Comparison of Spectral-Domain versusTime-Domain Optical CoherenceTomography in Management of Age-RelatedMacular Degeneration with Ranibizumab

Kaori Sayanagi, MD, Sumit Sharma, BS, Takuhiro Yamamoto, MD, Peter K. Kaiser, MD

Purpose: To compare the ability to delineate and detect patterns of choroidal neovascularization (CNV)activity in patients with exudative age-related macular degeneration (AMD) after ranibizumab treatment betweentime-domain optical coherence tomography (TD-OCT) and 4 different spectral-domain optical coherence to-mography (SD-OCT) devices.

Design: Prospective, consecutive case series.Participants: Sixty-one eyes of 58 patients with exudative AMD after ranibizumab treatment were included

in this study.Methods: All patients were imaged with TD-OCT and at least 1 of 4 different SD-OCT devices at the same

visit after ranibizumab treatment. The OCT images were analyzed in a masked fashion by 2 independent graders(KS, TY) to delineate and detect the presence of CNV activity defined as the presence of subretinal fluid,intraretinal cysts, intraretinal fluid, sub–retinal pigment epithelium (sub-RPE) fluid, or a combination thereof. Theautomated evaluation of retinal thickness also was analyzed between devices.

Main Outcome Measures: Evidence of CNV activity on linear B-scans and 3-dimensional so-called cubescans on SD-OCT and linear B-scan on TD-OCT.

Results: In linear B-scan mode, all 4 SD-OCT devices were superior in their ability to delineate sub-RPE fluidcompared with TD-OCT (P�0.05). Three of 4 SD-OCT devices were superior in delineating intraretinal fluid, and2 of 4 SD-OCT devices were superior in delineating subretinal fluid and intraretinal cysts (P�0.05). In the3-dimensional so-called cube mode, all 4 SD-OCT devices were superior in detecting subretinal fluid and 2 of 4SD-OCT devices were superior in detecting sub-RPE and intraretinal fluid (P�0.05). There were significantcorrelations in center point thickness between all SD-OCT devices and TD-OCT (P�0.01), and 3 of 4 and 1 of 3SD-OCT devices showed significant differences from TD-OCT in center point thickness (P�0.01) and centersubfield thickness (P�0.001), respectively.

Conclusions: SD-OCT is superior to TD-OCT in evaluating for CNV activity in patients with wet AMD afterranibizumab injection. Retinal thickness measurements between SD-OCT and TD-OCT also were significantlydifferent.

Financial Disclosure(s): Proprietary or commercial disclosure may be found after the references.

Ophthalmology 2009;116:947–955 © 2009 by the American Academy of Ophthalmology.

Exudative age-related macular degeneration (AMD) is theleading cause of severe visual loss and blindness in elderlypatients in the West.1 Although several treatments for AMDincluding laser photocoagulation, photodynamic therapy,and surgery have been reported, none has stabilized orimproved visual acuity consistently.

Ranibizumab (Lucentis; Genentech, Inc., South SanFrancisco, CA) is a recombinant humanized monoclonalantibody antigen-binding fragment that neutralizes the bio-logical activities of all human vascular endothelial growthfactor A (VEGF-A) isoforms and the biologically activedegradation products, such as VEGF110, and was approvedby the United State Food and Drug Administration for thetreatment of neovascular AMD in June 2006.2 According to

the phase 3 clinical trials, the Minimally Classic/Occult

© 2009 by the American Academy of OphthalmologyPublished by Elsevier Inc.

Trial of the Anti-VEGF Antibody Ranibizumab in theTreatment of Neovascular AMD (MARINA) and Anti-VEGF Antibody for the Treatment of Predominantly Clas-sic Choroidal Neovascularization in AMD (ANCHOR),ranibizumab is the first AMD therapy to achieve visualimprovement in the patients with exudative AMD.3,4 Thebenefits apply to all angiographic subtypes of exudativeAMD and all lesion sizes. Furthermore, the rate of severeocular and systematic adverse events was low.

Optical coherence tomography (OCT) is a noninvasiveand useful diagnostic tool that provides cross-sectional im-ages of the retina. Although fluorescence angiography (FA)has been a common diagnostic tool to consider the treatmentor retreatment for exudative AMD,5,6 OCT-guided therapy

with ranibizumab was evaluated in a small case series of

947ISSN 0161-6420/09/$–see front matterdoi:10.1016/j.ophtha.2008.11.002

Ophthalmology Volume 116, Number 5, May 2009

patients with wet AMD. The Prospective OCT Imaging ofPatients with Neovascular AMD Treated with Intra-OcularLucentis (PrONTO) study used time-domain OCT (TD-OCT)and evaluated the central 1 mm of 6 6-mm radial line scans todetermine if a patient should receive treatment with ranibi-zumab. Using this OCT-guided model, patients required con-siderably fewer treatments than the 24 monthly injectionsperformed in the phase 3 studies, yet yielded comparablevisual results.7 The conventional, third-generation TD-OCT(Stratus; Carl Zeiss Meditec, Inc., Dublin, CA) has an axialresolution of less than 10 �m, but still has a relatively slowdata-acquisition speed (400 A scans/second). Recently de-veloped spectral-domain OCT (SD-OCT) devices offer im-proved image resolution of less than 5 to 7 �m and dramat-ically faster acquisition speeds (18 000 to 40 000 A scan/second) and provide detailed views of the intraretinalmicrostructures as well as 3-dimensional (3-D) images ofthe retina.8–10 In SD-OCT, spectrum is measured with a

Table 1. Patient Demographic Data

No. (%) orMean�Standard Deviation

Total patients 58Gender

Male 22 (38%)Female 36 (62%)

Age (yrs) 81.8�5.0Total eyes 61Lens status

Phakic 24 (39%)Aphakic 0 (0%)Pseudophakic 37 (61%)

No. of previous ranibizumab injections1 11 (18%)2 12 (20%)3 7 (11%)4 3 (5%)�5 28 (46%)

Other previous treatments*Focal laser 5 (8%)Photodynamic therapy 12 (20%)Intravitreal injection

Steroid 2 (3%)Pegaptanib 6 (10%)Bevacizumab 20 (33%)

*Includes multiple previous treatments.

Optical CoherenceTomography

Findings

Spectralis (n � 38)

Time-Domain OpticalCoherence Tomography

Spectral-Domain OpticalCoherence Tomography

Subretinal fluid 1.47�0.65 1.76�0.43Intraretinal cysts 1.34�0.71 1.74�0.45Intraretinal fluid 1.34�0.71 1.74�0.45Sub-RPE fluid 1.37�0.59 1.82�0.39

RPE � retinal pigment epithelium.

P values in bold are �0.05.

948

spectrometer and is converted to depth information by aFourier transformation, so that SD-OCT does not need tomove the mechanical reference arm, resulting in the fasteracquisition times and greater axial resolution compared withTD-OCT.11

This study evaluated the ability to identify choroidalneovascularization (CNV) activity on TD-OCT comparedwith 4 different SD-OCT devices and determined whetherthe SD-OCT devices provided any additional informationfor decisions about ranibizumab treatment in exudativeAMD.

Patients and Methods

After Cleveland Clinic Institutional Review Board approval, pa-tients with CNV resulting from AMD were included in the studyif they met the following inclusion criteria: age older than 50 years,ability to give written informed consent, and undergoing OCTevaluation to determine whether treatment should be performedwith ranibizumab. Patients with CNV resulting from other causes,those with severe uveitis, and those who had a history of vitrec-tomy in the study eye were excluded. All patients underwentTD-OCT examination and either 1 of 4 SD-OCT examinations atthe same visit through a dilated pupil by trained OCT technicians.

Time-Domain Optical Coherence TomographyExaminationThe Stratus OCT (software version 4.0) was used for the TD-OCTexamination. This OCT uses a superluminescent diode with acenter wavelength of 820 nm. All patients were imaged using a6-mm linear cross-hair pattern centered on the fovea (512 A-scans;scan length, 6.0 mm), and a fast macular thickness map pattern (6linear B-scans in an asterisk pattern spaced 30° apart; each B-scanconsisting of 512 A-scans, 6 mm in length). The automated anal-ysis package of the Stratus OCT was used to evaluate center pointthickness (CPT; retinal thickness at the intersection of the 6 radiallines) and center subfield thickness (CSF; retinal thickness in thecentral 1 mm diameter circle). The automated analysis package ofthe Stratus OCT defines retinal thickness for CPT and CSF as thedistance between the internal limiting membrane (ILM) and thephotoreceptor junction of the inner and outer segments.

Spectral-Domain Optical Coherence TomographyExaminationFor the SD-OCT examinations, 4 different SD-OCT devices wereused in this study: SOCT Copernicus (Copernicus; Reichert, Inc.,

Table 2. Grading Score on Spectral-Domain Optical Coherence

Copernicus (n � 34)

ValueTime-Domain Optical

Coherence TomographySpectral-Domain OpticalCoherence Tomography P Value

0.045 1.74�0.51 1.82�0.39 0.4920.003 1.50�0.62 1.74�0.45 0.0540.006 1.47�0.56 1.76�0.43 0.0100.001 1.47�0.61 1.88�0.33 <0.001

P

<

Sayanagi et al � SD-OCT vs. TD-OCT after Ranibizumab

Depew, NY), Heidelberg Spectralis HRA�OCT (Spectralis; Hei-delberg Engineering, Heidelberg, Germany), Cirrus HD-OCT (Cir-rus; Carl Zeiss Meditec, Inc., Dublin, CA), and 3-D OCT-1000(OCT-1000; Topcon, Inc., Paramus, NJ). Because each device hasdifferent acquisition protocols and analysis packages, the methodsdiffered between devices.

1. SOCT Copernicus (Copernicus). This SD-OCT uses a su-perluminescent diode with a center wavelength of 840 nm.The axial resolution of the SOCT Copernicus is less than 6�m with a data-acquisition speed of 25 000 A scans/second. For macular evaluation, a 7.0-mm line scan patterncentered on the fovea (7427 A-scans; scan length, 7.0 mm)and volume scan (50 B-scans�743 A-scans, covering aretinal area of 7.0�7.0 mm) were used to obtain a singlehigh-resolution, horizontal B-scan and 3-D image. TheCPT was evaluated automatically using the software withinthe Copernicus and was defined as the distance betweenILM to the junction of the inner and outer segments. It wasmeasured on both a single B-scan and a foveal B-scan thatwas selected from a 3-D volume scan image. Retinalthickness for CSF also was defined as the distance betweenthe ILM and the junction of the inner and outer segmentsand was measured automatically from the 3-D volumescans by the automated software within the central 1-mmcircle.

2. Heidelberg Spectralis HRA�OCT (Spectralis). This SD-OCT uses a superluminescent diode with a center wave-length of 870 nm. The axial resolution of the SpectralisOCT is less than 7 �m with a data-acquisition speed of40 000 A scans/second. For the study, patients were im-aged using a single, horizontal B-scan image centered onthe fovea (1536 A-scans; scan angle, 30°; scan length, 9mm) and a 3-D cube scan (25 B-scans�384 A-scans,covering a retinal area of 30°�20°). The scans were ob-tained using the automated averaging system (automaticreal time mean � 14) turned on to amplify the signals andto reduce noise within the images. The CPT was defined asthe distance between the ILM to the bottom of the retinalpigment epithelium (RPE) by the automatic segmentationalgorithms of the Spectralis software. The CPT was mea-sured on both a single B-scan and a foveal B-scan that wasselected from a 3-D image. Software to measure the CSFwas not available in the current version of the Spectralissoftware.

3. Cirrus HD-OCT (Cirrus). This SD-OCT uses a superlumi-nescent diode with a center wavelength of 840 nm. Theaxial resolution of the Cirrus HD-OCT is less than 5 �mwith a data-acquisition speed of 27 000 A scans/second.For the study, patients were imaged using a 5-line rasterscan centered on the fovea (4096 A-scans; scan length, 6.0

Tomography and Time-Domain Optical Coherence Tomography

Cirrus (n � 27)

Time-Domain OpticalCoherence Tomography

Spectral-Domain OpticalCoherence Tomography P Value

1.52�064 1.85�0.46 0.0081.22�0.75 1.71�0.45 0.0021.30�0.72 1.78�0.42 0.0091.48�0.51 1.93�0.27 <0.001

mm), which consists of 5 closely spaced horizontal lines,and a macular cube 512�128 scan (128 B-scans�512A-scans, covering a retinal area of 6.0�6.0 mm). The CPTwas measured manually using the software calipers on afoveal B-scan image that was selected from a 3-D image asthe distance between the ILM to the top of the RPE. TheCSF was measured automatically by the software and wasdefined as the distance between ILM and the middle ofRPE on 3-D cube scans.

4. Three-dimensional OCT-1000 (OCT-1000). This SD-OCTuses a superluminescent diode with a center wavelength of830 nm. The axial resolution of the 3-D OCT-1000 is lessthan 6 �m with the data-acquisition speed of 18 000 Ascans/second. For the study, patients were imaged using asingle line-scan centered on the fovea (4096 A-scans; scanlength, 6.0 mm) and a 3-D scan (128 B-scans�512 A-scans, covering a retinal area of 6.0�6.0 mm). The CPTwas measured manually using the built-in calipers on botha single B-scan and a foveal B-scan that was selected froma 3-D image as the distance between ILM and the top ofRPE. The CSF was measured automatically on a 3-Dimage by the software and was defined as the distancebetween the ILM and the top of the RPE.

Optical Coherence Tomography Scan Comparison

All OCT images from the TD- and SD-OCT devices were re-viewed at 2 different times on different days by 2 independentgraders (KS, TY) who have experience in interpreting OCTs toevaluate for CNV activity. The CNV activity was defined as thepresence of subretinal fluid, intraretinal cysts, intraretinal fluid,sub-RPE fluid, or a combination thereof. Each OCT image wasgraded using the following grading criteria: 0 � activity was notdetected; 1 � activity was detected, but the details were not clearlyobserved; and 2 � activities as well as the details were clearlyobserved. The number of scans that had visible leakage to evaluatethe ability to detect any CNV leakage also was counted. Disagree-ment was resolved after discussion, so that investigators agreed onall the images. White-on-black images were used for all evalua-tions, and contrast and brightness were adjusted on each image toobtain the best possible image quality for analysis.

Statistic Analysis

Statistical analysis was performed with SigmaStat SPSS softwareversion 2.03 (SPSS, Inc., Chicago, IL). Comparison for significantdifference of grading score, CPT, and CSF between TD-OCT andSD-OCT images was analyzed using the paired t test or Wilcoxonsigned-rank test. Comparison for significant differences in the rateof detecting CNV activity between TD- and SD-OCT was ana-

OCT-1000 (n � 18)

Time-Domain OpticalCoherence Tomography

Spectral-Domain OpticalCoherence Tomography P Value

1.56�0.62 1.72�0.46 0.3751.56�0.62 1.55�0.51 1.0001.56�0.62 1.72�0.46 0.3131.33�0.59 1.77�0.43 0.020

949

Ophthalmology Volume 116, Number 5, May 2009

lyzed using the McNemar test. A statistical correlation betweenCPT of TD-OCT and CPT, CSF, or both, of SD-OCT images wasanalyzed using the Spearman rank-sum test. P values less than0.05 were considered significant.

Results

A total of 61 eyes and 58 patients were included in this study. Thepatient demographics are shown in the Table 1. Three eyes under-went 2 examinations, and 1 eye underwent 3 examinations atdifferent follow-up visits, yielding a total of 66 image sets foranalysis.

The mean grading score and ability to delineate CNV activityon TD-OCT and SD-OCT are shown in Table 2. All 4 SD-OCTdevices were significantly superior to TD-OCT in delineatingsub-RPE fluid (P�0.05). Three of 4 SD-OCT devices (Spectralis,Copernicus, and Cirrus) were superior in delineating intraretinalfluid (P�0.05). Two of 4 SD-OCT devices (Spectralis and Cirrus)were superior in delineating subretinal fluid and intraretinal cysts(P�0.05).

The prevalence of OCT findings on the single linear B-scan ofTD-OCT and SD-OCT devices, as well as the 3-D cube scans ofeach SD-OCT device, are shown in Tables 3 and 4, respectively.In linear B-scan mode, one SD-OCT device (Spectralis) wassuperior to TD-OCT in detecting subretinal fluid (P�0.05), andanother SD-OCT device (OCT-1000) was superior in detectingintraretinal cysts (P�0.05). In contrast, all 4 SD-OCT deviceswere superior to TD-OCT in detecting subretinal fluid in 3-Dcube mode. Two of 4 SD-OCT devices (Spectralis and Cirrus)were superior in detecting intraretinal fluid (P�0.05), and theother 2 SD-OCT devices (Copernicus and OCT-1000) weresuperior in detecting sub-RPE fluid (P�0.05) in 3-D cubemode.

The mean CPT and CSF of both TD-OCT and each SD-OCT

Table 3. Prevalence of Coexistence of Optical Coherence Tomand Spectral-Domain Optical Co

OpticalCoherence

TomographyFindings

Spectral-DomainOptical Coherence

Tomography

Presence on Time-DomCoherence Tomo

Presence on Spectral-Domain Optical

Coherence Tomography

Ab

Coh

Subretinal fluid Spectralis (n � 38) 9/38 (24%)Copernicus (n � 34) 9/34 (26%)Cirrus (n � 27) 5/27 (19%)OCT-1000 (n � 18) 6/18 (33%)

Intraretinal cysts Spectralis 15/38 (39%)Copernicus 7/34 (21%)Cirrus 11/27 (41%)OCT-1000 5/18 (28%)

Intraretinal fluid Spectralis 21/38 (55%)Copernicus 11/34 (32%)Cirrus 16/27 (59%)OCT-1000 7/18 (39%)

Sub-RPE fluid Spectralis 19/38 (50%)Copernicus 12/34 (35%)Cirrus 12/27 (44%)OCT-1000 9/18 (50%)

RPE � retinal pigment epithelium.P values in bold are �0.05.

device are shown Table 5. There were significant correlations in

950

CPT and CSF between TD-OCT and the SD-OCT devices(P�0.01). The CPTs of Spectralis and Copernicus in both linearB-scan and 3-D scan modes were significantly thicker (P�0.001)and thinner (P�0.01) than those of TD-OCT, respectively. TheCPT on Cirrus in 3-D scan mode (CPT on Cirrus in linear B-scanmode is not available) was significantly thinner, whereas CSF onCirrus was significantly thicker than that of TD-OCT (P�0.001).There was no significant difference between OCT-1000 and TD-OCT in CPT and CSF and between Copernicus and TD-OCT inCSF.

Case Reports

Case 1. A 67-year-old woman with CNV resulting from exudativeAMD underwent 1 previous intravitreal injection of ranibizumabbefore inclusion in the study. Her best-corrected visual acuity(BCVA) was 20/400. Three SD-OCT devices (Copernicus, OCT-1000, and Cirrus) detected and clearly delineated the intraretinaland sub-RPE fluid, and on the OCT-1000 image, the cysticchanges also were observed. Although Stratus TD-OCT detectedand delineated the sub-RPE fluid, the corresponding intraretinalfluid was visualized only as hyporeflectivity (Fig 1).

Case 2. An 82-year-old man with CNV resulting from exuda-tive AMD underwent 5 previous intravitreal injections of ranibi-zumab before inclusion in the study. His BCVA was 20/40�2. Insingle linear B-scan modes, the Cirrus detected and clearly delin-eated both subretinal and sub-RPE fluid in addition to drusen. OnOCT-1000 and Copernicus scans, sub-RPE fluid was visualizedclearly, but the subretinal fluid was not observed clearly. StratusTD-OCT and Spectralis visualized only sub-RPE fluid (Fig 2). Incontrast, on 3-D cube scans, all SD-OCT devices visualized sub-RPE fluid and subretinal fluid. The subretinal fluid was observed mostclearly on the higher and lower side of the fovea (Fig 3).

Case 3. A 76-year-old woman with CNV resulting from exu-dative AMD underwent 1 previous intravitreal injection of ranibi-

hy Findings on Time-Domain Optical Coherence Tomographyce Tomography (Single B-Scan)

Opticaly

Absence on Time-Domain OpticalCoherence Tomography

PValue

on Spectral-in Optical

Tomography

Presence on Spectral-Domain Optical

Coherence Tomography

Absence on Spectral-Domain Optical

Coherence Tomography

(0%) 6/38 (16%) 23/38 (61%) 0.041(0%) 3/34 (9%) 22/34 (65%) 0.248(0%) 3/27 (11%) 19/27 (70%) 0.248(0%) 1/18 (6%) 11/18 (61%) 1.000(5%) 4/38 (11%) 17/38 (45%) 0.683(9%) 2/34 (6%) 22/34 (65%) 1.000(7%) 4/27 (15%) 10/27 (37%) 0.683(6%) 2/18 (11%) 10/18 (56%) 0.045(3%) 5/38 (13%) 11/38 (29%) 0.221(15%) 4/34 (12%) 14/34 (41%) 1.000(0%) 5/27 (19%) 6/27 (22%) 0.074(11%) 2/18 (11%) 7/18 (39%) 0.617(8%) 5/38 (13%) 11/38 (29%) 0.724(6%) 2/34 (6%) 18/34 (53%) 0.618(19%) 1/27 (4%) 9/27 (33%) 0.221(0%) 2/18 (11%) 7/18 (39%) 0.480

ographeren

aingraph

senceDomaerence

0/380/340/270/182/383/342/271/181/385/340/272/183/382/345/270/18

zumab before inclusion in the study. Her BCVA was 20/400. In

Sayanagi et al � SD-OCT vs. TD-OCT after Ranibizumab

3-D scan mode, a foveal cut showed the intraretinal and sub-RPEfluid. In the superior and inferior scanned areas, Cirrus and Spec-tralis detected and clearly delineated subretinal fluid that could notbe observed at the fovea (Fig 4).

Case 4. An 89-year-old man with CNV resulting from exuda-tive AMD underwent 3 previous intravitreal injections of ranibi-zumab before inclusion in the study. His BCVA was 20/200�1.The Spectralis achieved a reduction in eye movement artifacts,whereas Stratus SD-OCT showed a wavy boundary. The Spectralisdetected and clearly delineated sub-RPE and intraretinal fluid (Fig 5).

Discussion

Ranibizumab is the first Food and Drug Administration-ap-proved therapy for CNV resulting from exudative AMD to

Table 4. Prevalence of Coexistence of Optical Coherence Tomand Spectral-Domain Optical Coher

OpticalCoherence

TomographyFindings

Spectral-DomainOptical Coherence

Tomography

Presence on Time-DomCoherence Tomo

Presence on Spectral-Domain Optical

Coherence Tomography

Ab

Coh

Subretinal fluid Spectralis (n � 22) 5/22 (23%)Copernicus (n � 33) 9/33 (27%)Cirrus (n � 28) 5/28 (18%)OCT-1000 (n � 18) 6/18 (33%)

Intraretinal cysts Spectralis 8/22 (36%)Copernicus 8/33 (24%)Cirrus 13/28 (46%)OCT-1000 6/18 (33%)

Intraretinal fluid Spectralis 10/22 (45%)Copernicus 13/33 (39%)Cirrus 16/28 (57%)OCT-1000 8/18 (44%)

Sub-RPE fluid Spectralis 12/22 (45%)Copernicus 18/33 (55%)Cirrus 17/28 (61%)OCT-1000 9/18 (50%)

RPE � retinal pigment epithelium.P values in bold are �0.05.

Table 5. Comparison and Correlation of Macular ThicknessCoherence Tomography and Spectral

Spectral-Domain Optical CoherenceTomography

Spectral-Domain OptCoherence Tomograp

TD-OCT (center point thickness) versusSpectralis (linear B-scan) 340.5�123.2Copernicus (linear B-scan) 184.8�115.6OCT-1000 (linear B-scan) 265.3�104.9Spectralis (3-dimensional scan) 319.4�124.0Copernicus (3-dimensional scan) 201.0�193.7Cirrus (3-dimensional scan) 182.3�103.4OCT-1000 (3-dimensional scan) 212.1�97.8

TD-OCT (center subfield thickness) vsCopernicus 267.9�169.2Cirrus 314.3�75.9OCT-1000 245.5�83.9

TD-OCT � time-domain optical coherence tomography.

P values in bold are �0.05.

improve the mean visual acuity. Although the optimal dosingschema has not been determined, frequent ranibizumab treat-ments are thought to be necessary to achieve these beneficialeffects. The pivotal phase 3 trials (MARINA and ANCHOR)used monthly injections, whereas another randomized studywith quarterly injections after 3 baseline monthly injectionshad worse visual results (Phase IIIb, multi-center, random-ized, double-masked, sham infection-control study of theefficacy and safety of ranibizumab [PIER]).12 In contrast, anOCT-guided pilot study, the PrONTO study, used a similarnumber of treatments as the PIER study, but had compara-ble visual results as monthly injections.7 It is important torecognize that PIER was a phase 3 randomized, controlledtrial, whereas PrONTO was a 40-patient, nonrandomized,

hy Findings on Time-Domain Optical Coherence TomographyTomography (3-Dimensional Scan)

Opticaly

Absence on Time-Domain OpticalCoherence Tomography

PValue

on Spectral-in Optical

Tomography

Presence on Spectral-Domain Optical

Coherence Tomography

Absence on Spectral-Domain Optical

Coherence Tomography

2 (0%) 6/22 (27%) 11/22 (50%) 0.0413 (0%) 12/33 (36%) 12/33 (36%) 0.0018 (0%) 9/28 (32%) 14/28 (50%) 0.0088 (0%) 7/18 (39%) 5/18 (28%) 0.0232 (0%) 2/22 (9%) 12/22 (55%) 0.4803 (9%) 2/33 (6%) 20/33 (61%) 1.0008 (4%) 7/28 (25%) 7/28 (25%) 0.0778 (0%) 4/18 (22%) 8/18 (44%) 0.1342 (0%) 9/22 (41%) 3/22 (14%) 0.0083 (9%) 8/33 (24%) 9/33 (27%) 1.0008 (4%) 9/28 (32%) 2/28 (7%) 0.0278 (6%) 4/18 (22%) 5/18 (28%) 0.3712 (9%) 2/22 (9%) 6/22 (27%) 0.6173 (3%) 8/33 (24%) 6/33 (18%) 0.0468 (4%) 7/28 (25%) 3/28 (11%) 0.0778 (0%) 7/18 (39%) 2/18 (11%) 0.023

etinal Thickness in Central Area on Time-Domain Opticalain Optical Coherence Tomography

Time-Domain OpticalCoherence Tomography

ComparisonP Value

Correlation P Value,Correlation Coefficient

231.4�75.8 <0.001 <0.0001, 0.617239.0�99.2 <0.001 <0.0001, 0.614242.6�90.7 0.491 0.00184, 0.678236.1�85.2 <0.001 0.008, 0.548238.8�102.7 0.002 0.0183, 0.436241.4�81.8 <0.001 <0.0001, 0.604239.8�92.7 0.404 <0.0001, 0.743

257.8�92.8 0.261 <0.0001, 0.625256.5�68.8 <0.001 <0.0001, 0.858264.1�83.1 0.325 <0.0001, 0.868

ograpence

aingraph

senceDomaerence

0/20/30/20/10/23/31/20/10/23/31/21/12/21/31/20/1

and R-Dom

icalhy

951

Ophthalmology Volume 116, Number 5, May 2009

noncontrolled investigator-sponsored trial, so the results ofthe PIER and PrONTO studies cannot be compared directly.Nevertheless, most physicians use OCT to guide treatmentdecisions with ranibizumab in CNV resulting from AMD,instead of monthly injections, as was performed in theclinical trials.

Previously when considering retreatment in AMD, FAcommonly was used as an indicator of CNV activity; how-ever, several reports have indicated poor agreement in in-terpretation of FA in AMD among the doctors, especiallyafter treatment.13–15 Optical coherence tomography is anoninvasive and powerful diagnostic tool that providescross-sectional views of the retina. Several reports havevalidated a good agreement in interpretation of TD-OCT

Figure 1. Single B-scan image on (top left) time-domain optical coherspectral-domain optical coherence tomography (SD-OCT) devices: (boCopernicus (Co; Reichert Inc., Depew, NY), and (bottom right) the Cirrdelineate the intraretinal and sub–retinal pigment epithelium (sub–RPEdelineates the cystic changes (arrowhead). Stratus visualized the sub-RPE fl(arrowhead).

Figure 2. Single B-scan image on (top left) time-domain optical coherspectral-domain OCT (SD-OCT) devices: (middle left) the OCT-1000 (ODepew, NY); (middle right) the Cirrus (Ci; Carl Zeiss Meditec, Inc., DubCirrus clearly visualized both subretinal and sub–retinal pigment epitheli(asterisk). The OCT-1000 clearly visualized sub-RPE fluid and drusen (arsubretinal fluid was not observed clearly on both devices (arrowhead). The

asterisk); however, the Spectralis delineated it more clearly than the Stratus d

952

images between observers and good correlation between FAand TD-OCT.15,16

Spectral-domain OCT has higher axial resolution andfaster acquisition speeds compared with TD-OCT that allowSD-OCT to provide not only the detailed views of intrareti-nal microstructure, but also 3-D images of the macula.These enhanced views should offer physicians a better viewof the macula than previous TD-OCT devices. The currentstudy investigated the ability to delineate and detect theCNV activity in patients with exudative AMD treated withintravitreal injection(s) of ranibizumab and found that SD-OCT demonstrated a significantly better ability to delineateand detect CNV activity compared with TD-OCT. In singlelinear B-scan mode, Spectralis and Cirrus were the best at

tomography (Stratus [S]; Carl Zeiss Meditec, Inc., Dublin, CA) and 3left) the OCT-1000 (O; Topcon, Inc., Paramus, NJ), (top right) thei; Carl Zeiss Meditec, Inc., Dublin, CA). The 3 SD-OCT devices clearly

(arrowhead and arrows, respectively), and in addition, the OCT-1000rrows), but showed hyporeflectivity corresponding to the intraretinal fluid

tomography (Stratus [S]; Carl Zeiss Meditec, Inc., Dublin, CA) and 4con, Inc., Paramus, NJ); (bottom left) the Copernicus (Co; Reichert Inc.,A); and (bottom right) the Spectralis (Sp; Vista, Heidelberg, Germany).ub–RPE) fluid (arrowhead and arrow, respectively) in addition to drusennd asterisk). The Copernicus visualized sub-RPE fluid (arrow); however,us and Spectralis visualized only sub-RPE fluid and drusen (arrowhead and

encettomus (C) fluiduid (a

ence; Toplin, Cum (srow aStrat

id.

Sayanagi et al � SD-OCT vs. TD-OCT after Ranibizumab

detecting subretinal fluid and intraretinal cysts, respectively.In single linear B-scan mode, SD-OCT scans the retinausing 3 to 14 times more A-scans per B-scan (Copernicus7427, HRA 1536, Cirrus 4096, and OCT-1000 4096 vs.Stratus 512 A-scans) and with 45 to 100 times faster acqui-sition speeds (Copernicus, 25 000 A-scans/second; HRA,40 000 A-scans/second; Cirrus, 27 000 A-scans/second;and OCT-1000, 18 000 A-scans/second vs. Stratus, 400A-scans/second), resulting in higher axial resolution, im-proved signal-to-noise ratios, and reduced eye movementartifacts. Therefore, it is not surprising that the SD-OCTdevices were superior in their ability to delineate CNVactivity over TD-OCT. In 3-D cube mode, SD-OCT scansthe retina faster, but with almost the same axial resolution asTD-OCT (Copernicus, 743 A-scans; HRA, 384 A-scans;Cirrus, 512 A-scans; OCT-1000, 512 A-scans vs. Stratus,512 A-scans) to produce a 3-D image of the macula. Three-dimensional scans allow visualization of the entire scannedarea, resulting in a superior ability to detect CNV activityover linear scans and the TD-OCT’s radial line/fast macularthickness map scans.

This study also compared retinal thickness measure-ments between the devices and found significant differencesin CPT and CSF between the TD-OCT and SD-OCT de-vices. The CPT of Spectralis was significantly thicker than

Figure 3. (First column) Infrared-red fundus image and (Second–Fifth cooptical coherence tomography (OCT) devices: (Second column) OCT-1Depew, NY), (Fourth column) Cirrus (Carl Zeiss Meditec, Inc., Dublin,on compound fundus image (a, b, c) indicate location of OCT scans, respside of the fovea (arrowheads).

Figure 4. (First column) Infrared-red fundus image, (Second column) sin(TD-OCT; Stratus; Carl Zeiss Meditec, Inc., Dublin, CA), and (Third aspectral-domain OCT (SD-OCT) devices: (Third column) Cirrus (CarHeidelberg, Germany). Black lines on compound fundus images (a, b, c) inintraretinal (arrowheads) and sub–retinal pigment epithelium fluid. In add

inferior scanned areas (arrows).

that of Stratus, whereas those on the Copernicus and Cirruswere thinner. There was no significant difference foundbetween the OCT-1000 and Stratus in CPT and CSF andbetween the Copernicus and Stratus in CSF. As mentionedin the previous studies,17,18 the major differences betweendevices is based on what line the software uses to determinethe outer retinal boundary. In addition, the method used tocalculate CPT and CSF differs between TD- and SD-OCTdevices. The Stratus calculates the CPT and CSF thicknessby analyzing the 6 linear B-scans of the fast macular thick-ness map or the 6-radial line scan pattern, whereas SD-OCTcalculates CPT and CSF by analyzing the 3-D cube data thatincludes 25 to 128 B-scans in the measured retinal area. Thenumber of retinal points analyzed to determine CPT andCSF on SD-OCT are significantly greater than that of Stra-tus TD-OCT; thus, Stratus may miss a region if it is notscanned in 1 of the 6 linear B-scans, whereas the 3-D cubescans of SD-OCT should not miss areas. Another possiblereason for the differences between devices is that the seg-mentation algorithms seemed to identify correctly the reti-nal boundaries better in the SD-OCT devices. Time-domainOCT has been shown to have poor boundary detection incases of CNV secondary to AMD.19,20

In summary, all the SD-OCT devices had higher detec-tion rates of CNV activity compared with TD-OCT after

s) B-scan images selected from 3-dimensional image on 4 spectral-domainTopcon, Inc., Paramus, NJ), (Third column) Copernicus (Reichert Inc.,and (Fifth column) Spectralis (Vista, Heidelberg, Germany). Black linesely. The subretinal fluid is observed most clearly on the higher and lower

scan image at the fovea using time-domain optical coherence tomographyourth columns) B-scan images selected from 3-dimensional images on 2ss Meditec, Inc., Dublin, CA) and (Fourth column) Spectralis (Vista,e location of OCT scans, respectively. Both TD- and SD-OCT visualizedCirrus and Spectralis clearly visualized subretinal fluid on the superior and

lumn000 (CA),ectiv

gle B-nd Fl Zeidicat

ition,

953

Ophthalmology Volume 116, Number 5, May 2009

ranibizumab injections for exudative AMD than TD-OCT.Moreover, the SD-OCT 3-D cube scans had higher detec-tion rates than the SD-OCT linear B-scans. Because thedistribution and number of scans were different among thedevices, comparison among the individual SD-OCT devicescould not be performed in this study. Larger investigationsare necessary to see if the differences in CNV activitydetected by SD- versus TD-OCT devices will lead to im-proved functional or visual outcomes after anti-VEGF ther-apy, or both.

References

1. Eye Diseases Prevalence Research Group. Prevalence of age-related macular degeneration in the United States. Arch Oph-thalmol 2004;122:564–72.

2. Kaiser PK, Do DV. Ranibizumab for the treatment of neovas-cular AMD. Int J Clin Pract 2007;61:501–9.

3. Rosenfeld PJ, Brown DM, Heier JS, et al, MARINA StudyGroup. Ranibizumab for neovascular age-related macular de-generation. N Engl J Med 2006;355:1419–31.

4. Brown DM, Kaiser PK, Michels M, et al, ANCHOR StudyGroup. Ranibizumab versus verteporfin for neovascularage-related macular degeneration. N Engl J Med 2006;55:1432–44.

5. Van de Moere A, Sandhu SS, Talks SJ. Correlation of opticalcoherence tomography and fundus fluorescein angiographyfollowing photodynamic therapy for choroidal neovascular

Figure 5. (Top) Single B-scan image using time-domain optical coheren(Bottom) spectral-domain OCT (Spectralis [Sp]; Vista, Heidelberg, Germaof the faster acquisition speed. Stratus shows a wavy boundary, whereasintraretinal fluid.

membrane. Br J Ophthalmol 2006;90:304–6.

954

6. Eter N, Spaide RF. Comparison of fluorescein angiographyand optical coherence tomography for patients with choroidalneovascularization after photodynamic therapy. Retina 2005;25:691–6.

7. Fung AE, Lalwani GA, Rosenfeld PJ, et al. An optical coher-ence tomography-guided, variable dosing regimen with intra-vitreal ranibizumab (Lucentis) for neovascular age-relatedmacular degeneration. Am J Ophthalmol 2007;143:566–83.

8. Wojtkowski M, Leitgeb R, Kowalczyk A, et al. In vivo humanretina imaging by Fourier domain optical coherence tomogra-phy. J Biomed Opt 2002;7:457–63.

9. Nassif N, Cense B, Park BH, et al. In vivo human retinalimaging by ultrahigh-speed spectral domain optical coherencetomography. Opt Lett 2004;29:480–2.

10. Leitgeb R, Drexler W, Unterhuber A, et al. Ultrahigh resolu-tion Fourier domain optical coherence tomography. Opt Ex-press [serial online] 2004;12:2156–65. Available at: http://www.opticsinfobase.org/oe/abstract.cfm?uri�oe-12-10-2156.Accessed November 2, 2008.

11. Chen TC, Cense B, Pierce MC, et al. Spectral domain opticalcoherence tomography: ultra-high speed, ultra-high resolutionophthalmic imaging. Arch Ophthalmol 2005;123:1715–20.

12. Regillo CD, Brown DM, Abraham P, et al, PIER Study Group.Randomized, double-masked, sham-controlled trial of ranibi-zumab for neovascular age-related macular degeneration:PIER Study year 1. Am J Ophthalmol 2008;145:239–48.

13. Friedman SM, Margo CE. Choroidal neovascular membranes:reproducibility of angiographic interpretation. Am J Ophthal-mol 2000;130:839–41.

14. Hole FG, Jorzik J, Schutt F, et al. Agreement among

mography (OCT; Stratus [S]; Carl Zeiss Meditec, Inc., Dublin, CA) andpectral-domain OCT offers a reduction in eye movement artifacts becauseralis clearly delineates the sub–retinal pigment epithelium (arrows) and

ce tony). SSpect

ophthalmologists in evaluating fluorescein angiograms in

Sayanagi et al � SD-OCT vs. TD-OCT after Ranibizumab

patients with neovascular age-related macular degeneration forphotodynamic therapy eligibility (FLAP-study). Ophthalmology2003;110:400–5.

15. van Velthoven ME, de Smet MD, Schlingemann RO, et al.Added value of OCT in evaluating the presence of leakage inpatients with age-related macular degeneration treated withPDT. Graefes Arch Clin Exp Ophthalmol 2006;244:1119–23.

16. Zhang N, Hoffmeyer GC, Young ES, et al. Optical coherencetomography reader agreement in neovascular age-related maculardegeneration. Am J Ophthalmol 2007;144:37–44.

17. Forooghian F, Cukras C, Meyerle CB, et al. Evaluation of

time domain and spectral domain optical coherence tomog-

S. Sharma - Travel Funding - Reichert, Inc.

raphy in the measurement of diabetic macular edema. InvestOphthalmol Vis Sci 2008;49:4290–6.

18. Leung CK, Cheung CY, Weinreb RN, et al. Comparison ofmacular thickness measurements between time domain andspectral domain optical coherence tomography. Invest Oph-thalmol Vis Sci 2008;49:4893–7.

19. Ray R, Stinnett SS, Jaffe GJ. Evaluation of image artifactproduced by optical coherence tomography of retinal pathol-ogy. Am J Ophthalmol 2005;139:18–29.

20. Leung CK, Chan WM, Chong KK, et al. Alignment artifactsin optical coherence tomography analyzed images. Ophthal-

mology 2007;114:263–70.

Footnotes and Financial Disclosures

Originally received: May 12, 2008.Final revision: September 25, 2008.Accepted: November 3, 2008.Available online: February 20, 2009. Manuscript no. 2008-570.

Cole Eye Institute, Cleveland Clinic Foundation, Cleveland, Ohio.

Financial Disclosure(s):The author(s) have made the following disclosure(s):

P. K. Kaiser - Scientific Advisory Boards - Heidelberg, Carl Zeiss Meditec.

Supported by Reichert, Inc., Depew, New York; Heidelberg Engineering,Heidelberg, Germany; and Carl Zeiss Meditec, Dublin, California.

Correspondence:

Peter K. Kaiser, MD, Cole Eye Institute, The Cleveland Clinic Foundation,9500 Euclid Avenue, Desk i3, Cleveland, OH 44195. E-mail: pkkaiser@

aol.com.

955