Asymmetry parameters comparing retinal thickness in fellow eyes or hemispheres in the same subject have clinical advantages compared with raw measurements. In contrast to raw measurements, asymmetry parameters are theoretically not biased by factors that affect the eyes symmetrically such as age, race, and sex.
Posterior pole retinal thickness asymmetry analysis was proposed to be a useful tool in detecting early glaucomatous changes
4,19–24; however, to date there is sparse normative data available regarding physiological macular asymmetry in healthy subjects.
In our study, we examined physiological retinal thickness asymmetries in healthy adults using the entire 64 cell PPAA-protocol of a SPECTRALIS SD-OCT device.
Our data show that statistically significant physiological asymmetries in intra- and interocular retinal thickness exist. In our dataset, the grand mean interocular retinal thickness asymmetry was 5.6 μm (95% CI: 4.6–6.5) and the grand mean intraocular thickness asymmetry was 8.3 μm (95% CI: 6.8–9.9) in the right eye and 8.4 μm (95% CI: 6.7–10.0) in the left eye.
Sullivan-Mee et al.
24 investigated retinal asymmetry in 50 subjects with early primary open-angle glaucoma (POAG) and 50 control subjects using the PPAA-protocol of the SPECTRALIS SD-OCT device. This study enrolled subjects aged over 40 years, and recorded the grand mean, superior and inferior macular thickness asymmetries, but not the difference in every single cell of the PPAA-protocol. The study found a grand mean interocular retinal thickness asymmetry of 2 μm (0.0–5.3) and a grand mean intraocular hemisphere asymmetry of 2.6 μm (0.0–7.0) in the control group. In the PAOG group these values were 10 (0.0–27.0) and 8.0 μm (0.0–23.0), respectively. The fact that we in our study found a slightly higher inter- and intraocular retinal thickness asymmetry in healthy individuals may arise from either the smaller number of patients in the above study or from the differences in inclusion criteria.
Two other studies have previously investigated physiological interocular retinal thickness asymmetries using commercial OCT software (Cirrus OCT with Macular Cube 200 × 200 protocol; Carl Zeiss Meditec, Dublin, CA, USA).
32,33 Altemir at al.
32 found 23.2 μm to be the interocular tolerance limit for average macular thickness asymmetry in 357 healthy children (mean age: 9 ± 1.7 years), and Dalgliesh et al.
33 found 8 μm to be the interocular tolerance limit of average macular thickness asymmetry in 1500 adolescent subjects (mean age: 17.3 ± 0.51 years). A study measuring the central retinal thickness values using six different OCT instruments found similar results with the Cirrus HD-OCT and the SPECTRALIS OCT. This may arise from the similar image processing algorithms of the two OCT systems, both including the RPE in the retinal segmentation.
2 Optical coherence tomography software differences need to be addressed when selecting cutoff values for asymmetry analysis to detect early glaucomatous changes.
20,23
To evaluate local differences in asymmetry, we investigated asymmetry across the entire posterior pole thickness map. The results indicated that the intra- as well as the interocular retinal thickness asymmetry increases toward the periphery of the macula, especially toward the nasal corners where the 8 × 8 grid in the posterior pole thickness map overlaps the upper and lower temporal vascular arches. Although ICC is relatively high for all cells, ICC shows the same trend, with higher within-subject variation toward the periphery. This marked asymmetry in the peripheral zones is supported by the study of Um et al.,
22 where an increased interocular asymmetry in zones representing the inferior and superior periphery of the grid were observed in healthy controls, glaucoma suspects, and glaucoma patients using a modified PPAA protocol.
Furthermore, a study by Yamashita et al.
34 found the intraocular symmetry in retinal thickness to be lowest in the peripheral nasal areas of macula. They found that the difference between the supra- and infratemporal artery and vein angle correlated significantly with the higher retinal thickness asymmetry in the peripheral nasal areas.
The segmentation algorithm in the SPECTRALIS SD-OCT device includes the retinal vessels in the retinal thickness measurements.
25,26 Therefore, it is important to take the effect of the retinal vessels on the retinal thickness into account. A new segmentation algorithm that subtracts the thickness of the vessels from the retinal thickness measurement is necessary, if accurate retinal thickness and retinal thickness asymmetry measurements are needed.
The central 20° area assessed by the PPAA corresponds closely with the 24-2 visual field test. A recent report
21 showed a pointwise relationship between visual field sensitivity (VFS) and macular thickness determined by the SPECTRALIS SD-OCT device using a modified 16-cell posterior pole thickness grid. The study revealed that VFS and mean retinal thickness showed the strongest correlation centrally, and weakening correlation toward the peripheral cells of the grid. This finding supports that the PPAA of peripheral cells—that are influenced by the vascular structures—may have a lower value in glaucoma diagnostics.
The posterior pole asymmetry analysis evaluates retinal asymmetry by measuring all layers of the macula, thus symmetry/asymmetry is biased by other layers than the RNFL and the RGC layer. A modified PPAA software assessing asymmetry only regarding the RNFL-RGC complex could help us to take full advantage of this technology in the diagnostic of glaucoma.
We found a slight statistically significant average age and sex effect on retinal thickness asymmetry of 0.04 μm/year (0.02–0.06 μm) and 0.54 μm (0.19–0.88 μm), respectively, for men compared with women. However, in evaluating these results, it must be taken into account, that our study is slanted toward younger women. The magnitude of the age and sex effect on asymmetry should be considered with caution, since the residual plots of the linear model showed slight deviation from the Gaussian assumption. However, the findings of significant age and sex effects were supported by the more complicated and less interpretable generalized linear model based on the gamma distribution.
A handful of studies have investigated the relation between age and sex and retinal thickness using the SD-OCT.
35–39 Two of these publications found greater foveal thickness among men
37,38 and a third one showed a significant increase in central foveal subfield thickness with age.
39 To the best of our knowledge, there has been no other investigation that assessed the possible effect of age and sex on posterior pole asymmetry to date.
Our study has certain limitations. We enrolled only healthy adults aged between 18 and 45 years, to eliminate any structural change of the retina due to any age-related macular pathology, primarily AMD. Additionally, subtle changes in the refractive media might affect data acquisition and automated segmentation with age.
40 Most patients in ophthalmology, however, are aged older than 45 years and are often troubled by more than one ocular disease.
Our study is also limited by our choice to only enroll individuals with a refractive error between −1.5 and + 1.5D (SE) in order to eliminate the effect of anisometropia or high ammetropia. Therefore, our normal values only reflect retinal thickness asymmetry in the near emmetropic and the isometropic. A recent study by Kim et al.
41 investigating healthy Korean eyes (RE: −3.97 ± 2.84 D) showed that retinal thickness correlated positively with RE with regional variations within the 8 × 8 posterior pole grid.
41
Finally, the PPAA analysis does not take the anatomical symmetric placement of the arcuate nerve fibers around the temporal raphe into account. Instead the grid is placed symmetrically to the fovea–disc axis, a symmetry line that correlates better with the anatomical symmetry of the nasal fibers, than the symmetry of the temporal fibers.
A recent study by Huang et al.
42 found that the geometry of the temporal raphe was neither strictly horizontal to the fovea nor to the optic disc. Through the investigation of 11 healthy subjects using adaptive optics scanning laser ophthalmoscope, they found the angle between the temporal raphe and the fovea disc axis to be 170.3 ± 3.6° and that the angle between the temporal raphe and a horizontal line through fovea varied from −9 to 6°, with a mean of −1.67 ± 4.8°.
As a consequence hereof, the PPAA causes incorrect pairing of regions in the hemisphere asymmetry analysis. For the time being the manufacturer is analyzing the problem. However, at present it is not possible to place the asymmetry grid around the anatomical symmetry line.
In conclusion, statistically significant physiological asymmetries in inter- and intraocular central retinal thickness were found in healthy Caucasians when using the PPAA protocol of a SPECTRALIS SD-OCT device. Normal variation, as presented in this article, must be considered when early signs of glaucoma or other pathologies are evaluated based on the retinal thickness asymmetry.
It is important to take into consideration that the PPAA protocol includes the retinal vessels in the thickness measurements, and that the protocol alignment is not completely in accordance with the anatomical symmetry of the retinal nerve fibers. These limitations of the PPAA-protocol lead to increased variability of the asymmetry measurements across the asymmetry grid.