ccgm y cpi oct sp

Macular Ganglion Cell/Measurements by Spectral DTomography for Detection

e

ZAC

enhance the performance of the GCL/IPL measures. Cirrus High-Definition OCT (Cirrus HD-OCT, Carl ZeissInquiries to Kouros Nouri-Mahdavi, 100 Stein Plaza, Los Angeles, CACombining the best parameters from each algorithmimproved detection of glaucoma (P[ 0.04). CONCLUSIONS: Regional GCL/IPL measures derivedfrom Cirrus HD-OCT performed as well as regionalRNFL outcomes for detection of early glaucoma. Usingthe GC/IPL to OR ratio did not enhance the performance

Meditec, Dublin, California, USA) now provide a ganglioncell analysis in which GCL/IPL thickness measurementsand various graphic and statistical analyses are provided.There is preliminary evidence that such GCL/IPL measure-ments may perform as well as RNFL measures for detectionof early glaucoma.6,7 Prior studies have reported age andaxial length to be the main factors affecting variability ofthe GCL/IPL or GCC measurements.8,9 A recent studyfound that the GCC/total retinal thickness (TRT) ratioperformed better than all other RNFL or macularparameters for detection of glaucoma.10 This finding, ifconfirmed, suggests that the thickness of the inner retinaltissues is related to the density and thickness of the outer

Accepted for publication Aug 1, 2013.From the Glaucoma Division, Jules Stein Eye Institute, David Geffen

School of Medicine, University of California Los Angeles, Los Angeles,California (all).

90095; e-mail: [email protected] to Retinal Nerv

KOUROS NOURI-MAHDAVI, SARA NOWROOSHANE KNIPPING, JOANN GI

PURPOSE: To evaluate the performance of ganglion celllayer/inner plexiform layer (GCL/IPL) measurementswith spectral-domain optical coherence tomography(Cirrus HD-OCT) for detection of early glaucoma andto compare results to retinal nerve fiber layer (RNFL)measurements. DESIGN: Cross-sectional prospective diagnostic study. METHODS: We enrolled 99 subjects, including 59 eyeswith glaucoma (47 subjects) (mean deviation >L6.0dB) and 91 normal eyes (52 subjects). Patients under-went biometry and peripapillary and macular OCTimaging. Performance of the GCL/IPL and RNFL algo-rithms was evaluated with area under receiver operatingcharacteristic curves (AUC), likelihood ratios, and sensi-tivities/specificities adjusting for covariates. Combinationof best parameters was explored. RESULTS: Average (SD) mean deviation in the glau-coma group wasL2.5 (1.9) dB. Onmultivariate analyses,age (P< 0.001) and axial length (P [ 0.03) predictedGC/IPL measurements in normal subjects. No significantcorrelation was found between average or regional GC/IPL thickness and respective outer retina (OR) thicknessmeasurements (P > 0.05). Average RNFL thicknessperformed better than average GCL/IPL measurementsfor detection of glaucoma (AUC [ 0.964 vs 0.937;P [ 0.04). The best regional measures from each algo-rithm (inferior quadrant RNFL vs minimum GCL/IPL)had comparable performances (P [ 0.78). Entering theGC/IPL to OR ratio into prediction models did not0002-9394/$36.00http://dx.doi.org/10.1016/j.ajo.2013.08.001

2013 BY ELSEVIER INC.Inner Plexiform Layeromain Optical Coherenceof Early Glaucoma andFiber Layer Measurements

IZADEH, NARIMAN NASSIRI, NILA CIRINEO,ONI, AND JOSEPH CAPRIOLI

of GCL/IPL parameters. Combining the best measures fromthe 2 algorithms improved detection of glaucoma. (Am JOphthalmol 2013;156:12971307. 2013 by ElsevierInc. All rights reserved.)

THE HALLMARK OF GLAUCOMA IS LOSS OF THE

retinal ganglion cell axons that leads to a typicaloptic neuropathy. Qualitative and quantitative

evaluation of the optic nerve head and the retinal nervefiber layer (RNFL) has been used to detect glaucomatousdamage. There is evidence that macular ganglion cell losscan be demonstrated in early experimental glaucoma.1 Inaddition, recent publications suggest that evidence of glau-comatous damage can be observed in the inner retina organglion cell complex (GCC), ie, the combined thicknessof the RNFL, the ganglion cell layer (GCL), and the innerplexiform layer (IPL), early during the disease process bymeans of spectral domain optical coherence tomography(SD-OCT).2,3 A recent study showed that GCC losscould be detected in eyes with preperimetric glaucoma.4

It has been suggested that the macular GCL thicknessmay be the most relevant parameter to measure in glau-coma.5 However, given the resolution of the current gener-ation of SD-OCTs, accurate measurement of the GCLalone is not practical. Hence, some SD-OCT manufac-turers have developed segmentation algorithms for calcu-lating the GCC thickness or the combined thickness ofthe GCL and IPL. The software versions 6.0 or higher of1297ALL RIGHTS RESERVED.

ISADORANota adhesivasello distintivo

evidence of glaucomatous damage at the level of the optic

nerve head (see below).All subjects underwent a thorough eye examination on

the day of imaging, including visual acuity, automatedrefraction, measurement of intraocular pressure (IOP),gonioscopy, slit-lamp exam, dilated fundus exam, and stan-dard achromatic perimetry (SAP) or short-wavelengthautomated perimetry (SWAP) fields. Axial length andkeratometry weremeasured by IOLMaster (Carl ZeissMedi-tec), and those test results were considered reliable if thesignal to noise ratio for individual scans was above 3.Central corneal thickness measurements were measuredwith a DGH 55 Pachmate (DGH Technology, Exton,Pennsylvania, USA). The devices output is the averageof 10 reliable measurements of the central corneal thick-retinal cells and, therefore, the ratio of the inner to outerretinal (OR) layers may be a useful parameter to includein algorithms used for detection of glaucoma. The goal ofthe current study is to explore factors affecting the GCL/IPL thickness in a group of normal eyes from the Universityof California, Los Angeles (UCLA), OCT Imaging Study,so as to evaluate the performance of the newly definedGCL/IPL measurements for detection of early glaucomaand to compare the findings to RNFL thickness measure-ments. We also hypothesized that the outer retina thick-ness may predict the GCL/IPL thickness, so including theOR thickness or the GCL/IPL/OR ratio in the predictionmodels using macular measurements may improve theirperformance.

METHODS

THIS STUDYWASAPPROVEDBYTHE INSTITUTIONALREVIEW

board at the UCLA and was carried out in accordance withthe Declaration of Helsinki. We prospectively recruited 59eyes of 47 glaucoma subjects and 91 eyes of 52 normalsubjects between December 2010 and October 2012.Patients with clinical diagnoses of open-angle glaucomamade by an attending ophthalmologist were prospectivelyidentified and invited to be enrolled in the study if theymet the following criteria: age >_30 years, open angles,visual acuity >_20/80, visual field mean deviation (MD)>_6 dB, refractive error

amlion

aph

l Sub

(52

(69.2) 66.1 (66.0)

outer retina thickness or the ratio of GC/IPL to outer retinathickness was not a significant covariate in any of theglobal or regional multivariate models (P > 0.6 for all).The AUCs, sensitivities/specificities, and accuracies forthe global and regional GC/IPL and RNFL thicknessmeasures are presented in Table 3. As is evident inFigures 4 and 5, the AUC for the average RNFL thicknesswas superior to average GC/IPL measurements (AUC 0.964; 95% CI 0.936-0.991 vs 0.937; 95% CI: 0.895-0.980; P 0.04). The partial AUC (pAUC) differencebetween the average RNFL and average GC/IPL thicknessmeasurements with the cutoff point placed at 95%approached statistical significance (P 0.057) (Fig. 4).The performance of the minimum and inferotemporal

FIGURE 1. Association of the average ganglion cell/inner plex-iform layer thickness with the average retinal nerve fiber layerthickness as a function of diagnosis. Circles[ normal subjects;triangles[ glaucoma patients.

FIGURE 2. Association of the average ganglion cell/inner plex-iform layer thickness with the average outer retinal thicknessaccording to diagnosis. Circles[ normal subjects; triangles[glaucoma patients.detail. The control subjects were younger and had shorteraxial length on average (P< 0.001 for both). On univariateanalyses, the only factors found to be associated withthinner average RNFL thickness in the control groupwere older age (P 0.01), longer axial length (P 0.012), and potentially smaller disc area (P 0.073),whereas the SD-OCT signal strength (P 0.305) andspherical equivalent (P 0.344) did not show any associ-ation with the average RNFL thickness. On multivariateanalyses, older age (beta 0.43 m per year; P 0.003);longer axial length (beta 2.11 m per mm; P 0.018);and smaller disc area (beta 1.19 m per 0.1 mm2; P 0.056) were associated with thinner average RNFLmeasurements in normal eyes.Thinner global GC/IPL measurements in control

subjects were associated with advancing age (P 0.001),flatter keratometry (P 0.058), longer axial length (P 0.101), and male gender (P 0.149) on univariate anal-yses. Signal strength did not show a significant associationwith average GC/IPL thickness (P 0.339). On multivar-iate analyses, only age (beta0.28m per year; P 0.001)and axial length (beta 1.36 m per mm; P 0.03)predicted average GC/IPL thickness measurements innormal subjects. Interaction of age and axial length wasnot significant in multivariate analyses (P 0.140). Alower average GC/IPL thickness predicted a worse MD inthe normal control group (beta 1.45 dB per micron,P 0.017), but the association disappeared after correctingfor age and axial length. The average GC/IPL thicknesscorrelated significantly with the average RNFL thicknessin both the normal and the glaucoma groups (r 0.693and 0.628, respectively; P < 0.001 for both) (Fig. 1),whereas the association with the disc area was not statisti-cally significant (r 0.121, P 0.252 in the control groupand r 0.020, P 0.876 in the glaucoma group). Nosignificant association was found between the average orregional GC/IPL thickness and average or regional outerretina thickness measurements (r 0.029, P 0.801for normal subjects and r0.156, P 0.251 for glaucomaeyes for global measurements; P> 0.05 for all regional asso-ciations) (Fig. 2).

PERFORMANCE OF THE TWO ALGORITHMS FOR DETEC-TION OF GLAUCOMA: Table 2 compares the global andregional/sectoral thickness measurements for RNFL andGC/IPL between the 2 groups, and Figure 3 shows thedistribution of the difference between glaucoma andcontrol groups for the average and regional GC/IPL andRNFL thickness measurements. All potentially significantcovariates (ie, parameters with P< 0.15 on univariate anal-yses or factors that make biological sense, such as disc area)were included in the final stepwise multivariate analysesalong with each global or regional RNFL or GC/IPLpredictor and logit scores were calculated for each outcometo calculate AUCs. The final logistic models included age,signal strength, axial length, and disc size as covariates. The1300 AMERICAN JOURNAL OF DECEMBER 2013OPHTHALMOLOGY

Plemet

rou

1.

0.

0.

1.

1.

1.

1.

0.

1.

1.

1.

1.

1.

2.

1.

1.TABLE 2.Comparison of Global andRegional Ganglion Cell/InnerControl and Early Peri

OCT Parameters Normal G

GC/IPL parameters

Average GC/IPL (mean 6 SE) 81.1 (6

Minimum GC/IPL (mean 6 SE) 79.2 (6

Superotemporal GC/IPL (mean 6 SE) 80.5 (6

Superior GC/IPL (mean 6 SE) 81.8 (6

Superonasal GC/IPL (mean 6 SE) 82.3 (6

Inferonasal GC/IPL (mean 6 SE) 80.6 (6

Inferior GC/IPL (mean 6 SE) 79.9 (6

Inferotemporal GC/IPL (mean 6 SE) 81.7 (6

RNFL parameters

Average RNFL (mean 6 SE) 94.0 (6

Superior quadrant (mean 6 SE) 115.7 (6

Nasal quadrant (mean 6 SE) 72.9 (6

Inferior quadrant (mean 6 SE) 124.0 (6

Temporal quadrant (mean 6 SE) 63.5 (6

1 oclock sector (mean 6 SE) 104.8 (6

2 oclock sector (mean 6 SE) 89.7 (6

3 oclock sector (mean 6 SE) 61.7 (6GC/IPL thickness (the best performing regional GC/IPLmeasures) was similar to that of the inferior quadrantRNFL thickness, the best performing regional RNFLoutcome (AUCs 0.962 [95% CI: 0.915-0.985] vs 0.955[95% CI: 0.906-0.981] vs 0.977 [95% CI: 0.943-0.996],P 0.71 and 0.23, respectively; P for pAUC differenceof minimum and inferotemporal GC/IPL vs the inferiorquadrant RNFL 0.77 and 0.82, respectively). Whensensitivity and specificities were evaluated, the sensitivityfor the minimum GC/IPL thickness or inferotemporalGC/IPL sector (81.3%, 95% CI: 70.9%-91.7% and84.9%, 95% CI: 75.5%-94.3%, respectively) were compa-rable to but lower than that of the inferior quadrantRNFL (89.8%, 95% CI: 82.0-97.6%) when compared athigh specificity (>_95%). As mentioned above, the ROCsand sensitivities/specificities were all adjusted for theconfounding factors listed above. We also hypothesizedthat the adding the best regional GC/IPL parameter tothe best regional RNFL parameter could potentiallyimprove discrimination of glaucomatous from normaleyes although, given the very good performance of both

4 oclock sector (mean 6 SE) 66.5 (61.









GC/IPL ganglion cell/inner plexiform layer; OCT optical coherenc

VOL. 156, NO. 6 PERFORMANCE OF MACULARxiform Layer (GC/IPL), and RNFL ThicknessMeasurements in theric Glaucoma Groups

p Glaucoma Subjects P value

0) 66.5 (61.3)

the highest positive LR at the 1% level (18.89, 95% CI:7.20-49.60). An RNFL sector demonstrating P > 5% alsohad the best negative LR (0.05, 95%CI: 0.01-0.19), closely

FIGURE 3. Bar graph demonstrates the average difference in the reparameters in the glaucoma and control groups. Whiskers representthickness; IN[ inferonasal; IT[ inferotemporal; SN[ superon

TABLE 3. Area Under the Receiver Operating Characteristic CurvesParameters for Detection of Early Glaucoma as Measured

AUC (95% CI) Sensitivity Specificity Accuracy

RNFL parameters

Average RNFL 0.964 (0.936-0.991) 88.1% 91.2% 90.0%

Superior

quadrant

0.936 (0.898-0.974) 84.8% 86.8% 86.0% N

Inferior

quadrant

0.976 (0.952-0.999) 93.2% 91.2% 92.0% T

1 oclock sector 0.896 (0.844-0.947) 81.4% 83.5% 82.4% 2

3 oclock sector 0.866 (0.809-0.923) 83.1% 80.2% 81.6% 4

5 oclock sector 0.906 (0.858-0.953) 91.5% 75.8% 83.7% 6

7 oclock sector 0.964 (0.933-0.995) 93.2% 93.4% 93.3% 8

9 oclock sector 0.861 (0.802-0.920) 81.4% 82.4% 81.9% 1

11 oclock

sector

0.919 (0.876-0.961) 88.1% 82.4% 85.3% 1

GC/IPL

parameters

Average

GC/IPL

0.937 (0.889- 0.972) 86.4% 87.9% 87.3% M

Superior

GC/IPL

0.906 (0.848-0.948) 76.3% 86.8% 85.0% I

Inferonasal

GC/IPL

0.891 (0.833-0.938) 72.9% 82.4% 78.7% S

Inferotemporal

GC/IPL

0.955 (0.906-0.981) 86.4% 94.5% 91.3% S

AUC area under curve; GC/IPL ganglion cell/inner plexiform layelayer.

1302 AMERICAN JOURNAL OFfollowed by the RNFL quadrant thickness and GC/IPLsectors (LR 0.08, 95% CI: 0.03-0.20 and 0.11, 95% CI:0.05-0.24), respectively.

tinal nerve fiber layer and the ganglion cell/inner plexiform layer1 standard error. GC[ ganglion cell layer/inner plexiform layerasal; ST[ superotemporal.

for Retinal Nerve Fiber and Ganglion Cell/Inner Plexiform Layer

by Spectral Domain Optical Coherence Tomography

AUC (95% CI) Sensitivity Specificity Accuracy

asal quadrant 0.889 (0.837-0.942) 74.6% 85.7% 81.3%

emporal

quadrant

0.880 (0.826-0.933) 71.2% 82.4% 78.0%

oclock sector 0.875 (0.815-0.936) 93.2% 80.2% 86.7%

oclock sector 0.877 (0.821-0.932) 88.1% 75.8% 82.0%

oclock sector 0.952 (0.917-0.987) 91.5% 86.8% 89.2%

oclock sector 0.867 (0.810-0.923) 96.6% 62.6% 79.6%

0 oclock sector 0.898 (0.847-0.949) 91.5% 76.9% 84.2%

2 oclock sector 0.907 (0.860-0.955) 81.4% 90.1% 85.7%

inimum GC/IPL 0.962 (0.915-0.985) 91.5% 94.5% 93.3%

nferior GC/IPL 0.938 (0.889-0.972) 83.1% 91.2% 88.0%

uperotemporal

GC/IPL

0.926 (0.873-0.963) 83.3% 74.6% 89.0%

uperonasal GC/

IPL

0.871 (0.809-0.922) 71.2% 83.5% 78.7%

r; OCT optical coherence tomography; RNFL retinal nerve fiber

DECEMBER 2013OPHTHALMOLOGY

DISCUSSION

INTEREST IN MACULARMEASUREMENTS FOR DETECTION OF

glaucoma dates back more than a decade. Using the RetinalThickness Analyzer, Zeimer and colleagues were able todetect decreased macular thickness in eyes with establishedglaucoma.16 Similar findings were reported with earliergenerations of OCT.1723 With the advent of SD-OCTs,the quality of OCT images has improved considerably,allowing better segmentation and delineation of variousmacular layers. This has led to renewed interest inmeasuring the inner retinal layers, where significant lossof cells occurs in glaucoma.24,25 Current SD-OCT devices

FIGURE 4. Receiver operating characteristic curves comparingperformance of average retinal nerve fiber layer and ganglioncell/inner plexiform layer thickness measurements for detectionof early perimetric glaucoma.

FIGURE 5. Receiver operating characteristic curves comparingperformance of the best retinal nerve fiber layer parameter (infe-rior quadrant thickness) and the best ganglion cell/inner plexi-form (GC/IPL) layer parameter (minimum GC/IPL) fordetection of early perimetric glaucoma.

VOL. 156, NO. 6 PERFORMANCE OF MACULARhave adopted different approaches in this regard. Initially,algorithms were developed to measure the combined thick-ness of the central macular RNFL, GCL, and IPL called theganglion cell complex, or GCC.2 Such measurements haveshown very good reproducibility.26 In the study by Tan andassociates, the GCC global loss volume performed as well asthe time-domain average RNFL. Ganglion cell complexmeasurements have been reported to be able to detectevidence of early damage in the seemingly noninvolved

27

FIGURE 6. Receiver operating characteristic curves comparingperformance of the inferior quadrant thickness alone (bestretinal nerve fiber layer parameter) with the combined perfor-mance of the inferior quadrant and the minimum ganglioncell/inner plexiform layer thickness for detection of early peri-metric glaucoma.hemifield of glaucomatous eyes. Seong and colleaguesdemonstrated that the GCC parameters performed aswell as the RNFL thickness measurements in glaucomatouseyes with early damage and in eyes in which the visual fieldloss was within the central 10 degrees.28 A review of thecurrent literature with regard to GCC parameters demon-strates that the performance of inner macular parametersapproaches that of the RNFL, although RNFL parametersstill tend to carry out this task slightly better.2,28,29 Thisdifferential performance seems to be primarily a functionof the location of glaucomatous damage and the area ofthe macula scanned.28,30

The goal of the current study was to evaluate fully theperformance of the macular GC/IPL thickness measure-ments as determined by the most recent Cirrus HD-OCTsoftware and compare its performance to that of circumpa-pillary RNFL thickness measurements for discriminationbetween eyes with early perimetric glaucoma (MD>6.0 dB) from normal control eyes. We first evaluatedthe predictive factors determining the GC/IPL thicknessin the normal control eyes. Specifically, we were interestedin the potential correlation between the inner retina thick-ness and the outer retina thickness in the central macula.Preliminary data show that the ratio of the GCC to TRT

1303SD-OCT IN GLAUCOMA

Flaggra

.47

.71

.71

.25

.22

.02

could potentially improve the discriminatory power of theSD-OCT measurements for detection of glaucoma.10

The global and regional GC/IPL thickness measurementsdiminished with advancing age and longer axial length. Thisis consistent with the available reports in the literature onthis topic.9,31 We found no evidence of any potentialcorrelations between the outer retinal thickness and theGC/IPL measurements. When we entered either theabsolute outer retina thickness or the ratio of the GC/IPLto outer retina thickness as covariates in multivariatelogistic regression models, this finding was confirmed; thatis, neither covariate improved prediction of glaucomawhen added to the multivariate models. This is in contrastto the findings by Kita and colleagues.10 They reported thatthe parameters average, superior, or inferior GCC/TRT ratioactually outperformed the corresponding GCC parameters.

TABLE 4. Frequency of Statistically Abnormal Measurements asPrintouts of Cirrus High-Definition Optical Coherence Tomo

OCT Parameters

Normal Group Glaucoma Group

P < 5% P < 1% P < 5% P < 1%

Average RNFL 10 (11%) 8 (8.8%) 42 (71.2%) 25 (42.4%) 6

Average

GC/IPL

10 (11%) 5 (5.5%) 37 (62.7%) 30 (50.9%) 5

RNFL

quadrant

11 (12.1%) 9 (9.9%) 55 (93.2%) 49 (83.1%) 7

RNFL sectors 27 (29.7%) 4 (4.4%) 57 (96.6%) 49 (83.1%) 3

GC/IPL

sectors

8 (8.8%) 6 (6.6%) 53 (89.8%) 49 (83.1%) 10

Min GC/IPL 7 (7.7%) 6 (6.6%) 50 (84.8%) 42 (71.2%) 11

GC/IPL ganglion cell/inner plexiform layer; Min minimum; OCTRacial differences, differences in the axial length of the studysamples, or the macular area examined may be partiallyresponsible for the inconsistent findings. This may also bedue, in part, to the fact that the retinal ganglion cells(RGCs) do not directly overlie the corresponding rods andcones in the perifoveal area of the macula.32 One could arguethat measuring macular or ganglion cell (GCC or GC/IPL)and outer retina volumes may be better indicators of the rela-tionship between the inner and outer retinal layers. We arecurrently exploring this issue in further detail.We then compared the performance of the GC/IPL

thickness measurements for detection of early glaucomato that of RNFL thickness measurements after adjustingfor relevant covariates, such as age, axial length, OCTsignal strength, and disc size. Our results showed that globalRNFL measurements (ie, average RNFL thickness) weresuperior to global GC/IPL thickness values for detectionof early glaucoma according to the AUC values; however,based on results provided by the normative database, the 2global parameters actually demonstrated similar perfor-

1304 AMERICAN JOURNAL OFmances. On the other hand, based on the AUC and partialAUC calculations, regional GC/IPL measurements(minimum GC/IPL and inferotemporal GC/IPL)performed as well as regional RNFLmeasurements (inferiorquadrant). This finding suggests that in the macular region,similar to the peripapillary RNFL, early damage can be andoften is localized. Alternatively, one may speculate thatbecause of averaging, global measures are inherently lesslikely to detect early localized loss. One important issuethat needs to be emphasized is the fact that macular SD-OCT imaging is actually measuring an area encompassingonly half of the RGCs in the eye (in the central macula),whereas RNFL imaging aims to evaluate the entirecomplement of the RGC axons. The fact that the globalRNFL measurements were superior to the correspondingGC/IPL measurements would be expected. The enrolled

ged on the Retinal Nerve Fiber Layer and Ganglion Cell Analysisphy in the Normal Control Subjects and Glaucoma Patients

Positive Likelihood Ratio Negative Likelihood Ratio

P < 5% P < 1%

P < 5% considered

abnormal

Only P < 1%

considered

abnormal

(3.53-11.88) 4.82 (2.33-9.96) 0.32 (0.22-0.49) 0.63 (0.50-0.79)

(3.08-10.6) 9.25 (3.81-22.50) 0.42 (0.30-0.59) 0.52 (0.40-0.68)

(4.41-13.49) 8.39 (4.47-15.78) 0.08 (0.03-0.20) 0.18 (0.11-0.33)

(2.37-4.48) 18.89 (7.2-49.60) 0.05 (0.01-0.19) 0.18 (0.10-0.31)

(5.24-19.92) 12.60 (5.76-27.53) 0.11 (0.05-0.24) 0.18 (0.10-0.32)

(5.36-22.63) 10.80 (4.90-23.79) 0.16 (0.09-0.30) 0.30 (0.21-0.46)

optical coherence tomography; RNFL retinal nerve fiber layer.group of glaucoma patients had early perimetric glaucomawith a median interquartile range MD of 2.4 (3.6to1.1) dB; hence, our findings are quite relevant for earlydetection or confirmation of glaucoma. We also found thatcombining the best RNFL and GC/IPL parameters fromCirrus HD-OCT actually improved its ability to discrimi-nate early glaucomatous eyes from normal control eyes.This effect was evident when coefficients for partialAUCs at very high cutoff point for specificity (95%) werecompared where it most matters (Fig. 6). This means thatdespite the very good isolated performance of RNFL andGC/IPL parameters and the potential for a ceiling effect,combining the best parameters from the 2 imaging algo-rithms could be of value for early detection of glaucoma,although the clinical relevance of the magnitude of thiseffect may be debatable. The utility of combining variousOCT parameters has been previously explored with bothtime-domain and spectral-domain OCTs.3335 The studyby Huang and associates used linear discriminantfunction to combine various OCT parameters. These

DECEMBER 2013OPHTHALMOLOGY

FOppor AdNeratioa coingbase of the Cirrus HD-OCT takes into consideration onlythe effect of aging, and factors such as myopia and raceare not accounted for. The latter could potentially havebiased our findings because in contrast to AUCs derivedfrom multivariate logistic regressions, the statistical signif-icance of SD-OCT parameters could not be corrected fordifferences in clinical and demographic characteristicsbetween the control and glaucoma groups.Mwanza and colleagues recently explored the diagnostic

performance of the new GC/IPL parameters and found thatminimum and inferotemporal GC/IPL thickness measure-ments were the best parameters to discriminate betweeneyes with early glaucoma (defined as eyes with MD >6dB, similar to our study) and normal eyes and their perfor-mance was as good as peripapillary RNFL thicknessmeasurements.6 Because of the varying inclusion criteria,the SD-OCT devices used, and the analysis algorithms, itis difficult to compare results across studies concerningthis subject. The only available study comparing the utilityof GCC to GC/IPL thickness measurements did not finda difference in the AUCs for these 2 parameters.7 Newlydefined parameters, such as global loss volume and focalloss volume, have been designed to improve detectionof early glaucomawithRTVue (Optovue, Fremont, Califor-nia, USA) SD-OCT. The index global loss volume seems to

ALL AUTHORSHAVE COMPLETED AND SUBMITTED THE ICMJEand the following were reported. Research reported in this publicationwas sufor Prevention of Eye Disease and the National Eye Institute of Health undetant for Allergan. Dr Caprioli is a consultant forAllergan,Merck,Alcon, anin Design of study (K.N.); Analysis and interpretation (K.N., S.N.); PrepaFinal approval of manuscript (K.N., S.N., N.N., N.C., S.K., J.G., J.C.); Datresources (K.N., J.G., J.C.); Statistical expertise (K.N.); Obtaining of fundsupport (S.K.).

REFERENCES

1. Desatnik H, Quigley HA, Glovinsky Y. Study of centralretinal ganglion cell loss in experimental glaucoma inmonkey eyes. J Glaucoma 1996;5(1):4653.VOL. 156, NO. 6 PERFORMANCE OF MACULARically, it has been shown that the inferior macular nervefibers project to the inferior disc39 and, therefore, it seemsnatural that the inferotemporal GC/IPL would perform aswell as the inferior RNFL parameters. It has yet to be deter-mined whether GC/IPL measurements have any advantageover GCC measurements for detection of early glaucoma7;but very good reproducibility for global and regional GC/IPL measurements has been reported.40 This is consistentwith other studies exploring reproducibility of varioussublayers of the retina.2,41,42

The current study enrolled consecutive subjects willingto participate in a tertiary setting, and the glaucoma grouptended to be somewhat older and had higher axial lengths.We addressed this issue by using multivariate analysis,adjusting for potential differences in covariates.In summary, we explored the performance of the GC/IPL

parameters to discriminate between a group of patientswith early perimetric glaucoma and normal subjects. Thefindings confirm that inner macular parameters performas well as RNFL parameters in such patients. We also foundthat combining the macular and RNFL data could poten-tially enhance SD-OCTs performance for detection ofearly glaucoma. The GC/IPL to outer retina ratio did notenhance glaucoma detection in this study; however, thissubject requires further exploration.

RM FOR DISCLOSUREOF POTENTIAL CONFLICTS OF INTEREST,rted by an award from theGeraldOppenheimer Family FoundationCenterward Number K23EY022659 to Dr Nouri-Mahdavi. Dr Giaconi is a consul-wWorldMedical. DrNouri-Mahdavi is a consultant toAllergan. Involvedn of manuscript (K.N., S.N.); Critical revision of manuscript (J.G., J.C.);llection (K.N., S.N., N.N., N.C., S.K.); Provision of materials, patients, or(K.N.); Literature search (K.N.); Administrative, technical, and logistic

2. Tan O, Chopra V, Lu AT, et al. Detection of macularganglion cell loss in glaucoma by Fourier-domain opticalcoherence tomography. Ophthalmology 2009;116(12):23052314. e23012302.

3. Kotera Y, Hangai M, Hirose F, Mori S, Yoshimura N. Three-dimensional imaging of macular inner structures in glaucomalikelihood ratio. It must be noted that the normative data-the performance of the color probability maps in the GC/IPL printout. The high rate of false-positive RNFL findingsin myopic individuals is a well-known issue.36 Our study isone of the first to explore the likelihood ratios for the CirrusGC/IPL parameters. As mentioned above, regional GC/IPLmeasures performed as well as those of RNFL, although anabnormal RNFL clock sector at 1% significance level hadthe highest positive likelihood ratio, and the presence ofa normal clock sector (P> 5%) led to the highest negative

for detection of early glaucomatous damage. The minimum(followed by the inferotemporal) GC/IPL thickness had thebest AUC among GC/IPL parameters, which was compa-rable to that of the inferior quadrant RNFL, the bestregional parameter in the current study. The best regionalRNFL and GC/IPL parameters for detection of early glau-coma were located in the inferior hemifield (inferotemporalGC/IPL segment vs inferior quadrant), which is consistentwith the finding that the inferior rim is the most commonsite of early disc and RNFL glaucomatous damage.38 Specif-findings need to be implemented in the clinical software incurrent SD-OCT machines so that the full potential ofcombining various structural measures can be verified andused.There are scant data in the literature as yet concerningbe better suited to this aim, according to the data in theliterature.2,37 However, most of the studies based on theRTVue SD-OCT have looked at global GCC or GCCthickness in the superior or inferior hemiretina. In contrast,Cirrus HD-OCT provides regional GC/IPL measurements1305SD-OCT IN GLAUCOMA

by using spectral-domain optical coherence tomography.Invest Ophthalmol Vis Sci 2011;52(3):14121421.

4. Lisboa R, Paranhos A Jr, Weinreb RN, Zangwill LM,Leite MT, Medeiros FA. Comparison of different spectraldomain OCT scanning protocols for diagnosing preperi-metric glaucoma. Invest Ophthalmol Vis Sci 2013;54(5):34173425.

5. Nakano N, Hangai M, Nakanishi H, et al. Macular ganglioncell layer imaging in preperimetric glaucoma with specklenoise-reduced spectral domain optical coherence tomog-raphy. Ophthalmology 2011;118(12):24142426.

6. Mwanza JC, Durbin MK, Budenz DL, et al. Glaucoma diag-nostic accuracy of ganglion cell-inner plexiform layer thick-ness: comparison with nerve fiber layer and optic nervehead. Ophthalmology 2012;119(6):11511158.

7. Kotowski J, Folio LS, Wollstein G, et al. Glaucoma discrimi-nation of segmented cirrus spectral domain optical coherencetomography (SD-OCT) macular scans. Br J Ophthalmol 2012;96(11):14201425.

8. KimNR, Kim JH, Lee J, Lee ES, Seong GJ, Kim CY. Determi-nants of perimacular inner retinal layer thickness in normaleyes measured by Fourier-domain optical coherence tomog-raphy. Invest Ophthalmol Vis Sci 2011;52(6):34133418.

9. Mwanza JC, Durbin MK, Budenz DL, et al. Profile and predic-tors of normal ganglion cell-inner plexiform layer thicknessmeasured with frequency domain optical coherence tomog-raphy. Invest Ophthalmol Vis Sci 2011;52(11):78727879.

10. Kita Y, Kita R, Takeyama A, Takagi S, Nishimura C,Tomita G. Ability of optical coherence tomography-determined ganglion cell complex thickness to total retinalthickness ratio to diagnose glaucoma. J Glaucoma 2012[Epub ahead of print].

11. Curcio CA, Allen KA. Topography of ganglion cells inhuman retina. J Comp Neurol 1990;300(1):525.

12. Bengtsson B, Heijl A. False-negative responses in glaucomaperimetry: indicators of patient performance or test reli-ability? Am J Ophthalmol 2000;130(5):689.

13. Bengtsson B. Reliability of computerized perimetric thresholdtests as assessed by reliability indices and threshold reproduc-ibility in patients with suspect and manifest glaucoma. ActaOphthalmol Scand 2000;78(5):519522.

14. JunoyMontolio FG,Wesselink C, Gordijn M, Jansonius NM.Factors that influence standard automated perimetry testresults in glaucoma: test reliability, technician experience,time of day, and season. Invest Ophthalmol Vis Sci 2012;53(11):70107017.

15. Johnson CA, Sample PA, Cioffi GA, Liebmann JR,Weinreb RN. Structure and function evaluation (SAFE):I. criteria for glaucomatous visual field loss using standardautomated perimetry (SAP) and short wavelength auto-mated perimetry (SWAP). Am J Ophthalmol 2002;134(2):177185.

16. Zeimer R, Asrani S, Zou S, Quigley H, Jampel H. Quantita-tive detection of glaucomatous damage at the posterior poleby retinal thickness mapping: a pilot study. Ophthalmology1998;105(2):224231.

17. Bagga H, Greenfield DS, Knighton RW. Macular symmetrytesting for glaucoma detection. J Glaucoma 2005;14(5):358363.

18. Guedes V, Schuman JS, Hertzmark E, et al. Optical coher-ence tomography measurement of macular and nerve fiber1306 AMERICAN JOURNAL OFlayer thickness in normal and glaucomatous human eyes.Ophthalmology 2003;110(1):177189.

19. Ishikawa H, Stein DM, Wollstein G, Beaton S, Fujimoto JG,Schuman JS. Macular segmentation with optical coherencetomography. Invest Ophthalmol Vis Sci 2005;46(6):20122017.

20. Lederer DE, Schuman JS, Hertzmark E, et al. Analysis ofmacular volume in normal and glaucomatous eyes usingoptical coherence tomography. Am J Ophthalmol 2003;135(6):838843.

21. Wollstein G, Schuman JS, Price LL, et al. Optical coherencetomography (OCT) macular and peripapillary retinal nervefiber layer measurements and automated visual fields. Am JOphthalmol 2004;138(2):218225.

22. TanO, Li G, Lu AT, Varma R, Huang D.Mapping of macularsubstructures with optical coherence tomography for glau-coma diagnosis. Ophthalmology 2008;115(6):949956.

23. Medeiros FA, Zangwill LM, Alencar LM, et al. Detection ofglaucoma progression with stratus OCT retinal nerve fiberlayer, optic nerve head, andmacular thickness measurements.Invest Ophthalmol Vis Sci 2009;50(12):57415748.

24. Glovinsky Y, Quigley HA, Pease ME. Foveal ganglion cellloss is size dependent in experimental glaucoma. InvestOphthalmol Vis Sci 1993;34(2):395400.

25. Frishman LJ, Shen FF, Du L, et al. The scotopic electroretino-gram of macaque after retinal ganglion cell loss from experi-mental glaucoma. Invest Ophthalmol Vis Sci 1996;37(1):125141.

26. Hirasawa H, Araie M, Tomidokoro A, et al. Reproducibilityof thickness measurements of macular inner retinal layersusing SD-OCT with or without correction of ocular rotation.Invest Ophthalmol Vis Sci 2013;54(4):25622570.

27. Takagi ST, Kita Y, Yagi F, Tomita G. Macular retinalganglion cell complex damage in the apparently normalvisual field of glaucomatous eyes with hemifield defects.J Glaucoma 2012;21(5):318325.

28. SeongM, Sung KR, Choi EH, et al. Macular and peripapillaryretinal nerve fiber layer measurements by spectral domainoptical coherence tomography in normal-tension glaucoma.Invest Ophthalmol Vis Sci 2010;51(3):14461452.

29. Mori S, Hangai M, Sakamoto A, Yoshimura N. Spectral-domain optical coherence tomography measurement ofmacular volume for diagnosing glaucoma. J Glaucoma 2010;19(8):528534.

30. Morooka S, Hangai M, Nukada M, et al. Wide 3-dimensionalmacular ganglion cell complex imaging with spectral-domainoptical coherence tomography in glaucoma. Invest Ophthal-mol Vis Sci 2012;53(8):48054812.

31. Koh VT, Tham YC, Cheung CY, et al. Determinants ofganglion cell-inner plexiform layer thickness measured byhigh-definition optical coherence tomography. InvestOphthalmol Vis Sci 2012;53(9):58535859.

32. Raza AS, Cho J, de Moraes CG, et al. Retinal ganglion celllayer thickness and local visual field sensitivity in glaucoma.Arch Ophthalmol 2011;129(12):15291536.

33. Medeiros F, Zangwill L, Bowd C, Vessani R, Susannajr R,Weinreb R. Evaluation of retinal nerve fiber layer, opticnerve head, and macular thickness measurements for glau-coma detection using optical coherence tomography. Am JOphthalmol 2005;139(1):4455.

34. Huang JY, Pekmezci M, Mesiwala N, Kao A, Lin S. Diag-nostic power of optic disc morphology, peripapillary retinalDECEMBER 2013OPHTHALMOLOGY

nerve fiber layer thickness, and macular inner retinal layerthickness in glaucoma diagnosis with Fourier-domain opticalcoherence tomography. J Glaucoma 2011;20(2):8794.

35. Wang M, Lu AT, Varma R, Schuman JS, Greenfield DS,Huang D. Combining information from 3 anatomic regionsin the diagnosis of glaucoma with time-domain optical coher-ence tomography. J Glaucoma 2012 [Epub ahead of print].

36. Aref AA, Sayyad FE, Mwanza JC, Feuer WJ, Budenz DL.Diagnostic specificities of retinal nerve fiber layer, optic nervehead, and macular ganglion cell-inner plexiform layermeasurements in myopic eyes. J Glaucoma 2012 [Epub aheadof print].

37. Nukada M, Hangai M, Mori S, et al. Imaging of localizedretinal nerve fiber layer defects in preperimetric glaucomausing spectral-domain optical coherence tomography. J Glau-coma 2012 [Epub ahead of print].

38. Jonas JB, Gusek GC, Naumann GO. [The parapapillaryregion of normal and glaucoma eyes. I. Planimetric values

of 312 glaucoma and 125 normal eyes]. Klin Monatsbl Augenh1988;193(1):5261.

39. Hood DC, Raza AS, de Moraes CG, Liebmann JM, Ritch R.Glaucomatous damage of the macula. Prog Retin Eye Res2013;32:121.

40. Mwanza JC, Oakley JD, Budenz DL, Chang RT, Knight OJ,Feuer WJ. Macular ganglion cell-inner plexiform layer: auto-mated detection and thickness reproducibility with spectraldomain-optical coherence tomography in glaucoma. InvestOphthalmol Vis Sci 2011;52(11):83238329.

41. Garas A, Vargha P, Hollo G. Reproducibility of retinal nervefiber layer and macular thickness measurement with theRTVue-100 optical coherence tomograph. Ophthalmology2010;117(4):738746.

42. Arintawati P, Sone T, Akita T, Tanaka J, Kiuchi Y. the appli-cability of ganglion cell complex parameters determined fromSD-OCT images to detect glaucomatous eyes. J Glaucoma2012 [Epub ahead of print].VOL. 156, NO. 6 PERFORMANCE OF MACULAR 1307SD-OCT IN GLAUCOMA

Biosketch

Kouros Nouri-Mahdavi, MD, MSc, is Assistant Professor of Ophthalmology at the Glaucoma Division, Stein Eye Institute,

University of California Los Angeles. His research interests include exploring functional tests for detection of glaucoma or

its progression, enhancing the role of spectral-domain optical coherence tomography imaging in glaucoma, and the study of

treatment outcomes in glaucoma.1307.e1 DECEMBER 2013AMERICAN JOURNAL OF OPHTHALMOLOGY

Biosketch

Sara Nowroozizadeh, MD, received her medical degree from Shiraz University of Medical Sciences, Shiraz, Iran, and

finished her residency in Ophthalmology at Khalili Eye Hospital at the same University. She is currently an

International Fellow at the Glaucoma Division, Stein Eye Institute, University of California Los Angeles.VOL. 156, NO. 6 1307.e2PERFORMANCE OF MACULAR SD-OCT IN GLAUCOMA

Macular Ganglion Cell/Inner Plexiform Layer Measurements by Spectral Domain Optical Coherence Tomography for Detection of Early Glaucoma and Comparison to Retinal Nerve Fiber Layer MeasurementsMethodsResultsPerformance of the two algorithms for detection of glaucoma

DiscussionReferences

ccgm y cpi oct sp

Documents