To the best of our knowledge, this is the first longitudinal prospective study of the VF characteristics of NMOSD-ON during acute attack with a follow-up period at least 6 months after onset. The VF defect of ON, pattern as well as severity, evolves during the process of acute attack until the inflammation finally stables. The results of VF might be influenced by the course, treatment regimen, experience of VF readers et al., among which course was the main factor. This study clarified the VF defect pattern, severity and recovery trend of two types of ON with different pathogenesis mechanism (astrocyte disease versus primary demyelination).
Pattern of VF defect
VF not only reflects the severity of ON damage, but also reflects the localization of optic nerve damage by pattern. Most previous studies of VF defect pattern in acute phase of IDON reported that central scotoma was the dominant pattern at onset [12, 19, 20], but diffuse or nerve fiber bundle abnormality dominated 6 months after onset [10, 11], particularly when tested by Goldmann perimeter. Our study by static perimeter with 30–2 test program showed that diffuse abnormality at onset of VF defect in IDON group gradually evolved to nerve fiber bundle abnormality (18.9%), neurologic defect (13.9%) and central scotoma (8.9%) at ≥ 6 m. So, it was nerve fiber bundle abnormality rather than central scotoma was the dominant defect in the IDON group at ≥ 6 m. The difference might be due to the different devices that were used. Also, the pattern difference might also be influenced by the course of ON attack. A central scotoma mainly caused by the involvement of papillomacular bundles, may be covered by the diffuse involvement at onset. Therefore, if the VF of the enrolled patients were examined in early stage (within 8 days), the pattern of VF defect was diffuse abnormality. While the VF examined in later stage of ON (14–28 days or more), the pattern may be central scotoma.
Study about the pattern of VF in the acute phase of NMOSD-ON was extremely rare. It was reported that 76% of the 51 NMO-ON attacks had central scotoma of VF defect by Goldmann manual perimeter [12]. However, the exact course at which the VF tested was not mentioned. Our study by static perimeter showed that although diffuse pattern dominated most of the acute phase in NMOSD-ON group, the most common pattern gradually evolved to nerve fiber bundle abnormality (23.9%) at ≥ 6 m, rather than central scotoma. The percentage of neurologic abnormality of VF defect (quadrant, altitudinal, hemianopia and three quadrants pattern) in NMOSD-ON was higher than that of IDON. At 1 m, the altitudinal hemianopia in NMOSD-ON was found to be significantly higher than that of IDON. This finding was consistent with Nakajima et al.’s study [12], showing whenever visual acuity permitted, NMO patients showed much higher neurologic abnormality of VF defect, 10% for altitudinal, 6% for quadrant, 4% for three quadrants, 2% for hemianopia, and 2% for bitemporal hemianopia, while these patterns were extremely rare in MS-ON. However, in ONTT [4] study which contained mostly IDON, 14.9% of eyes showed altitudinal pattern at baseline. Therefore, the altitudinal pattern might not be characteristic of NMOSD-ON. Besides differences in neurologic abnormality pattern, NMOSD-ON also showed much less central scotoma of VF defect than that in IDON (15.8% in IDON group versus 0% in NMOSD-ON group), as early as at 1 m. This might indicate that the probability of papillomacular bundles involvement occurring alone in IDON, and its recovery might be earlier than other patterns defect in IDON. As to the mechanism of neurologic abnormality of VF defect, since NMOSD-ON is subject to damage the posterior part of orbital segment of optic nerve, and the nerve fiber bundles mainly run on the lateral side of it. So, whether this part is particularly prone to be attacked by AQP4-Ab due to the weakness of blood-optic nerve barrier remains to be investigated. As seen from the analyses above, the pattern of VF defect is helpful to diagnose and distinguish optic neuritis clinically.
Recovery pattern of VF defect
This pattern of recovery might indicate which part of the optic nerve tends to recover first: from peripheral to central, vice versa, or quadrantal, and might indicate the pathological process involved during recovery. Our results showed that both types of ON mainly recovered quadrant-wise, without a trend of clockwise or counter-clockwise. But due to the small sample size, further investigation was needed to make a definite conclusion and the mechanism remained to be investigated.
Severity and prognosis of VF
Numerous studies [10, 11, 21] showed that the visual function prognosis of NMOSD-ON was worse than that of IDON. The final MD of NMOSD-ON varied between -4.76 ~ -10.8 [10, 11, 21], while that of IDON varied between -4.6 ~ 7.9 [10, 11]. In this study, the severity of VF damage in NMOSD-ON was significantly more severe than that of IDON at all follow-ups, and there were significantly higher percentage of patients of NMOSD-ON (19.6%) still failed VF exam at ≥ 6 m supported that NMOSD-ON had a worse prognosis. However, there was no consensus as to the time point when the process of recovery terminated in ON (the end of acute phase), especially in NMOSD-ON. As to IDON, it was usually defined as 6 m ~ 9 m after onset [10, 11], while in NMOSD-ON, this time point varied in literature from 3 m ~ 8 m [10, 11, 21]. In this study, by detailed and regular longitudinal observation of BCVA, VF it was found that the recovery process of NMOSD-ON might terminated as early as 1 m after onset. The short period of recovery in NMOSD suggests that early treatment may be most effective.
The return of visual field function after acute optic neuritis was independent of the depth of the original defect (diffuse or localized VF defects), except for patients with the most severe defects [22]. This study also demonstrated that there was no correlation between severity of visual damage (BCVA, MD as well as PSD) at onset and that at the end of acute phase. However, a positive correlation detected between MD at 1 m and final MD was found in NMOSD-ON. A similar correlation was found for BCVA in NMOSD-ON. This was further support that the visual function of NMOSD-ON recovered as early as 1 m after onset and the visual function at this time point might be an indicator of the prognosis in NMOSD-ON. However, a correlation with final visual function of IDON was only found later (at 3 m), indicating a slower process in BCVA recovery.
PSD of VF
There was little research about the PSD of VF in ON, which is an index for VF irregularity. Merle [10] found the PSD was significantly higher in MS than that in NMO among the first episode, while no significant difference was found in relapsed ON. This was contradictory to our study, which found PSD higher in NMOSD-ON than that in IDON at each follow-up. The difference in PSD of these two studies may be due to a difference in baseline severity of the disease. But the statistically significant difference of PSD was not found in the first episode ON patients. In this study, PSD reduced gradually during follow-ups in both groups. The tendency of PSD in the two groups showed that the irregularity of local VF was getting smaller with the recovery of ON. Throughout the course of the disease, the irregularity of local VF in NMOSD-ON patients was more obvious than that in IDON, suggesting the fact that NMOSD-ON may lead to more severe VF damage. A positive correlation was found between 1 m and final PSD, as well as between 3 m and ≥ 6 m. This might indicate the visual function at 1 m after onset is an indicator of the final prognosis in NMOSD-ON and IDON as the change of MD above.
BCVA recovery
The BCVA recovery course of NMOSD-ON and IDON had seldom been reported before. Hiroki Masuda [9] reported that BCVA recovery from ON was poorer in NMOSD than that in MS and was negatively affected by pervious ON attack, but no longitudinal VF study was found. In this study during the follow-ups, there was an obvious different trend of BCVA and VF recovery between the two groups. In IDON, the recovery manifested a gradual and smooth trend. However, in NMOSD, the recovery manifested as a prompt recovery at 1 m then leveled off. And a positive correlation was found between at 1 m and final BCVA in NMOSD-ON group only. While a positive correlation was found between 3 m and final BCVA in IDON. This might indicate that the visual functions of 1 m in NMOSD-ON and 3 m in IDON were related to its prognosis.
Visual function in first episode ON
Analyzing the VF defect in the first episode ON attack could exclude the influence from previous attacks. In this study, VF and BCVA damage in the first episode ON sub-group were more severe in NMOSD-ON than that of IDON, just as that in the mix group. Although only BCVA showed significant difference, it might be due to the sample size being not large enough. Our results were consistent with the other two studies. In Merle et al.’s study [10], for the eyes of first episode ON, ETDRS visual acuity, contrasts vision (Sloan chart 1.25%), MD in the NMO group were worse than that in MS-ON. However, no difference in visual acuity (Snellen) and retinal nerve fiber layer (RNFL)was found between them. In Fernandes et al.’s study [11], the number of eyes with normal VF (p < 0.001) and average VF MD (better than -3.0) after a single episode of ON (p < 0.001) was significantly less in NMO-ON than that in MS-ON. These results further supported that NMOSD-ON has more severe visual damage and poorer recovery than that in IDON. As to the pattern of VF in first episode ON, no significant difference could be found between the sub-groups in this study. However, central scotoma was only seen in IDON. This was consistent with previous studies indicating that central scotoma was the characteristic VF defect pattern in MS [12, 19]. The reason why papillomacular bundles were more vulnerable to damage in IDON was still unclear.
Limitations
There were several limitations to our study. First, single center design might cause a certain degree of patient selection bias. Second, the sample size was not large enough. In our study, many VF subcategories had very small sample size, which disables the statistical comparison between the two groups. A larger sample size is needed to see if there were real differences. We believed that if the amount of patients with first episode was larger, more differences might be revealed between the two groups. Third, more indexes of visual function, e.g., color vision and contrast sensitivity, as well as structure index, e.g., RNFL and ganglion cell layer (GCL) analyses, should be observed. The correlation between these indexes should be analyzed to further understand the clinical characteristics of NMOSD-ON. Last, the data of another common type of ON, MOGAD related ON, was not included in this study. More prospective, multicenter, longitudinal cohort clinical study with larger sample size of patients of different types of ON has already been under investigation in our group.