ORIGINAL STUDY
Interocular Retinal Nerve Fiber Layer Thickness Symmetry Value in Normal Young Adults Donghyun Jee, MD,* Seung Woo Hong, MD,*w Youn Hea Jung, MD,* and Myung Douk Ahn, MD*
Purpose: The aim of this study was to determine the normal range of interocular peripapillary retinal nerve fiber layer (RNFL) thickness symmetry value in young adults. Factors affecting interocular RNFL thickness symmetry were also investigated. Materials and Methods: Both eyes of 241 ophthalmologically and neurologically normal subjects were scanned using optical coherence tomography (200200 optic disc cube protocol). The effect of ocular cyclotorsion on interocular RNFL thickness symmetry was determined and mathematically accounted for. Symmetry value between the right and left RNFL thickness values was calculated. Linear regression analyses were used to identify correlations between the corrected interocular symmetry value and interocular differences in refractive error, axial length, superior and inferior temporal retinal artery and vein location, and optical coherence tomography signal strength. Results: The mean interocular RNFL thickness symmetry value before and after correction of ocular cyclotorsion was 0.8791 ± 0.0665 and 0.9044 ± 0.0571 (P < 0.001), respectively. Interocular differences in axial length, inferior temporal retinal arcade location, and refractive error were weakly and negatively correlated with the ocular cyclotorsion-corrected symmetry value (P < 0.05). Anatomic differences between the eyes in the location of the superior temporal retinal artery and vein were strongly correlated with the corrected symmetry value (P < 0.01, R2 = 0.208). Conclusions: On the basis of the data obtained here from normal individuals, a corrected symmetry value of 21 mm Hg; reproducible VF defect in either eye; unreliable VF testing (false-positive/false-negative rate >15%, fixation loss >20%); and any suspicious RNFL defect on fundus photographic images.
Volume 23, Number 8, October/November 2014
www.glaucomajournal.com |
e125
Jee et al
J Glaucoma
Volume 23, Number 8, October/November 2014
FIGURE 1. Effect of ocular cyclotorsion on interocular retinal nerve fiber layer (RNFL) thickness symmetry value. In the original scan result (A), the symmetry value was 76%. After an 8.4-degree clockwise rotation of the right eye’s RNFL thickness profile (B), the symmetry increased to 95%.
Each normal subject underwent a comprehensive ophthalmologic evaluation. Manifest refractions were performed with an autorefractometer (Canon R-F10; Canon Inc., Japan) and VF examinations were conducted using the SITA fast VF protocol of the Humphrey VF analyzer (HFA II 750-4.1 2005; Carl Zeiss Meditec Inc.). Axial length measurements were taken using the IOL Master (Carl Zeiss Meditec Inc.) and optic disc; fundus and RNFL photographs were obtained with a digital fundus cameras (Canon CF-60UD; Canon Inc.). All images were digitally recorded and analyzed by a single evaluator (S.W.H.). After the subjects were aligned properly, the eyes of the normal subjects who met the eligible criteria were scanned using the Cirrus HD OCT system (version 4.0.0.64; Carl Zeiss Meditec Inc.) after pharmacologic pupil dilation. Three individual 200 200 cube optic disc scans were obtained. Only scans that did not have movement/blink artifacts and had a signal strength Z7 were accepted. Unacceptable scans were discarded and a new subject was considered. Three quality scans were obtained for each eye, and among these the scan with the highest signal strength and least eye movement was selected to be included in the analyses. The RNFL thickness value at each of the 256 measurement points (0 to 255) was recorded. Pearson correlation coefficient between the right and left RNFL thickness values was calculated for each subject and was defined as the interocular RNFL thickness symmetry value. Of the symmetry measurements, we chose Pearson correlation coefficient because the Cirrus OCT RNFL thickness report provides an interocular symmetry value, which is the Pearson correlation coefficient multiplied by 100. However, this symmetry value needs to be corrected for ocular cyclotorsion. To do this, we drew a line between the geographic centers of the optic nerve head and the fovea. The point at which this reference line crossed the
e126 | www.glaucomajournal.com
scan circle became the new reference point (point 0). Then, we rearranged the sequence of RNFL thickness result to the new reference point. Because the Cirrus OCT RNFL thickness map does not contain the fovea, we drew the reference line on the fundus photograph and moved it to the RNFL thickness map. This step-by-step process was as follows (Fig. 2). Step 1: On the fundus photograph, we outlined the optic disc margin and calculated the geographic center of the optic nerve head using image-editing software (Photoshop CS5, ver. 12.0.1; Adobe Inc., San Jose, CA). We located the geographic centers of the optic nerve head (point c) and the fovea (point f) and drew a line between the 2 points. Step 2: We marked the temporal border of the superior temporal retinal vein (point a) and the temporal border of inferior temporal retinal vein (point b) where they crossed the disc margin. Step 3: We constructed a triangle, using points a, b, and f as corners, and then measured the angle a between sides a-b and a-f, the angle b between sides a-b and b-f, and the angle y between side a-f and the line c-f. Step 4: On the RNFL thickness deviation map, we marked the points where the temporal border of the superior temporal retinal vein and the temporal border of the inferior temporal retinal vein crossed the disc margin (points a* and b*, respectively). Then, we drew a line between points a* and b*. Step 5: At point a*, we used a protractor to measure angle a and drew a line to the temporal direction. At point b*, we measured angle b and drew a line to the same direction. We then marked the point f* where the 2 lines met. Step 6: at point f*, we measured angle y and drew a line in the disc direction (blue line). Finally, we marked the point where the scan circle and the line met. This point was set as the new reference point (point 0). Vessel anatomy was also examined and quantified in the following manner (Fig. 3). On the RNFL thickness r
2013 Lippincott Williams & Wilkins
J Glaucoma
Volume 23, Number 8, October/November 2014
Interocular RNFL Symmetry Value in Normal Adults
FIGURE 2. Fundus photograph and retinal nerve fiber layer (RNFL) thickness deviation map image showing reference lines. The blue line on the fundus photograph (A) runs between the geographic centers of the optic nerve head (point c) and the fovea (point f). The red triangle has corners at the fovea (point f) and the points where the temporal border of the superior temporal retinal vein (point a) and the temporal border of inferior temporal retinal vein (point b) cross the disc margin. Side a-f and the line c-f make the angle y between them. B, On the RNFL thickness deviation map, a similar triangle was constructed with 2 corners at the points where the temporal border of the superior temporal retinal vein (point a*) and the temporal border of inferior temporal retinal vein (point b*) cross the disc margin and side a*-b*. The other corner of the triangle was point f*. Angle y was measured and a line was drawn in the direction of the disc (blue line). Finally, point r was marked where the scan circle (red circle) and the line met.
deviation map, where the new reference line was drawn, we marked the points where the scan circle and the outer borders of the superior and inferior temporal retinal artery and vein met. Next, the middle points between the outer borders on the scan circle were marked. From the center of the scan circle, the angle between the middle point of the superior temporal retinal artery and vein and the new reference point was measured, as was the angle between the middle point of the inferior temporal retinal artery and vein and the new reference point. Linear regression analyses were performed to determine the correlations between corrected RNFL thickness symmetry values and interocular differences in axial length, refractive error spherical equivalent (SE), scan signal strength, and superior temporal and inferior temporal artery and vein locations. The analyses were performed with the symmetry value as the dependent variable. Drawings and measurements on images were carried out using an image-editing software (Photoshop CS5, ver. 12.0.1; Adobe Inc., San Jose, CA). Statistical analyses and graph plotting were performed using SPSS statistical software (version 13.0, SPSS Corp., IL, Chicago), and statistical significance was defined as a P < 0.05.
Mean original interocular RNFL thickness symmetry value was 0.8791 ± 0.0665, which increased to 0.9044 ± 0.0571 after correction for ocular cyclotorsion (P < 0.001; paired t test). On an average, the new reference point for ocular cyclotorsion correction (new point 0, the point where reference line met the OCT scan circle) was point
RESULTS Study subjects were mostly men [230 men (95.4%) and 11 women (4.6%)] and had a mean age of 21.4 ± 1.6 years (range, 19 to 28 y). All subjects were ethnically Korean and their ocular information is summarized in Table 1. On an average, the angle between the superior retinal artery and vein line and the reference was smaller in the right eye (76.26 degrees) than in the left eye (79.34 degrees; P < 0.001; paired t test). The OCT scan signal strength was also higher in the right eye than in the left eye (P = 0.028). No statistically significant interocular differences were observed for SE, axial length, average RNFL thickness, or the angle between the inferior retinal artery and vein line and the reference line. Table 1 shows the difference between the right eye and the left eye. r
2013 Lippincott Williams & Wilkins
FIGURE 3. Measuring the angles between the new reference point (to correct for ocular cyclotorsion) and the middle points of the superior and inferior retinal arteries and veins allowed us to correct for ocular cyclotorsion. The points where the scan circle and the outer borders of the superior and inferior temporal retinal artery and vein met (yellow lines and green lines, respectively). The middle point between the outer borders of superior temporal retinal artery and vein (yellow arrowhead), and the middle point between outer borders of inferior temporal retinal artery and vein (green arrowhead) are also marked. The angle a lies between the middle point of the superior temporal retinal artery and vein and the new reference point (blue arrowhead). The angle b lies between the middle point of the inferior temporal retinal artery and vein and the new reference point.
www.glaucomajournal.com |
e127
Volume 23, Number 8, October/November 2014
250.53 ± 2.168 (of 256 point of RNFL thickness) in the right eye and point 251.16 ± 3.153 in the left eye. Figure 4 shows symmetry value distributions before and after correction for ocular cyclotorsion. Table 2 shows the percentile distributions of the original and corrected interocular RNFL thickness symmetry values. Figure 5 shows the univariate linear regression analysis of the effect of various interocular differences on the corrected RNFL thickness symmetry value. The absolute interocular difference of the angle between the superior retinal artery and vein line, and the reference line was strongly and negatively correlated with the RNFL thickness symmetry value (P < 0.001, R2 = 0.208). However, absolute interocular differences in SE, axial length, and the angle between the retinal artery and vein and the reference lines were weakly but statistically negatively correlated with the corrected RNFL thickness symmetry value (P < 0.05). The difference between mean signal strength in the right and left eye scans was not significant (P = 0.10). Interocular SE differences were moderately correlated with interocular axial length differences (r = 0.484, P < 0.001). Interocular differences of SE and axial length were weakly correlated with interocular differences in the angle difference between the inferior retinal artery and vein line and the reference line (r = 0.184, P < 0.001 and r = 0.247, P < 0.001, respectively). Other interocular differences were not significantly correlated with each other.
0.64 ± 0.64 (0-2) 0.028 0.129 ± 0.901 ( 2 to + 2) 8.44 ± 1.109 (7-10) 8.56 ± 0.982 (7-10)
3.16 ± 2.59 (0-14.55) 0.098 0.436 ± 4.073 ( 14.11 to 14.55) 98.21 ± 8.87 (74-124.85) 98.64 ± 8.78 (72.62-124.53)
5.66 ± 5.35 (0-33.75) 0.353 0.467 ± 7.784 ( 33.75 to 22.50) 295.76 ± 10.45 (260.16-319.22) 296.24 ± 10.63 (55.94-320.63)
0.180 ± 0.212 (0-1.76) 5.60 ± 5.06 (0-28.13) 0.210 < 0.001 0.022 ± 0.277 ( 1.31 to + 1.76) 3.081 ± 6.900 ( 28.13 to 15.47) 24.850 ± 1.2092 (22.75-29) 79.34 ± 9.89 (43.59-108.28) 24.872 ± 1.2318 (21.56-28.77) 76.26 ± 9.67 (46.41-99.84)
e128 | www.glaucomajournal.com
Signed difference (right eye left eye). *Paired t test.
Absolute Difference
0.084
Right Eye
P* Signed Difference
0.0565 ± 0.703 ( 2.5 to + 3.38)
Left Eye
2.449 ± 2.3445 ( 10.125 to + 1.385) 2.505 ± 2.3522 ( 11 to + 4.125)
DISCUSSION
Spherical equivalent of refractive error (range) (D) Axial length (range) (mm) Angle between the superior retinal artery and vein line and reference line (range) (deg.) Angle between the inferior retinal artery and vein and reference line (range) (deg.) Average retinal nerve fiber layer thickness (range) (mm) Signal strength of scan
Ocular variables
TABLE 1. Summary of Ocular Information From Normal Subjects (n = 241)
J Glaucoma
0.4632 ± 0.5316 (0-3.375)
Jee et al
In the previous studies, the RNFL assessment was based on comparisons with normative thickness data, which showed relatively good sensitivity and specificity for glaucomatous damage detection.12–21 However, some patients have unique RNFL thickness profiles that largely deviate from the normative RNFL thickness profile. In these patients, comparing RNFL thickness with normative data does not provide useful information. Glaucomatous damage would decrease interocular RNFL thickness symmetry independent of a patient’s RNFL thickness profile configuration. Because little is known about interocular RNFL symmetry in healthy subjects, we examined this parameter to better understand how this information can be used to detect glaucomatous damage. The previous studies have investigated interocular RNFL thickness symmetry in normal adults.22–24 Mwanza et al22 used spectral-domain OCT and reported that an interocular average RNFL thickness difference of >9 mm may be indicative of glaucomatous damage. Budenz reported similar results by using time-domain OCT.23 Our study is somewhat different from these previous studies because we focused on comparing RNFL thickness symmetry between the right and left eyes. Further, our study used a new technique to remove ocular cyclotorsional effects from interocular RNFL symmetry data. There are other ways to statistically evaluate symmetry between 2 variables. Ghadiali et al24 used the coefficient of determination (R2) to successfully evaluate and compare interocular RNFL thickness symmetry profiles of multiple subjects. The coefficient of determination reflects the absolute thickness difference between the right and left eye. Whereas Pearson correlation coefficient reflects the correlation between the right and left eye. We chose to use the Pearson correlation coefficient for symmetry evaluation because glaucoma practitioners are familiar with OCT r
2013 Lippincott Williams & Wilkins
J Glaucoma
Volume 23, Number 8, October/November 2014
Interocular RNFL Symmetry Value in Normal Adults
FIGURE 4. Histogram showing the distribution of interocular retinal nerve fiber layer (RNFL) thickness symmetry values. A, Original RNFL thickness data. B, Data that have been corrected for ocular cyclotorsion.
symmetry values and because the Pearson correlation coefficient is more convenient in statistical analyses. In addition, 2 coefficients of determination result from 1 subject (right eye, left eye) and one of these 2 values, alone, does not fully reflect interocular symmetry. We did not want to choose between the 2 numbers and we were unsure how to resolve the complication of having 2 values in statistical analyses, especially in analyzing the ocular factors affecting the symmetry value. Our data analyses removed ocular cyclotorsional effects on RNFL thickness profiles by creating and using a reference line that was drawn between the foveal and disc centers. The line largely corresponded to the horizontal raphe of the retina. Other geographic markers, such as retinal blood vessels, can also be used as a reference to correct for ocular cyclotorsion; however, blood vessel patterns are highly variable and can be affected by disease (eg, diabetes, glaucoma). The horizontal raphe is a suitable reference line because it has a relatively straight configuration and is easy to identify. Moreover, it is both the anatomic and functional divisor of the superior and inferior RNFL, and its location is minimally affected by disease processes. Even when subjects were properly aligned in the OCT machine before scanning, correcting for ocular cyclotorsion significantly increased the interocular RNFL thickness symmetry value. As shown in Figure 1, even a 6-pixel point rotation of the right eye can translate into a marked 19% difference in interocular symmetry. This finding suggests that scanning conditions can considerably affect OCT interocular RNFL thickness symmetry value results, and without correction for ocular cyclotorsion healthy normal individuals may produce abnormal RNFL symmetry value. The
symmetry value percentile distribution also changed. The first and fifth percentile of the original symmetry value was 0.7242 and 0.7697, respectively. After correcting for cyclotorsion, the first and fifth percentile of the symmetry value increased to 0.7714 and 0.8105, respectively. This finding suggests that the right and left eye RNFL thickness profiles are more highly correlated and that a symmetry value
Interocular Retinal Nerve Fiber Layer Thickness Symmetry Value in Normal Young Adults Donghyun Jee, MD,* Seung Woo Hong, MD,*w Youn Hea Jung, MD,* and Myung Douk Ahn, MD*
Purpose: The aim of this study was to determine the normal range of interocular peripapillary retinal nerve fiber layer (RNFL) thickness symmetry value in young adults. Factors affecting interocular RNFL thickness symmetry were also investigated. Materials and Methods: Both eyes of 241 ophthalmologically and neurologically normal subjects were scanned using optical coherence tomography (200200 optic disc cube protocol). The effect of ocular cyclotorsion on interocular RNFL thickness symmetry was determined and mathematically accounted for. Symmetry value between the right and left RNFL thickness values was calculated. Linear regression analyses were used to identify correlations between the corrected interocular symmetry value and interocular differences in refractive error, axial length, superior and inferior temporal retinal artery and vein location, and optical coherence tomography signal strength. Results: The mean interocular RNFL thickness symmetry value before and after correction of ocular cyclotorsion was 0.8791 ± 0.0665 and 0.9044 ± 0.0571 (P < 0.001), respectively. Interocular differences in axial length, inferior temporal retinal arcade location, and refractive error were weakly and negatively correlated with the ocular cyclotorsion-corrected symmetry value (P < 0.05). Anatomic differences between the eyes in the location of the superior temporal retinal artery and vein were strongly correlated with the corrected symmetry value (P < 0.01, R2 = 0.208). Conclusions: On the basis of the data obtained here from normal individuals, a corrected symmetry value of 21 mm Hg; reproducible VF defect in either eye; unreliable VF testing (false-positive/false-negative rate >15%, fixation loss >20%); and any suspicious RNFL defect on fundus photographic images.
Volume 23, Number 8, October/November 2014
www.glaucomajournal.com |
e125
Jee et al
J Glaucoma
Volume 23, Number 8, October/November 2014
FIGURE 1. Effect of ocular cyclotorsion on interocular retinal nerve fiber layer (RNFL) thickness symmetry value. In the original scan result (A), the symmetry value was 76%. After an 8.4-degree clockwise rotation of the right eye’s RNFL thickness profile (B), the symmetry increased to 95%.
Each normal subject underwent a comprehensive ophthalmologic evaluation. Manifest refractions were performed with an autorefractometer (Canon R-F10; Canon Inc., Japan) and VF examinations were conducted using the SITA fast VF protocol of the Humphrey VF analyzer (HFA II 750-4.1 2005; Carl Zeiss Meditec Inc.). Axial length measurements were taken using the IOL Master (Carl Zeiss Meditec Inc.) and optic disc; fundus and RNFL photographs were obtained with a digital fundus cameras (Canon CF-60UD; Canon Inc.). All images were digitally recorded and analyzed by a single evaluator (S.W.H.). After the subjects were aligned properly, the eyes of the normal subjects who met the eligible criteria were scanned using the Cirrus HD OCT system (version 4.0.0.64; Carl Zeiss Meditec Inc.) after pharmacologic pupil dilation. Three individual 200 200 cube optic disc scans were obtained. Only scans that did not have movement/blink artifacts and had a signal strength Z7 were accepted. Unacceptable scans were discarded and a new subject was considered. Three quality scans were obtained for each eye, and among these the scan with the highest signal strength and least eye movement was selected to be included in the analyses. The RNFL thickness value at each of the 256 measurement points (0 to 255) was recorded. Pearson correlation coefficient between the right and left RNFL thickness values was calculated for each subject and was defined as the interocular RNFL thickness symmetry value. Of the symmetry measurements, we chose Pearson correlation coefficient because the Cirrus OCT RNFL thickness report provides an interocular symmetry value, which is the Pearson correlation coefficient multiplied by 100. However, this symmetry value needs to be corrected for ocular cyclotorsion. To do this, we drew a line between the geographic centers of the optic nerve head and the fovea. The point at which this reference line crossed the
e126 | www.glaucomajournal.com
scan circle became the new reference point (point 0). Then, we rearranged the sequence of RNFL thickness result to the new reference point. Because the Cirrus OCT RNFL thickness map does not contain the fovea, we drew the reference line on the fundus photograph and moved it to the RNFL thickness map. This step-by-step process was as follows (Fig. 2). Step 1: On the fundus photograph, we outlined the optic disc margin and calculated the geographic center of the optic nerve head using image-editing software (Photoshop CS5, ver. 12.0.1; Adobe Inc., San Jose, CA). We located the geographic centers of the optic nerve head (point c) and the fovea (point f) and drew a line between the 2 points. Step 2: We marked the temporal border of the superior temporal retinal vein (point a) and the temporal border of inferior temporal retinal vein (point b) where they crossed the disc margin. Step 3: We constructed a triangle, using points a, b, and f as corners, and then measured the angle a between sides a-b and a-f, the angle b between sides a-b and b-f, and the angle y between side a-f and the line c-f. Step 4: On the RNFL thickness deviation map, we marked the points where the temporal border of the superior temporal retinal vein and the temporal border of the inferior temporal retinal vein crossed the disc margin (points a* and b*, respectively). Then, we drew a line between points a* and b*. Step 5: At point a*, we used a protractor to measure angle a and drew a line to the temporal direction. At point b*, we measured angle b and drew a line to the same direction. We then marked the point f* where the 2 lines met. Step 6: at point f*, we measured angle y and drew a line in the disc direction (blue line). Finally, we marked the point where the scan circle and the line met. This point was set as the new reference point (point 0). Vessel anatomy was also examined and quantified in the following manner (Fig. 3). On the RNFL thickness r
2013 Lippincott Williams & Wilkins
J Glaucoma
Volume 23, Number 8, October/November 2014
Interocular RNFL Symmetry Value in Normal Adults
FIGURE 2. Fundus photograph and retinal nerve fiber layer (RNFL) thickness deviation map image showing reference lines. The blue line on the fundus photograph (A) runs between the geographic centers of the optic nerve head (point c) and the fovea (point f). The red triangle has corners at the fovea (point f) and the points where the temporal border of the superior temporal retinal vein (point a) and the temporal border of inferior temporal retinal vein (point b) cross the disc margin. Side a-f and the line c-f make the angle y between them. B, On the RNFL thickness deviation map, a similar triangle was constructed with 2 corners at the points where the temporal border of the superior temporal retinal vein (point a*) and the temporal border of inferior temporal retinal vein (point b*) cross the disc margin and side a*-b*. The other corner of the triangle was point f*. Angle y was measured and a line was drawn in the direction of the disc (blue line). Finally, point r was marked where the scan circle (red circle) and the line met.
deviation map, where the new reference line was drawn, we marked the points where the scan circle and the outer borders of the superior and inferior temporal retinal artery and vein met. Next, the middle points between the outer borders on the scan circle were marked. From the center of the scan circle, the angle between the middle point of the superior temporal retinal artery and vein and the new reference point was measured, as was the angle between the middle point of the inferior temporal retinal artery and vein and the new reference point. Linear regression analyses were performed to determine the correlations between corrected RNFL thickness symmetry values and interocular differences in axial length, refractive error spherical equivalent (SE), scan signal strength, and superior temporal and inferior temporal artery and vein locations. The analyses were performed with the symmetry value as the dependent variable. Drawings and measurements on images were carried out using an image-editing software (Photoshop CS5, ver. 12.0.1; Adobe Inc., San Jose, CA). Statistical analyses and graph plotting were performed using SPSS statistical software (version 13.0, SPSS Corp., IL, Chicago), and statistical significance was defined as a P < 0.05.
Mean original interocular RNFL thickness symmetry value was 0.8791 ± 0.0665, which increased to 0.9044 ± 0.0571 after correction for ocular cyclotorsion (P < 0.001; paired t test). On an average, the new reference point for ocular cyclotorsion correction (new point 0, the point where reference line met the OCT scan circle) was point
RESULTS Study subjects were mostly men [230 men (95.4%) and 11 women (4.6%)] and had a mean age of 21.4 ± 1.6 years (range, 19 to 28 y). All subjects were ethnically Korean and their ocular information is summarized in Table 1. On an average, the angle between the superior retinal artery and vein line and the reference was smaller in the right eye (76.26 degrees) than in the left eye (79.34 degrees; P < 0.001; paired t test). The OCT scan signal strength was also higher in the right eye than in the left eye (P = 0.028). No statistically significant interocular differences were observed for SE, axial length, average RNFL thickness, or the angle between the inferior retinal artery and vein line and the reference line. Table 1 shows the difference between the right eye and the left eye. r
2013 Lippincott Williams & Wilkins
FIGURE 3. Measuring the angles between the new reference point (to correct for ocular cyclotorsion) and the middle points of the superior and inferior retinal arteries and veins allowed us to correct for ocular cyclotorsion. The points where the scan circle and the outer borders of the superior and inferior temporal retinal artery and vein met (yellow lines and green lines, respectively). The middle point between the outer borders of superior temporal retinal artery and vein (yellow arrowhead), and the middle point between outer borders of inferior temporal retinal artery and vein (green arrowhead) are also marked. The angle a lies between the middle point of the superior temporal retinal artery and vein and the new reference point (blue arrowhead). The angle b lies between the middle point of the inferior temporal retinal artery and vein and the new reference point.
www.glaucomajournal.com |
e127
Volume 23, Number 8, October/November 2014
250.53 ± 2.168 (of 256 point of RNFL thickness) in the right eye and point 251.16 ± 3.153 in the left eye. Figure 4 shows symmetry value distributions before and after correction for ocular cyclotorsion. Table 2 shows the percentile distributions of the original and corrected interocular RNFL thickness symmetry values. Figure 5 shows the univariate linear regression analysis of the effect of various interocular differences on the corrected RNFL thickness symmetry value. The absolute interocular difference of the angle between the superior retinal artery and vein line, and the reference line was strongly and negatively correlated with the RNFL thickness symmetry value (P < 0.001, R2 = 0.208). However, absolute interocular differences in SE, axial length, and the angle between the retinal artery and vein and the reference lines were weakly but statistically negatively correlated with the corrected RNFL thickness symmetry value (P < 0.05). The difference between mean signal strength in the right and left eye scans was not significant (P = 0.10). Interocular SE differences were moderately correlated with interocular axial length differences (r = 0.484, P < 0.001). Interocular differences of SE and axial length were weakly correlated with interocular differences in the angle difference between the inferior retinal artery and vein line and the reference line (r = 0.184, P < 0.001 and r = 0.247, P < 0.001, respectively). Other interocular differences were not significantly correlated with each other.
0.64 ± 0.64 (0-2) 0.028 0.129 ± 0.901 ( 2 to + 2) 8.44 ± 1.109 (7-10) 8.56 ± 0.982 (7-10)
3.16 ± 2.59 (0-14.55) 0.098 0.436 ± 4.073 ( 14.11 to 14.55) 98.21 ± 8.87 (74-124.85) 98.64 ± 8.78 (72.62-124.53)
5.66 ± 5.35 (0-33.75) 0.353 0.467 ± 7.784 ( 33.75 to 22.50) 295.76 ± 10.45 (260.16-319.22) 296.24 ± 10.63 (55.94-320.63)
0.180 ± 0.212 (0-1.76) 5.60 ± 5.06 (0-28.13) 0.210 < 0.001 0.022 ± 0.277 ( 1.31 to + 1.76) 3.081 ± 6.900 ( 28.13 to 15.47) 24.850 ± 1.2092 (22.75-29) 79.34 ± 9.89 (43.59-108.28) 24.872 ± 1.2318 (21.56-28.77) 76.26 ± 9.67 (46.41-99.84)
e128 | www.glaucomajournal.com
Signed difference (right eye left eye). *Paired t test.
Absolute Difference
0.084
Right Eye
P* Signed Difference
0.0565 ± 0.703 ( 2.5 to + 3.38)
Left Eye
2.449 ± 2.3445 ( 10.125 to + 1.385) 2.505 ± 2.3522 ( 11 to + 4.125)
DISCUSSION
Spherical equivalent of refractive error (range) (D) Axial length (range) (mm) Angle between the superior retinal artery and vein line and reference line (range) (deg.) Angle between the inferior retinal artery and vein and reference line (range) (deg.) Average retinal nerve fiber layer thickness (range) (mm) Signal strength of scan
Ocular variables
TABLE 1. Summary of Ocular Information From Normal Subjects (n = 241)
J Glaucoma
0.4632 ± 0.5316 (0-3.375)
Jee et al
In the previous studies, the RNFL assessment was based on comparisons with normative thickness data, which showed relatively good sensitivity and specificity for glaucomatous damage detection.12–21 However, some patients have unique RNFL thickness profiles that largely deviate from the normative RNFL thickness profile. In these patients, comparing RNFL thickness with normative data does not provide useful information. Glaucomatous damage would decrease interocular RNFL thickness symmetry independent of a patient’s RNFL thickness profile configuration. Because little is known about interocular RNFL symmetry in healthy subjects, we examined this parameter to better understand how this information can be used to detect glaucomatous damage. The previous studies have investigated interocular RNFL thickness symmetry in normal adults.22–24 Mwanza et al22 used spectral-domain OCT and reported that an interocular average RNFL thickness difference of >9 mm may be indicative of glaucomatous damage. Budenz reported similar results by using time-domain OCT.23 Our study is somewhat different from these previous studies because we focused on comparing RNFL thickness symmetry between the right and left eyes. Further, our study used a new technique to remove ocular cyclotorsional effects from interocular RNFL symmetry data. There are other ways to statistically evaluate symmetry between 2 variables. Ghadiali et al24 used the coefficient of determination (R2) to successfully evaluate and compare interocular RNFL thickness symmetry profiles of multiple subjects. The coefficient of determination reflects the absolute thickness difference between the right and left eye. Whereas Pearson correlation coefficient reflects the correlation between the right and left eye. We chose to use the Pearson correlation coefficient for symmetry evaluation because glaucoma practitioners are familiar with OCT r
2013 Lippincott Williams & Wilkins
J Glaucoma
Volume 23, Number 8, October/November 2014
Interocular RNFL Symmetry Value in Normal Adults
FIGURE 4. Histogram showing the distribution of interocular retinal nerve fiber layer (RNFL) thickness symmetry values. A, Original RNFL thickness data. B, Data that have been corrected for ocular cyclotorsion.
symmetry values and because the Pearson correlation coefficient is more convenient in statistical analyses. In addition, 2 coefficients of determination result from 1 subject (right eye, left eye) and one of these 2 values, alone, does not fully reflect interocular symmetry. We did not want to choose between the 2 numbers and we were unsure how to resolve the complication of having 2 values in statistical analyses, especially in analyzing the ocular factors affecting the symmetry value. Our data analyses removed ocular cyclotorsional effects on RNFL thickness profiles by creating and using a reference line that was drawn between the foveal and disc centers. The line largely corresponded to the horizontal raphe of the retina. Other geographic markers, such as retinal blood vessels, can also be used as a reference to correct for ocular cyclotorsion; however, blood vessel patterns are highly variable and can be affected by disease (eg, diabetes, glaucoma). The horizontal raphe is a suitable reference line because it has a relatively straight configuration and is easy to identify. Moreover, it is both the anatomic and functional divisor of the superior and inferior RNFL, and its location is minimally affected by disease processes. Even when subjects were properly aligned in the OCT machine before scanning, correcting for ocular cyclotorsion significantly increased the interocular RNFL thickness symmetry value. As shown in Figure 1, even a 6-pixel point rotation of the right eye can translate into a marked 19% difference in interocular symmetry. This finding suggests that scanning conditions can considerably affect OCT interocular RNFL thickness symmetry value results, and without correction for ocular cyclotorsion healthy normal individuals may produce abnormal RNFL symmetry value. The
symmetry value percentile distribution also changed. The first and fifth percentile of the original symmetry value was 0.7242 and 0.7697, respectively. After correcting for cyclotorsion, the first and fifth percentile of the symmetry value increased to 0.7714 and 0.8105, respectively. This finding suggests that the right and left eye RNFL thickness profiles are more highly correlated and that a symmetry value
