PCSin glaucoma.

Jung et al. [11] investigated structural and functional char-acteristics of isolated PCS and PNS in NTG patients withrelatively mild VF defects (MD>−10 dB). In their report,patients with isolated PCS showedmore structural deteriorationin the temporal and inferotemporal sectors of ONH and RNFLwhen using optical imaging devices, and had more frequentsuperior hemifield defects in the VF tests. The fovea is knownto be located below the center of the optic disc in most eyes[20]. Considering this spatial relationship between the foveaand ONH, the inferotemporal sector of peripapillary RNFLmay contain many more RGC axons originating from the areaadjacent to the papillomacular bundle than the superotemporalsector. Furthermore, the RGC axons from the superiorparafoveal region corresponding to the inferior paracentral areaof the VF test location are thought to merge more diversely intothe ONH than those from the inferior one, because their spatialcorrelation with the VF test location has been reported to bemuch weaker [21]. Consequently, glaucomatous damage in thesuperotemporal sector of the ONH can induce more peripheralVF defect than that in the inferotemporal sector of the ONH.This relationship between RNFL and VF defects in the maculararea, considering the relative locations of the fovea and ONH,was well-illustrated in a previous report.[22]

In addition to the spatial relationship between the fovea andthe ONH, we posulated that the distribution and length of theRGC axons of the parafoveal area in relation to the ONHcould also be contributing factors related to paracentralVF involvement. In our study, the degree of central VF

Table 1 Comparison betweencentral visual field- sparing andcentral visual field-invadinggroups in patients with normal-tension glaucoma

CCT, central corneal thickness;GAT, Goldmann applanation to-nometry; OPA, ocular pulse am-plitude; MD, mean deviation;PSD, pattern standard deviation;VFI, visual field index; CIGTS,Collaborative Initial GlaucomaTreatment Study; VDD, verticaldisc diameter; HDD, horizontaldisc diameter; VHD ratio, VDD/HDD; DFD, distance betweendisc margin and fovea; PPA,peripapillary atrophy; RNFL, ret-inal nerve fiber layer; OCT, opti-cal coherence tomography; I/Sratio, inferior quadrant RNFLthickness/superior quadrantRNFL thickness

Parameter Sparing group (n = 50) Invading group (n= 38) P-value

Age (years) 49.4 ± 14.6 53.4 ± 9.9 0.132

Male/female ratio 34/16 20/18 0.142

CCT (μm) 551.6 ± 38.6 538.6 ± 33.7 0.101

SE (diopter) −2.360 ± 2.526 −1.586 ± 2.352 0.146

GAT (mmHg) 14.5 ± 3.4 14.3 ± 2.7 0.861

OPA (mmHg) 2.43 ± 0.88 2.91 ± 0.94 0.017

MD (dB) −3.065 ± 1.850 −3.062 ± 1.540 0.995

PSD (dB) 6.382 ± 2.790 6.964 ± 2.409 0.307

VFI (%) 91.9 ± 4.2 88.4 ± 5.0 <0.001

Central 4 CIGTS score 0.2 ± 0.6 5.6 ± 2.6 <0.001

VDD (mm) 1.825 ± 0.205 1.753 ± 0.260 0.148

HDD (mm) 1.586 ± 0.221 1.619 ± 0.203 0.478

VHD ratio 1.167 ± 0.177 1.090 ± 0.138 0.023

DFD (mm) 3.877 ± 0.278 3.642 ± 0.401 0.002

PPA area (mm2) 1.278 ± 0.926 0.959 ± 0.650 0.074

PPAwidth (mm) 0.407 ± 0.298 0.298 ± 0.210 0.031

RNFL thickness measured by OCT (μm)

Average 80.73 ± 13.07 80.98 ± 11.18 0.923

Temporal quadrant 71.44 ± 17.60 62.45 ± 15.71 0.015

Superior quadrant 95.94 ± 22.55 111.13 ± 17.59 0.001

Nasal quadrant 63.88 ± 16.43 68.34 ± 14.59 0.189

Inferior quadrant 91.50 ± 22.05 81.92 ± 13.80 0.021

I/S ratio 1.014 ± 0.368 0.754 ± 0.173 <0.001

Table 2 Factors significantly correlated with Collaborative Initial Glau-coma Treatment Study score of central 4 points of visual field test in total88 eyes with normal tension glaucoma

Correlationcoefficient (p-value)

Beta coefficient by multiplelinear regression (p-value)*

DFD −0.354 (0.001) −0.355 (0.002)VHD ratio −0.221 (0.038) 0.027 (0.813)

RNFL thickness ofsuperior quadrant

0.304 (0.004) 0.129 (0.371)

I/S ratio −0.332 (0.002) −0.180 (0.203)OPA 0.222 (0.037) (0.066)

* Model summary : CIGTS score of central 4 points=12.07−3.211×DFD(R2 =0.259, p =0.013)

CIGTS, Collaborative Initial Glaucoma Treatment Study; DFD, distancebetween disc margin and fovea; VHD ratio, vertical disc diameter/hori-zontal disc diameter; RNFL, retinal nerve fiber layer; I/S ratio, inferiorquadrant RNFL thickness/superior quadrant RNFL thickness; OPA, ocu-lar pulse amplitude

Graefes Arch Clin Exp Ophthalmol (2014) 252:307–314 311

involvement was quantified using the central 4 CIGTS scoreon a total deviation map in the VF test, and correlated signif-icantly with DFD, VHD ratio, RNFL thickness of superiorquadrant, I/S ratio using OCT RNFL measurement, and OPA.In a multiple regression analysis, only DFD correlated signif-icantly with central 4 CIGTS score (Table 2). NTG patientswho had shorter DFD showed a tendency to have higherscores for central VF involvement. Among these factors,

VHD ratio correlated significantly with DFD, indicating thata less vertically oval shape of the ONHwas related to a shorterDFD. The peripapillary RNFL thickness of the temporalsector correlated significantly with DFD in total NTG patients.In particular, both temporal and nasal RNFL thicknessescorrelated significantly with DFD in the central VF-sparinggroup (Table 3). Because we considered that the temporalsector of the ONH in the sparing group may have less

Table 3 Correlation betweendistance between disc margin tofovea and other structural param-eters in total normal-tensionglaucoma patients, sparing group,and invading group

Pearson correlation coefficient r(p-value)

NTG, normal-tension glaucoma;SE, spherical equivalent; CCT,central corneal thickness; VDD,vertical disc diameter; HDD, hor-izontal disc diameter; VHD ratio,VDD/HDD; DA, disc area; RA,rim area; VCD ratio, vertical cupto disc ratio; PPA, peripapillaryatrophy; RNFL, retinal nerve fi-ber layer; OCT, optical coherencetomography; I/S ratio, inferiorquadrant RNFL thickness/superi-or quadrant RNFL thickness

Parameter Total NTG patients (n=88) Sparing group (n =50) Invading group (n =38)

Age −0.196 (0.067) −0.262 (0.067) −0.039 (0.818)

SE −0.206 (0.054) −0.397 (0.004) 0.056 (0.740)

CCT −0.020 (0.851) −0.011 (0.940) −0.166 (0.320)

VDD 0.226 (0.034) −0.026 (0.860) 0.343 (0.035)

HDD −0.288 (0.007) −0.543 (<0.001) −0.028 (0.866)

VHD ratio 0.484 (<0.001) 0.510 (<0.001) 0.410 (<0.001)

DA −0.043 (0.692) −0.351 (0.012) 0.213 (0.200)

RA −0.124 (0.251) −0.273 (0.055) 0.044 (0.793)

VCD ratio −0.005 (0.960) −0.092 (0.526) 0.199 (0.232)

PPA area 0.243 (0.023) 0.088 (0.541) 0.361 (0.026)

PPAwidth 0.422 (<0.001) 0.502 (<0.001) 0.270 (0.101)

RNFL thickness measured by OCT

Temporal 0.268 (0.012) 0.484 (<0.001) −0.080 (0.631)

Superior −0.064 (0.557) 0.199 (0.167) −0.103 (0.540)

Nasal −0.208 (0.051) −0.345 (0.014) −0.005 (0.977)

Inferior 0.039 (0.719) −0.039 (0.790) −0.066 (0.692)

Average 0.015 (0.892) 0.123 (0.396) −0.091 (0.587)

I/S ratio 0.071 (0.508) −0.143 (0.322) −0.051 (0.762)

Table 4 Comparison between subgroups of normal-tension glaucoma and normal group

Parameter Sparing group (n = 50) Invading group (n = 38) Normal (n = 42) P-value**

Age (years) 49.6 ± 14.7 (0.948)* 53.0 ± 10.0 (0.470)* 50.2 ± 15.2 0.375

Male/female 34/16 (0.015)* 20/18 (0.382)* 18/24 0.050

SE (diopter) −2.360 ± 2.526 (0.008)* −1.586 ± 2.352 (0.368)* −0.952 ± 1.997 0.016

DFD (mm) 3.873 ± 0.280 (0.005)* 3.653 ± 0.354 (0.894)* 3.670 ± 0.277 0.001

DA (mm2) 2.259 ± 0.514 (0.016)* 2.234 ± 0.525 (0.015)* 2.572 ± 0.638 0.010

RA (mm2) 1.095 ± 0.349 (<0.001)* 1.135 ± 0.323 (<0.001)* 1.757 ± 0.382 <0.001

VCD ratio 0.717 ± 0.147 (<0.001)* 0.725 ± 0.068 (<0.001)* 0.533 ± 0.095 <0.001

VDD (mm) 1.824 ± 0.207 (0.081)* 1.756 ± 0.232 (0.004)* 1.932 ± 0.289 0.007

HDD (mm) 1.585 ± 0.223 (0.006)* 1.620 ± 0.213 (0.060)* 1.932 ± 0.251 0.010

VHD ratio 1.167 ± 0.178 (0.291)* 1.031 ± 0.136 (0.468)* 1.124 ± 0.117 0.053

Average RNFL (μm) 80.73 ± 13.07 (<0.001)* 80.98 ± 11.18 (<0.001)* 106.470 ± 2.04 <0.001

Temporal quadrant RNFL (μm) 71.44 ± 17.61 (0.375)* 62.45 ± 15.71 (0.001)* 75.36 ± 12.159 0.001

Superior quadrant RNFL (μm) 95.94 ± 22.55 (<0.001)* 111.13 ± 17.59 (<0.001)* 130.36 ± 15.811 <0.001

Nasal quadrant RNFL (μm) 63.88 ± 16.43 (<0.001)* 68.34 ± 14.59 (<0.001)* 85.57 ± 13.608 <0.001

Inferior quadrant RNFL (μm) 91.50 ± 22.05 (<0.001)* 81.92 ± 13.80 (<0.001)* 134.36 ± 13.752 <0.001

( )*, p-values compared with normal group; **, ANOVAwith Dunnett’s post hoc comparison test; SE, spherical equivalent; DFD, distance between discmargin and fovea; DA, disc area; RA, rim area; VCD ratio, vertical cup to disc ratio; VDD, vertical disc diameter; HDD, horizontal disc diameter; VHDratio, VDD/HDD; RNFL, retinal nerve fiber layer thickness

312 Graefes Arch Clin Exp Ophthalmol (2014) 252:307–314

glaucomatous damage than that in the invading group, weassessed whether there was any relationship between DFDand circumpapillary RNFL thickness in undamaged healthyeyes. Although disc size, gender ratio, and SE of the normalgroup were not well-matched with the glaucoma patients(especially the sparing group), we found that the temporalRNFL thickness in the sparing group was not significantlydifferent from that of the normal group, while other sectorswere significantly thinner in all glaucoma groups than in thenormal group (Table 4). In these normal patients, the sametendency between DFD and circumpapillary RNFL thicknesswas found as in the sparing group (Table 3). According tothese findings, we suggest that a shorter DFD may be accom-panied by thinner temporal circumpapillary RNFL thicknessin undamaged eyes. It is also possible that the RGC axons maybe distributed more nasally in eyes with shorter DFD, butmore temporally in eyes with longer DFD. If the same amountof glaucomatous damage took place, the functional deficit inthe receptive field corresponding to the site of damage couldbe more profound in patients with thinner RNFL, becausethey have fewer reservoirs to resist than those with a thickerone. VHD ratio and temporal PPAwidth correlated positivelywith DFD in our study (Table 3). These findings suggest thatthe eyes with less vertically oval discs and narrower temporalPPA could be associated with shorter DFD, which may berelated to the initial PCS in NTG patients.

Although it was not statistically significant, the sparinggroup showed slightly more myopic refractive errors thanthe invading group (Table 1). The wider PPA and morevertically oval ONH of the sparing group could be due tomyopic changes in the ONH, as reported previously [23]. Theextent of PPA seemed not to be associated with initial PCS inour study, although the presence and extent of the beta zonePPA have been reported to correlate with glaucoma develop-ment [24] and progression, structurally [25] and functionally[26]. This may be because there are regions of the ONH thatare more influenced by PPA, as reported previously [25, 27].

Although DFD was not significantly correlated with SE inall NTG patients in our study, it was negatively correlatedwithSE in the central VF-sparing group (Table 3). This is consis-tent with a previous report in which the axial length of the eyecorrelated positively with DFD [28]. The initial PCS groupshowed slightly more myopic refractive error than the initialPNS group in one study [11], but there was no significantdifference in SE between the groups in another study [10] andours. These findings suggest that the influence of myopia oncentral VF involvement in glaucoma may be minimal, al-though high myopia was associated with more diffuse PMBdefects [29], and the eyes with longer axial length showeddenser scotoma in VF test [30] in previous reports.

Current optical imaging devices can measure ONH param-eters [31] and DFD n[28] with high accuracy and reproduc-ibility. The ONH parameters and DFD were measured by disc

photography with a semi-automated algorithm in our study.Although the reproducibility of this ODP measuring wasreported to be comparable with that of Cirrus OCT in aprevious study [19], it could be a limitation of this study. Wedid not measure the axial length in each patient because of theretrospective study design, so the influence of axial length oninitial PCS in glaucoma could not be examined. The durationof glaucoma can influence structural changes in ONH orRNFL thickness, but this could not be compared betweenthe invading and sparing groups due to the retrospective studydesign. Although normal subjects were enrolled not to com-pare them with glaucoma subjects but to examine the relation-ship between DFD and peripapillary RNFL thickness, somedemographic differences between normal and glaucomagroups may influence the results.

In conclusion, the initial presentation of central VF defectin glaucoma seems to be associated with both hemodynamicand structural factors, including DFD. NTG patients withhemodynamic or structural characteristics related to initialPCS should be managed with more attention to prevent dete-rioration of visual function. Additionally, a prospective longi-tudinal study should be conducted to assess risk factors relatedto initial PCS by observing preperimetric glaucoma patientswith these risk factors who develop PCS during the follow-upperiod.

The English in this document has been checked by at least two profes-sional editors, both native speakers of English. For a certificate, please see:

http://www.textcheck.com/certificate/T63BWA

Conflict of interest The authors have no conflict or financial interest inthe subject matter of this manuscript.

References

1. Quigley HA, Addicks EM (1981) Regional differences in the struc-ture of the lamina cribrosa and their relation to glaucomatous opticnerve damage. Arch Ophthalmol 99:137–143

2. Radius RL, Gonzales M (1981) Anatomy of the lamina cribrosa inhuman eyes. Arch Ophthalmol 99:2159–2162

3. Kolker AE (1977) Visual prognosis in advanced glaucoma: a compar-ison of medical and surgical therapy for retention of vision in 101 eyeswith advanced glaucoma. Trans Am Ophthalmol Soc 75:539–555

4. Fujita K, Yasuda N, Oda K, Yuzawa M (2006) Reading performancein patients with central visual field disturbance due to glaucoma.Nippon Ganka Gakkai Zasshi 110:914–918

5. Coeckelbergh TR, Brouwer WH, Cornelissen FW, Van Wolffelaar P,Kooijman AC (2002) The effect of visual field defects on drivingperformance: a driving simulator study. Arch Ophthalmol 120:1509–1516

6. Francis BA, Hong B,Winarko J, Kawji S, Dustin L, Chopra V (2011)Vision loss and recovery after trabeculectomy: risk and associatedrisk factors. Arch Ophthalmol 129:1011–1017

7. Su D, Park SC, Simonson JL, Liebmann JM, Ritch R (2012)Progression pattern of initial parafoveal scotomas in glaucoma.Ophthalmology 120:520–527

Graefes Arch Clin Exp Ophthalmol (2014) 252:307–314 313

8. Kimura Y, Hangai M, Morooka S, Takayama K, Nakano N, NukadaM, Ikeda HO, Akagi T, Yoshimura N (2012) Retinal nerve fiber layerdefects in highly myopic eyes with early glaucoma. InvestOphthalmol Vis Sci 53:6472–6478

9. Park HY, Jung KI, Na KS, Park SH, Park CK (2012) Visual fieldcharacteristics in normal-tension glaucoma patients with autonomicdysfunction and abnormal peripheral microcirculation. Am JOphthalmol 154(3):466.e1–475.e1

10. Park SC, De Moraes CG, Teng CC, Tello C, Liebmann JM, Ritch R(2011) Initial parafoveal versus peripheral scotomas in glaucoma:risk factors and visual field characteristics. Ophthalmology 118:1782–1789

11. Jung KI, Park HY, Park CK (2012) Characteristics of optic discmorphology in glaucoma patients with parafoveal scotoma comparedto peripheral scotoma. Invest Ophthalmol Vis Sci 53:4813–4820

12. LeeM, Cho EH, LewHM, Ahn J (2012) Relationship between ocularpulse amplitude and glaucomatous central visual field defect innormal-tension glaucoma. J Glaucoma 21:596–600

13. Kim DM, Seo JH, Kim SH, Hwang SS (2007) Comparison oflocalized retinal nerve fiber layer defects between a low-teen intra-ocular pressure group and a high-teen intraocular pressure group innormal-tension glaucoma patients. J Glaucoma 16:293–296

14. Mozaffarieh M, Flammer J (2013) New insights in the pathogenesisand treatment of normal tension glaucoma. Curr Opin Pharmacol 13:43–49

15. Thonginnetra O, Greenstein VC, Chu D, Liebmann JM, Ritch R,Hood DC (2010) Normal versus high tension glaucoma: a compar-ison of functional and structural defects. J Glaucoma 19:151–157

16. Budenz DL (1997) Glaucomatous visual field loss. In: Budenz DL(ed) Atlas of visual field. Lippincott-Raven, Philadelphia, pp 143–193

17. Hodapp E, Parrish RK, Anderson DR (1993) Clinical decisions inglaucoma. Mosby, St. Louis, pp 52–61

18. Gillespie BW,Musch DC, Guire KE, Mills RP, Lichter PR, Janz NK,Wren PA, Group CS (2003) The collaborative initial glaucoma treat-ment study: baseline visual field and test–retest variability. InvestOphthalmol Vis Sci 44:2613–2620

19. Lee M, Yoo H, Ahn J (2013) Comparison of disc analysis algorithmsprovided by Cirrus OCTand stereo optic-disc photography in normaland open angle glaucoma patients. Curr Eye Res 38:605–613

20. Rohrschneider K (2004) Determination of the location of the foveaon the fundus. Invest Ophthalmol Vis Sci 45:3257–3258

21. Ferreras A, Pablo LE, Garway-Heath DF, Fogagnolo P, Garcia-FeijooJ (2008) Mapping standard automated perimetry to the peripapillaryretinal nerve fiber layer in glaucoma. Invest Ophthalmol Vis Sci 49:3018–3025

22. Hood DC, Raza AS, de Moraes CG, Odel JG, Greenstein VC,Liebmann JM, Ritch R (2011) Initial arcuate defects within thecentral 10 degrees in glaucoma. Invest Ophthalmol Vis Sci 52:940–946

23. Kim TW, KimM,Weinreb RN,Woo SJ, Park KH, Hwang JM (2012)Optic disc change with incipient myopia of childhood.Ophthalmology 119(21–26):e21–e23

24. Tezel G, Dorr D, Kolker AE, Wax MB, Kass MA (2000)Concordance of parapapillary chorioretinal atrophy in ocular hyper-tension with visual field defects that accompany glaucoma develop-ment. Ophthalmology 107:1194–1199

25. Lee EJ, Kim TW, Weinreb RN, Park KH, Kim SH, Kim DM (2011)beta-Zone parapapillary atrophy and the rate of retinal nerve fiberlayer thinning in glaucoma. Invest Ophthalmol Vis Sci 52:4422–4427

26. Teng CC, DeMoraes CG, Prata TS, Liebmann CA, Tello C, Ritch R,Liebmann JM (2011) The region of largest beta-zone parapapillaryatrophy area predicts the location of most rapid visual field progres-sion. Ophthalmology 118:2409–2413

27. Kawano J, Tomidokoro A, Mayama C, Kunimatsu S, Tomita G,Araie M (2006) Correlation between hemifield visual field damageand corresponding parapapillary atrophy in normal-tension glauco-ma. Am J Ophthalmol 142:40–45

28. Chui TY, Zhong Z, Burns SA (2011) The relationship betweenperipapillary crescent and axial length: Implications for differentialeye growth. Vision Res 51:2132–2138

29. Chihara E, Tanihara H (1992) Parameters associated withpapillomacular bundle defects in glaucoma. Graefes Arch Clin ExpOphthalmol 230:511–517

30. Chen YF, Wang TH, Lin LL, Hung PT (1997) Influence of axiallength on visual field defects in primary open-angle glaucoma. JFormos Med Assoc 96:968–971

31. Chauhan BC, O’Leary N, Almobarak FA, Reis AS, Yang H, SharpeGP, Hutchison DM, Nicolela MT, Burgoyne CF (2013) Enhanceddetection of open-angle glaucoma with an anatomically accurateoptical coherence tomography-derived neuroretinal rim parameter.Ophthalmology 120:535–543

314 Graefes Arch Clin Exp Ophthalmol (2014) 252:307–314