shows the mean RNFL thickness [95% confidence interval (CI)] for the centered and displaced scans. No significant difference in the average RNFL thickness was found between centered scans (106.4 μm) and the average of the displaced scans combined (107.8 μm) using paired t test (P = 0.13). When the average RNFL thickness of the centered scans was compared with the average RNFL thickness of the displaced scans individually using ANOVA with post-hoc analysis, no difference was found between the average RNFL thickness of the nasally (105.2 μm), superiorly (106.2 μm), or inferiorly (104.1 μm) displaced scans and the centered scans. However, a significant difference (ANOVA with Dunnett's test: F = 8.82, P<0.0001) was found between temporally displaced scans (115.8 μm) and centered scans (106.4 μm).
Average RNFL Thickness Measurements for Centered and Displaced Scans
The increase in average RNFL thickness with temporal displacement was also confirmed in a subset of 2 eyes (), where progressive temporal displacements along the x axis resulted in consistent increases in average RNFL thickness. In contrast, no change in RNFL thickness was observed when the scan was displaced along the y axis (vertical) ().
The CV and ICC of the centered scans for assessing the intratest variability of the average RNFL thickness parameter were 1.75% and 0.95 (lower 95% CI = 0.92), respectively, indicating that the measurements were reproducible.
shows the mean (95% CI) RNFL thickness in microns for the superior, inferior, nasal, and temporal sectors for the centered and displaced scans. In general, significant differences in some sector RNFL measurements were found with each type of scan displacement with the nasally displaced scan showing statistically significant differences in all sectors. Predictably, superior and inferior RNFL thickness values changed most when the scan was displaced along the superior and inferior axis, whereas temporal and nasal RNFL thickness values changed most when the scan was displaced along the horizontal axis (, ). As expected, when the scan circle moved closer to the disc margin, the RNFL thickness increased. For example, the superior RNFL thickness increased from 133.7 μm when centered to 152.9 μm when displaced inferiorly.
Sectors RNFL Thickness Measurements for Centered and Displaced Scans
Double hump profile pattern for centered scans and scans displaced superiorly and inferiorly. *Sectors significantly changed compared with centered scans.
Double hump profile pattern for centered scans and scans displaced nasally and temporally. *Sectors significantly changed compared with centered scans.
To better evaluate the change in average RNFL thickness with temporal displacement, the superior and inferior sectors of the temporally and nasally displaced scans were compared with centered scans using ANOVA and Dunnett's test. For the superior sector, although no significant difference was found for the temporally displaced scan (P = 0.93), the nasally displaced RNFL thickness was significantly reduced (123.4 μm) compared with centered scans (133.7 μm, F = 6.13, P<0.05). Similar results were found for the nasally displaced RNFL with regard to the inferior sector (118.3 μm) showing thinner RNFL than the centered inferior sector scan (134.8 μm, F = 14.8, P<0.05). Conversely, the temporally displaced scan showed an average increase in inferior thickness of 6.1% compared with centered scans (142.9 and 134.8 μm, respectively).
The hypothesis was tested that the RNFL thickness increases in the sector closer to the disc and decreases in the opposite one so that the mean RNFL thickness of the 2 sectors may not change. The nasal and temporal sectors were examined on the horizontal axis by comparing the mean RNFL from the two sectors (nasal+temporal/2) in the centered scans and in the temporally and nasally displaced ones. After applying ANOVA and Dunnett's test, results showed a significant increase (F = 14.95, P<0.0001) in the mean for the displaced scans with the temporally displaced one showing the greater increase in thickness (+19.5% vs. +13.9%). In contrast, when the same analysis was carried out on the vertical axis by comparing the mean RNFL from the superior and inferior sectors (superior+inferior/2) in the centered scans and in the superiorly and inferiorly displaced ones, no statistically significant differences were found (134.2 vs. 131.4 and 130.6 μm respectively, F = 1.87, P = 0.16).
The average optic disc size measured using the Heidelberg Retina Tomograph was 2.15 ± 0.54 mm2. Differences in the measurements between centered and displayed scans remained significant (P<0.0001) after controlling for ONH size in multivariate analyses (data not shown).
A separate analysis was also conducted after excluding the 2 eyes with parapapillary atrophy; similar results to the ones described above were obtained for average and sectoral RNFL thickness in centered and displaced scans. Specifically, in the 2 eyes with parapapillary atrophy, a similar trend to the other eyes was observed with the average RNFL thickness increasing from 95.74 μm in OD and 92.75 μm in OS for the centered scans to 104.19 μm in OD and 101.87 μm in OS for the temporally displaced scans.
illustrates hypothetical diagnostic errors that could result from improper scan alignment. On the basis of these suggestions, when a sectoral RNFL thickness decrease is detected over time, the average RNFL thickness should be checked along with the other sectors measurements to determine the likelihood that such a change is real. If the average RNFL thickness is increased, a temporal shift of the scan at the follow-up visit compared with baseline might have occurred. In the presence of a coexisting decrease in the average RNFL thickness, a true change in the sectoral RNFL thickness may be likely if repeatable, although a false change may also be indicated if the baseline scan was displaced temporally. In contrast, when the average RNFL thickness is unchanged at follow-up and increased thickness is detected in the opposite sector, a true sectoral change is very unlikely as shifts along the vertical axis could be exclusively responsible for this finding.
FIGURE 5 Flowchart of the hypothetical spectrum of events that could imply false RNFL estimates owing to improper scan alignment. Careful overall assessment of sectoral and average OCT RNFL parameters can help identify scan alignment issues. Note: results are (more ...)
provides an example of a sample RNFL scan with scan circle displacement occurring along the vertical axis and superimposed RNFL double hump profile patterns resulting from centered and displaced scans. Apparent sectoral thinning is visible in the displaced scans although the opposite sectoral RNFL thickness is markedly increased. Similarly, for the same eye, shows the double hump profile patterns from scans displaced along the horizontal axis. The average RNFL thickness is increased in the temporally displaced scan.
FIGURE 6 A, Sample RNFL scan with displacement occurring along the vertical axis. The superimposed RNFL double hump profile patterns show apparent sectoral thinning in the displaced scans (bottom 2 scans) compared with the centered scan (top scan) whereas the (more ...)