To assess the reproducibility of circumpapillary retinal nerve fiber layer (cpRNFL) thickness measurement (measurement agreement) and its color-coded classification (classification agreement) by Cirrus spectral domain optical coherence tomography (OCT) in pseudophakic eyes.
Two-hundred five participants having glaucoma or glaucoma suspected eyes underwent two repeated Cirrus OCT scans to measure cpRNFL thickness (optic disc cube 200 × 200). After classifying participants into three different groups according to their lens status (clear media, cataract, and pseudophakic), values of intra-class coefficient (ICC), coefficient of variance, and test-retest variability were compared between groups for average retinal nerve fiber layer (RNFL) thicknesses and that corresponding to four quadrant maps. Linear weighted kappa coefficients were calculated as indicators of agreement of color code classification in each group.
ICC values were all excellent (generally defined as 0.75 to 1.00) for the average and quadrant RNFL thicknesses in all three groups. ICC values of the clear media group tended to be higher than those in the cataract and pseudophakic groups for all quadrants and average thickness. Especially in the superior and nasal quadrants, the ICC value of the cataract group was significantly lower than that of the clear media and pseudophakic groups. For average RNFL thickness, classification agreement (kappa) in three groups did not show a statistically significant difference. For quadrant maps, classification agreement (kappa) in the clear media group was higher than those in the other two groups.
Agreement of cpRNFL measurement and its color code classification between two repeated Cirrus OCT scans in pseudophakic eyes was as good as that in eyes with clear crystalline lens. More studies are required to ascertain the effect of lens status on the reproducibility of Cirrus OCT according to different stages of glaucoma patients.
Optical coherence tomography (OCT) has become a popular means of diagnosing and determining progression of glaucoma as it allows noninvasive imaging of the retinal nerve fiber layer (RNFL). Cirrus spectral domain (SD) OCT (software version 4.0.1; Carl Zeiss Meditec, Dublin, CA, USA) utilizes SD OCT technology to acquire OCT data with better resolution (5 µm compared with approximately 10 µm axial resolution in tissue) about 70 times faster (27,000 vs. 400 A scans/sec) than time-domain (TD) OCT technology. Previous studies confirmed that peripapillary RNFL thickness measurements by Cirrus SD OCT were very reproducible both in healthy and glaucomatous eyes [
When analyzing circumpapillary RNFL (cpRNFL) thickness in glaucoma patients, reproducibility is an important reliability index, regardless of the imaging technique used. There have been many potential explanations for variability in cpRNFL thickness measurements using OCT. Factors such as cataract, signal strength (SS), pupil size, type of scan, the quadrant measured, and the thickness of RNFL all may influence the overall scan quality and calculated RNFL thickness [
Glaucoma and cataract are frequent causes of vision loss among the elderly, and these conditions often coexist [
Thus, the current study aims to assess the reproducibility of RNFL thickness measurement and its color code classification using Cirrus SD OCT in pseudophakic eyes, and then compare the obtained results with those in clear media and cataract eyes simultaneously.
Glaucoma and glaucoma suspect patients were consecutively enrolled from the Glaucoma-Cataract Clinic of Siloam Eye Hospital from March 2012 to May 2012. The study was approved by the institutional review board and ethics committee of Siloam Eye Hospital, and complied using the tenets of the Declaration of Helsinki. Informed consent was obtained from all participants prior to their participation in the study.
Upon initial evaluation, all participants underwent a complete ophthalmologic examination including best-corrected visual acuity (BCVA), intraocular pressure measurement using a Goldmann applanation tonometer, and slit lamp and fundus examination. Perimetry was performed by experienced technicians using the Humphrey Visual Field Analyzer (Cirrus OCT, Carl Zeiss Meditec) with the 30-2 SITA Standard algorithm. All cataracts were staged according to the slit lamp-based Lens Opacities Classification System (LOCS) III.
Participants included those with a diagnosis of any form of glaucoma or glaucoma suspect with an age greater than 60 years. Glaucomatous eyes were defined as those having glaucomatous visual field (VF) defects confirmed by at least two reliable VF examinations and the presence of a compatible glaucomatous optic disc that showed increased cupping (a vertical cup-disc ratio >0.7), a difference in vertical cup-disc ratio of >0.2 between eyes, diffuse or focal neural rim thinning, disc hemorrhage, or RNFL defects. A glaucomatous VF defect was defined as having three or more significant (
All participants were categorized into three groups according to their lens status. Patients with a degree of cataract more than grade II by LOCS classification were grouped as the cataract group, and patients who had received uncomplicated cataract surgery with "in the bag" implantation of a Biovue (Ophthalmic Innovations International, Ontario, CA, USA) intraocular lens were grouped as the pseudophakic group. Participants with clear media were classified as clear media group.
All individuals were evaluated with the Cirrus SD OCT (software version 4.0.1, Carl Zeiss Meditec) to measure cpRNFL thickness (optic disc cube 200 × 200 scan) twice during the same day, with short breaks between each measurement. The OCT scan was performed by a single well-trained technician through a dilated pupil. Between each scan, the patient was asked to sit back and rest for a few minutes before proceeding to the next scan. The technician applied artificial tears before the subsequent scan, and then repositioned the participants.
Optic disc cube protocol generates a cube of data through a 6 mm square grid by acquiring a series of 200 horizontal scan lines each composed of 200 A-scans. For analysis, Cirrus algorithms identify the center of the optic disc and automatically place a calculation circle of 3.46 mm diameter around it. The anterior and posterior margins of the RNFL are delineated, and after extracting from the data cube 256 A-scan samples along the path of the calculation circle, the system calculates the cpRNFL thickness at each point on the circle. Scans with misalignment, segmentation failure, or decentration of the measurement circle were excluded from analysis. From the OCT database, the following data were collected: two repeated measures of cpRNFL thickness and their stoplight color code from quadrant maps, two repeated measures of average RNFL thickness with stoplight color classification, and SS. Scans with high SS values were defined as those with scores greater than or equal to 5 in two consecutive scans, while other scans were defined as scans with low SS (scans showing a SS of <5).
When data from both eyes were eligible for analysis, only one eye from each patient was randomly selected and used for data analysis. No participant was excluded based on SS score.
The method employed to determine the reproducibility of RNFL thickness measurement between two repeated OCT scans (measurement agreement) was the intra-class correlation coefficient (ICC). Linear weighted kappa coefficients were calculated as indicators of agreement for RNFL thickness classification between two repeated OCT scans (classification agreement). When the number of categories was less than two by two, the result was marked as 'not available' due to insufficient number of categories to perform the classification agreement test. Coefficient of variation (CV) and test-retest variability were also computed to determine the reproducibility of the Cirrus OCT scan used in this study. The CV, expressed as a percentage (%), was calculated as the square root of the variance divided by the mean RNFL thickness of two repeated measures. Test-retest variability, measured in micrometers, was calculated as twice the square root of the variance among two repeated measures.
Inter-scan measurement difference, calculated by the RNFL thickness difference obtained between two repeated measures, was compared among groups using ANOVA and post-hoc multiple comparisons test.
To determine the effect of signal strength on Cirrus OCT reproducibility, we performed Pearson's correlation analysis. After we categorized participants into two different SS groups (high SS and low SS), we calculated ICC and kappa values for each group.
Data were analyzed using statistical software SPSS ver. 15.0 (SPSS Inc., Chicago, IL, USA) and R software ver. 2.10.1 (GNU General Public License). A
Of the 300 individuals initially enrolled in the study, 95 eyes were excluded due to segmentation error, or scan circle decentration, in any of the repeated Cirrus OCT scans. In total, 205 eyes of 205 individuals were included for the final analysis. Participants' demographic and ophthalmic characteristics are shown in
After categorizing all participants into three groups (clear media, cataract, and pseudophakia), there were 73 eyes in the clear media group, 74 eyes in the cataract group, and 58 eyes in the pseudophakic group, respectively. Among those three groups, there was no statistically significant difference in relation to age, gender, and proportion of glaucoma or glaucoma suspect. Logarithm of the minimum angle of resolution BCVA in cataract group was significantly worse than that in the clear media group (
To compare the measurement reproducibility of RNFL thickness between the three groups, ICC, CV, and test-retest variability values were calculated. Also, linear weighted kappa coefficients were calculated as agreement indicators of the probability code results for RNFL thickness analysis. Those values are shown in
The relationship between SS and inter-scan measurement difference with respect to RNFL thickness is summarized in
As OCT manufacturers recommended that SS score of an acceptable image quality is 5 or higher, we categorized all participants into two groups according to their SS values, participants with high SS (SS ≥5 in two consecutive scans, n = 140) and those with low SS (SS <5 in either scan, n = 65), then calculated ICC and kappa values for each group (
In the present study, we evaluated the reproducibility of cpRNFL thickness measurement and its color code classification by Cirrus SD OCT in pseudophakic eyes, and then compared the results with those in clear media and cataract eyes. As a result, ICC and kappa values of pseudophakic eyes were as good as that of clear media eyes despite a slightly lower average signal strength. CV and test retest variability of the pseudophakic group were also similar to those of clear media group except in the inferior quadrant. To our best knowledge, there has been no prior study that simultaneously compares reproducibility of Cirrus SD OCT in three different groups (clear media, cataract, and pseudophakic eyes).
OCT is an important method used in diagnosing and determining the progression of glaucoma [
Cataract is a common cause of media opacity that can affect the quality of diagnostic imaging devices for glaucoma like OCT in the elderly. Clear optical media will optimize both forward and back light scattering, whereas media opacity caused by cataracts will increase light scattering and absorption. Previous studies using Stratus TD OCT reported that lens opacities may affect the RNFL thickness measurements [
One possible explanation for this is that the effect of low SS caused by lens opacity in the cataract group on RNFL measurement variability was minimized by exclusion of scans with low SS.
Results regarding the RNFL color code classification from Cirrus OCT were quite interesting. For quadrant maps, the color code classification of the nasal quadrant appeared to be the most variable among all three groups as previously investigated [
The result of inter-scan measurement difference, CV, and test-retest variability were found to be higher in the cataract group, while values of the remaining two groups (pseudophakic and clear media group) were similar to each other in temporal, superior, nasal quadrants, and average RNFL thickness. However, only in the inferior quadrant, the inter-scan measurement difference, CV, and test-retest variability in the pseudophakic group were higher than those of the clear media group. There are some potential explanations for this observation. Of several statistical methods for assessing reliability, CV and test-retest variability can be caused by both intra-individual variability and inter-individual variability, in contrast to ICC values. In this study population, the ratio of glaucoma patients to glaucoma suspects was higher in the pseudophakic group (25 : 33, 75.8%) than in the clear media group (30 : 43, 69.8%), as shown in
A previous study performed by the authors demonstrated that the presence of cataract significantly affects the RNFL thickness measurements by Cirrus SD OCT and Stratus TD OCT. The changes in the RNFL thickness measurements following cataract surgery were more prominent for Cirrus SD OCT than Stratus OCT [
As we expected, the average SS in the pseudophakic group was somewhat lower than that in the clear media group. Na et al. [
Cheng et al. [
There are several strengths of our study. First, a large population of 205 patients were enrolled indicating that our study has sufficient statistical power. Second, all participants were classified into three different groups according to lens status by a single experienced ophthalmologist (JHK). Third, as a single experienced technician operated all OCT scans, inter-operator variability was minimized.
The current study has several limitations. First, some patients with minimal lens opacity might be classified into clear media group considering the mean age of study population was over 60 years old. These ambiguities in group classification may have adverse effects on our results. Second, a direct comparison of reproducibility with Cirrus OCT and Stratus OCT RNFL thickness measurements in pseudophakic eyes was not allowed since we did not have both devices in our clinic. Third, due to limited participation of patients, we were not able to compare the reproducibility observed with preoperative and postoperative scans in the same patients. Further studies are required to ascertain the effect of lens status on reproducibility of Cirrus OCT according to different stages of glaucoma.
The present study demonstrated that the Cirrus OCT measurements of cpRNFL thickness were very reproducible in pseudophakic eyes, as well as in clear media eyes. However, even after successful cataract surgery, there is a possibility of SS attenuation in comparison with clear media eyes, probably due to the interference of light transmittance by implanted intraocular lens. Thus, when interpreting the results of Cirrus OCT scans in pseudophakic eyes, factors that can compromise the test results should be carefully considered.
No potential conflict of interest relevant to this article was reported.
Bland-Altman plot analysis of the average retinal nerve fiber layer (RNFL) thickness measured by two repeated Cirrus optical coherence tomography scans. Note the greater variation between two repeated average RNFL thickness measurements in the cataract group than in the other groups. (A) Clear media. (B) Cataract. (C) Pseudophakic. STD = standard deviation.
Participant characteristics
GS = glaucoma suspect; BCVA = best-corrected visual acuity; logMAR = logarithm of the minimum angle of resolution; SS = signal strength.
'≒' indicates similar values without statistically significant difference between two groups.
*Value from ANOVA or chi-square test; †Pairwise comparison after ANOVA with Bonferroni correction; ‡High SS (SS ≥5 in two consecutive scans) and low SS (SS <5 in either scan).
ICC and Fleiss' generalized kappa of probability codes of circumpapillary retinal nerve fiber layer thickness measurements obtained using Cirrus optical coherence tomography
ICC = intra-class correlation coefficient; NA = not available.
'≒' indicates similar values without statistically significant difference between two groups.
*ICC values with 95% confidence interval in parenthesis, calculated with the two-way random effects model using the absolute agreement definition; †Mean on top with 95% confidence interval in parentheses; ‡Due to insufficient number of categories to perform test.
Coefficient of variation and test-retest variability of circumpapillary retinal nerve fiber layer thickness measurements with the Cirrus optical coherence tomography
'≒' indicates similar values without statistically significant difference between two groups.
*Calculated as the square root of the variance divided by the mean thickness of two repeated measures; †Calculated as two times the square root of the variance of two repeated measures.
Comparison of inter-scan measurement difference between two repeated Cirrus optical coherence tomography scans among three groups
'≒' indicates similar values without statistically significant difference between two groups.
*Obtained by subtracting value of optical coherence tomography scan with lower one from that of scan with the greater value; †Value from ANOVA; ‡Pairwise comparison after ANOVA with Bonferroni correction.
Relationship between signal strength and inter-scan retinal nerve fiber layer thickness measurement difference of repeated optical coherence tomography scans
SS = signal strength.
*Obtained by obtaining the difference between optical coherence tomography scan values; †Correlation coefficient from Pearson correlation with
ICC and kappa values in each group classified by signal strength
ICC = intra-class correlation coefficient; NA = not available.
'≒' indicates similar values without statistically significant difference between two groups.
*ICC values with 95% confidence interval in parenthesis, calculated with the two-way random effects model using the absolute agreement definition; †Mean on top with 95% confidence interval in parentheses; ‡High SS (SS ≥5 in two consecutive scans) and low SS (SS <5 in either scan); §Due to insufficient number of categories to perform test.