Most cross-sectional studies to date addressing the atrophic stage (> 3 months) of NAION have reported a consistent conclusion of significant reduction of peripapillary radial capillaries (RPC) and9,12,15 capillaries in macular retina using OCT-A9,16. Also, these OCT-A derived capillary density changes in RPC were spatially correlated with the severity of visual field defect and thinning of peripapillary RNFL9,15. Thus, it has been well accepted that there is an obligatory loss of capillary density resulting from the loss of corresponding layers of neural tissue in the pathological progression of NAION9,12,15,16. A few studies concerning OCT-A in the acute stage of NAION show the defects in peripapillary microcirculation13,17,18. Song et al.19 and Rebolleda et,al.13 have reported that there were no significant differences of reduced peripapillary vessel density in acute and chronic stages. However, they assessed the measurements with smaller scanning area13, and compared the acute and chronic stages with the division of 21 days using two separate patient pools19, which may generally correspond to the comparison between 1–2 weeks and 1–2 months follow-up intervals in our cohort that correspond to the stage of resolution. In this cohort, in addition to the significant reduction of wRPC at enrollment compared to unaffected eyes and healthy controls, we also fund that wRPC in NAION-affected eyes decreased significantly from the acute stage to the stages of resolution and atrophy, along with a more drastic reduction of sRPC, suggesting a sectoral reduction of RPC as the ODE subsided.
The most plausible explanation for an apparent decrease in RPC density during the acute phase could be the mechanical compression and impedance of the flow in RPC from the edema of the optic nerve 13,17, and the local attenuated signal strength caused by the edema of the tissue and penetration of coherent light may also contribute a decrease in capillary density 17. In addition, it is possible that some patients are already beginning to show loss of axons in the setting of edema and axoplasmic flow of the remaining axons and that the reduced capillary density result from axon loss which would be masked by the edema1,4.
Compared to peripapillary RNFL thinning, the pattern of GCL/GCC thinning may reflect earlier structural changes caused by ischaemia of the optic nerve in NAION 20,21. Thinning of inner layers in macula could be detected within one week after the ischaemic event20,21 reflecting the early damage of RGC and the axons3,20−23, and peak at 3–6 months and last for 6–12 months3,22,23. Our results showed that wGCC and the corresponding wSVD in affected eyes were gradually decreased at 1–2 weeks follow-ups, and aggravated at 1–2 and 3–6 months follow-ups along with worsening superior hemifield defect. At the stage of resolution, we found a significant correlation between the reduction of wSVD and wGCC, which has not been detected in previous reports. Together with the significant association between wGCC thinning and wRNFL, our results provide further data to support the hypothesis that thinning of GCL in macula was secondary to the degeneration of peripapillary axons that are prone to suffer from hypoxia and ischmia20,23,24. There could be a vicious cycle about the neural-vascular interactions in the pathological progression of NAION1,24.
Although some patients showed mild reduction in wGCC and wSVD at the enrollment (average duration of onset was 5.67 ± 3.06 days), the overall difference among three groups was not significant enough to draw a confirmed conclusion. While, at the acute stage of NAION, the edema of RNFL in macula would mask the ability to detect significant thinning of the underlying layers of GCL and IPL23 and this may also explain that the correlation between wSVD and wGCC particularly occurred at the stage of resolution. However, the OCT-A derived capillary density in SVD was expected to show decrease due to both the edema of RNFL and/or early thinning of the underlying layers23,25, it might be that, as Augstburger et, al.9 proposed, the decrease of SVD in the 6 × 6 mm scan of OCT-A predominates near the large retinal vessels of the arcades and less at the parafoveal areas where the capillary network converges and forms an anastomotic ring9,24,25. And the axonal swelling or RNFL edema resulting from the blocked axoplasm and leading to impedance in flow of RPC and retinal vessels might be severer in the site closer to the optic disc. While, Fard et al. 26 have reported an early reduction of SVD and DVD in 6 × 6 mm scans but not of GCC at the acute stage within 2 weeks of onset. It might be that we missed the changes outside 3 × 3 mm scanning area in the early stage. And it is possible that oral corticosteroid therapy accelerates the regression of ODE14 and therefore makes the early reduction of SVD insignificant. Besides, an early reduction of wDVD has been detected in our study. It might be explained that the deep capillary vortexes27 might be earlier or more frequently to compensate for the lower blood flow and the changes of hypoxia and ischemia in macula as previously reported in patients with hypertension and diabetes mellitus28,29.
Contrary to previous results of decreased DVD in chronic stage with macular 6 × 6 mm OCT-A imaging9,16, our results with 3 × 3 mm scan show that the wDVD did not decrease significantly with time procession. This could be explained by the enhanced visibility of deep retinal complexes resulted from severer thinning of inner layers. Although grading accuracy for 3 × 3 mm scans was significantly higher than for 6 × 6 mm scans which has limited vascular depth discrimination and is prone to induce overestimation of the measurements and mix DVD with the SVD changes 25,30, the limited 3 × 3 mm scanning area of OCT-A could miss those changes outside the confined area and therefore underestimate the results.
There are several limitations in our study, including its retrospective nature, the small number of subjects, and some loss to follow-up at 3–6 months. Here, we focused on flow alterations with time, yet visual field loss of threshold was not analyzed in detail as previous studies showed a corresponding loss of structure and capillary density in the location of the visual field loss9,13,26. Additionally, vasculature in prelaminar and laminar slabs was not evaluated due to the current limits of the technology, which may be more related to the ischemia mechanism of PCA.