To evaluate the accuracy of 3 spectral-domain (SD) optical coherence tomography (OCT) devices (Topcon 3D-1000 [Topcon]; Cirrus HD [Carl Zeiss Meditec, Inc], and Spectralis OCT [Heidelberg Engineering]) before and after mydriasis for the diagnosis of diabetic macular edema.
Sixty-two eyes of 62 consecutive patients with diabetes without recent loss of vision referred for retinal control were assessed. Two scans were performed for each SD OCT instrument. Central retinal thickness was measured before and after pupil dilation. Pupil dynamic was studied using pMetrics pupillometer (iVIS Technologies), and lens opacity was measured by Pentacam densitometry (Oculus). The diagnostic accuracy of SD OCT devices was assessed by sensitivity, specificity, and area under the receiver operating characteristic curve. Logistic regression analysis was used to assess the effect of pupil size and lens opacity on the reliability of SD OCT in the acquisition of adequate images.
The area under the receiver operating characteristic curve for the Topcon 3D OCT device was 0.84, that for the Cirrus HD OCT device was 0.93, and that for the Spectralis OCT device was 0.91. Significant differences in area under the receiver operating characteristic curve before and after pupillary dilatation were not found. Sensitivity and specificity associated with the cutoff value for the best performance were 82% and 74% for the Topcon 3D OCT device, 90% and 87% for the Cirrus HD OCT device, and 90% and 84% for the Spectralis OCT device, respectively. The Topcon 3D OCT device had an 11.3% segmentation algorithm failure rate for the central millimeter of the fovea, and the nuclear lens density was significantly greater in these eyes than in those without failure (17.1 ± 1.1 mm vs 10.4 ± 0.2 mm; P < .05).
SD OCT is a useful tool to detect and to measure diabetic macular edema without the need for pupil dilatation.
Diabetes mellitus is widely accepted as the most common cause of avoidable loss of vision in working-age people in developed countries. The prevalence of diabetes for all age groups worldwide was estimated to be 2.8% in 2000 and 4.4% in 2030. The total number of people with diabetes is projected to rise from 171 million in 2000 to 366 million in 2030. Diabetic macular edema is a common complication of diabetic retinopathy and is the cause of most of the functional visual loss in patients with this condition. The presence of clinically significant macular edema increases the risk of moderate visual loss to approximately 30% to 50%, depending on the level of baseline visual acuity.
Diabetic macular edema was defined on the basis of stereoscopic fundus photography in Early Treatment Diabetic Retinopathy Study studies. However, this technique is complicated and difficult to use in daily practice and usually is replaced by contact fundus biomicroscopy, which has been found to be in close agreement with stereophotography. Noncontact fundus biomicroscopy generally is used for macular examination, although it has been shown to be slightly less sensitive than contact fundus biomicroscopy. Optical coherence tomography (OCT) is a new, noninvasive technique that provides retinal thickness profiles of the central retina with a resolution from 3.5 to 10 μm. In recent years, OCT has played an important role as a diagnostic and monitoring tool in many ocular diseases. Standard OCT assessment of diabetic macular edema has been adopted in multicenter trials in patients with diabetic retinopathy by the Diabetic Retinopathy Clinical Research network. Using Stratus OCT before and after pupil dilatation, the best quality of images and reproducibility for retinal nerve fiber layer thickness was found for dilated scanning. More recently, spectral-domain (SD; or Fourier) detection, in which a high-speed spectrometer is used to measure light echoes from all time delays simultaneously, has enhanced OCT capabilities. A recent study has shown that pupil dilatation did not influence reproducibility of retinal nerve fiber layer thickness and ganglion cell complex measurements made with the RTVue-100 Fourier-domain OCT device (Optovue Inc, Fremont, California, USA). According to this study, it was hypothesize that reliable measurements of macular thickness using SD OCT may be obtained without dilatation. Therefore, single measurements of central foveal thickness using an SD OCT device without pupil dilatation may constitute a new, useful, and cost-effective diagnostic tool. The objective of this cross-sectional study was to compare the usefulness and reliability of 3 SD OCT devices in the diagnosis of diabetic macular edema without the need of dilating the pupil.
A total of 62 consecutive patients with diabetes without recent loss of vision (in the 6 months before enrollment) referred by their primary care physicians to the ophthalmology services of the participating hospitals in compliance with the standard protocol for the care of patients with diabetes were recruited over a 9-month period. Patients were considered to be diabetic if they were taking any glucose-lowering medication. Exclusion criteria were patients with significant corneal opacities that could result in a poor OCT signal, patients with any ocular disease other than diabetes, and patients who had undergone any intraocular surgery, including cataract surgery. One eye per patient was studied (n = 62) and the study was limited to phakic eyes.
Noncontact lens biomicroscopy of the fundus was considered the gold standard. Clinically significant macular edema was defined according to the Early Treatment Diabetic Retinopathy Study criteria. The diagnosis of clinically significant macular edema (CSME) was made by an independent ophthalmologist (C.I.C.) who was blind to the results of the OCT measurements. Three commercially available SD OCT devices were used: a Topcon 3D-1000 (Topcon, Tokyo, Japan), a Cirrus HD (Carl Zeiss Meditec, Inc, Dublin, California, USA), and a Spectralis OCT (Heidelberg Engineering, Dossenheim, Germany). One eye of each patient was selected at random, and before pupil dilatation, pupillometry and measurements of the macular thickness with the 3 OCT devices were performed. Afterward and after pupil dilatation with 1 drop 1% tropicamide, lens opacity was measured by Pentacam densitometry (Oculus Pentacam HR; Oculus, Wetzlar, Germany), and the OCT macular scans with the 3 devices were repeated. Pupil measurement was performed with a pMetrics binocular pupillometer (iVIS Technologies, Taranto, Italy), and lens density using the Oculus Pentacam HR. The Pentacam HR captures Scheimpflug images of the anterior eye segment through a rotating measuring process that supplies pictures in 3 dimensions. The light conditions of the examination room were measured with a Digital Lux Meter (Sigma Instruments, New Dehli, India). An illumination of 25 lux was measured in the OCT examination room that corresponded to a light intensity defined by the pupillometer as “night driving.”
The ophthalmologist (F.J.L.M.) who performed OCT studies either before or after pupil dilatation was unaware of the diagnosis, that is, whether the patient had diabetic retinopathy or macular edema. The order of OCT acquisition was chosen randomly for each eye, and a minimum of 15 minutes elapsed between each examination. We attempted to use similar acquisition protocols on each instrument to assess central retinal thickness (mean thickness in the central 1000 μm diameter area) and midperipheral areas. All of the instruments provided a retinal thickness map based on the Early Treatment Diabetic Retinopathy Study model: 3 concentric circular areas are centered on the fixation point. The diameter of each area is 1, 3, and 6 mm for the 3 instruments. In each 1 of these 9 areas, the software calculates the mean value from all the thickness measurements on each single A scan. We used a 3-dimensional macular map (512 A-scans × 128 lines) with the Topcon 3D-1000, a macular cube (512 A-scans × 128 lines) with the Cirrus HD, and a fast macular map (768 A-scans × 25 lines) with the Spectralis.
In general, retinal thickness is measured automatically as the distance between the vitreoretinal interface and the anterior boundary of the retinal pigment epithelium. The location of these boundaries is determined by a thresholding algorithm that searches for the change in reflectivity present at each of these interfaces. Previous studies have shown that different SD OCT systems provide distinct values for central retinal thickness because different structures are used to define the outer border of the retina. For example, the Topcon 3D uses the top of the retinal pigment epithelium as a reference, whereas the Cirrus HD and Spectralis devices use the bottom of the retinal pigment epithelium. As a result, different systems cannot be used interchangeably for the measurement of macular thickness. Thus, although we considered edema detected by means of OCT to be present when foveal thickness was greater than a given cutoff point, we were unable to establish a common cutoff point of retinal thickness to diagnose macular edema when we used 3 different SD OCT instruments.
Statistical analyses were performed with Stata software version 10.1 (StataCorp LP, College Station, Texas, USA) adding the BLATPLOT, CONCORD, and DIAGT modules. Data were analyzed using Shapiro-Wilk test to evaluate the normality of sample distribution. The means and standard deviations were calculated for the central retinal thickness measurements from the 3 instruments before and after pupil dilatation. Differences between the central retinal thickness measurements obtained with the OCT devices before and after pupil dilatation were compared with the Mann–Whitney U test. We used the Wilcoxon paired rank-sum test to detect differences in signal intensity before and after pupil dilatation. Multivariate exact logistic regression was used to evaluate the influence of pupillary size and lens density on the reliability of nonmydriatic OCT in acquiring adequate measurements of the central macula. The ability of OCT without pupil dilatation to detect diabetic macular edema was analyzed using receiver operating characteristic (ROC) curves. The areas under the ROC curves (AUCs) were calculated and compared (chi-square test) to determine the most sensitive OCT device. We also selected the best cutoff for each OCT instrument (best tradeoff between sensitivity and specificity) and calculated the sensitivity, specificity, positive and negative predictive values, and positive and negative likelihood ratios for the diagnosis of diabetic macular edema without pupil dilatation. Concordance between nonmydriatic OCT and standard OCT for each device was characterized by the corresponding intraclass correlation coefficient with the 95% confidence interval for all macular regions obtained before and after pupil dilatation. A P value less than .05 was considered statistically significant.
There were 37 men and 25 women, with a mean age ± standard deviation of 61.6 ± 14.2 years. Type 1 diabetes mellitus was diagnosed in 12 patients, and type 2 diabetes mellitus was diagnosed in the remaining 50 patients. The mean duration of diabetes was 4.6 ± 1.0 years. Early nonproliferative diabetic retinopathy was found in 23 patients, moderate nonproliferative diabetic retinopathy was found in 13 patients, severe nonproliferative diabetic retinopathy was found in 15 patients, and proliferative diabetic retinopathy was found in 11 patients. CSME was diagnosed in 31 (50%) eyes by noncontact lens biomicroscopy. The mean corrected visual acuity (Snellen) was 0.69 ± 0.03.
Pupil dilation had no effect on the overall central retinal thickness measurements. Mean measurements for macular thickness in each of the 9 standard subfields obtained using the 3 OCT devices in mydriatic and nonmydriatic conditions are shown in Table 1 . In the nonmydriatic condition, the mean central retina thickness was 322.1 ± 18.2 μm for the Topcon 3D system, which was significantly lower than 334.5 ± 17.4 μm obtained with the Cirrus HD and 346.7 ± 17.4 μm obtained with the Spectralis ( P < .001, Wilcoxon signed-rank sum test).
|Topcon 3D OCT||Topcon 3D OCT (mydriasis)||P Value a||Cirrus HD OCT||Cirrus HD OCT (mydriasis)||P Value a||Spectralis OCT||Spectralis OCT (mydriasis)||P Value a|
|Central||322 ± 18||317 ± 18||.14||335 ± 17||335 ± 17||.92||347 ± 18||342 ± 18||.07|
|Inner nasal||315 ± 14||315 ± 15||.56||361 ± 12||358 ± 11||.12||382 ± 12||377 ± 12||.06|
|Outer nasal||280 ± 9||284 ± 9||.45||315 ± 7||315 ± 7||.73||333 ± 8||331 ± 9||.30|
|Inner temporal||311 ± 19||324 ± 16||.68||356 ± 13||354 ± 13||.79||373 ± 14||368 ± 13||.50|
|Outer temporal||252 ± 13||263 ± 11||.86||297 ± 8||294 ± 8||.65||312 ± 9||308 ± 8||.95|
|Inner superior||327 ± 15||335 ± 13||.40||363 ± 12||362 ± 11||.60||381 ± 13||376 ± 12||.41|
|Outer superior||279 ± 11||282 ± 9||.74||309 ± 10||305 ± 8||.77||320 ± 10||323 ± 10||.12|
|Inner inferior||328 ± 20||347 ± 16||.19||356 ± 13||355 ± 13||.93||375 ± 14||371 ± 13||.83|
|Inner nasal||284 ± 12||285 ± 9||.17||302 ± 8||305 ± 8||.89||315 ± 10||314 ± 9||.96|
The AUCs before and after pupil dilatation for the 3 SD OCT devices are shown in Table 2 , and the ROC curves are shown in Figure 1 . The AUC showed no difference between nonmydriatic and mydriatic OCT devices. Before pupil dilatation, there were statistically significant differences between AUC for the Topcon 3D compared with that of the Cirrus HD (0.84 vs 0.93; P < .001) and that of the Spectralis (0.84 vs 0.91; P = .03). Differences in AUCs between nonmydriatic Cirrus HD and Spectralis (0.93 vs 0.91; P = .390) were not observed. In nonmydriatic conditions, the best cutpoint of central retinal thickness for the diagnosis of diabetic macular edema was 265 μm in the case of the Topcon 3D device, 287 μm in case of the Cirrus HD device, and 299 μm in case of the Spectralis system. The corresponding sensitivity, specificity, percentage of correct classification, and the positive and negative likelihood ratios for each SD OCT system are detailed in Table 2 . Using these cutpoints, the diagnostic accuracy without pupil dilatation was 54.5% for the Topcon 3D device, 51.6% for the Cirrus HD device, and 53.2% for the Spectralis.
|AUC||P Value a||Cutoff value of best performance (μm)||Sensitivity (%)||Specificity (%)||Correctly classified (%)||LR+||LR−|
|Topcon 3D (mydriasis)||0.86||260||83||71||77||2.89||0.24|
|Cirrus HD (mydriasis)||0.93||276||94||81||87||4.93||0.08|