OCT-defined optic atrophy thresholds are associated with blunted optic nerve swelling in recurrent optic neuritis

Abstract

Background:

Acute optic neuritis (ON) can be associated with optic disc edema which can be quantified by optical coherence tomography (OCT) peripapillary retinal nerve fiber layer (RNFL) measurements. In patients with recurrent visual symptoms, objective measures can confirm a new clinical attack. Optic atrophy may preclude RNFL swelling during recurrent ON, but this has not been systematically evaluated.

Methods:

Patients with recurrent ipsilateral ON with OCT within 2 weeks of symptoms were retrospectively identified from 8 international centers.

Results:

We identified 112 patients (MOGAD, 45[40%]; MS, 36[32%], NMOSD, 20[18%], idiopathic ON, 11[10%]) with 137 episodes of recurrent ON (median age, 39 years [range 12-78]; female sex, 81 [72%]). Of patients with macular ganglion cell-inner plexiform layer (GCIPL) measures during acute recurrent ON, those with preserved GCIPL (≥65μm, n=36) showed greater RNFL thickness (mean 108μm [65–368], p<0.0001) than those with thin GCIPL (<65μm, n=30; mean 65μm [50–84]). Higher RNFL thickness measurements, which can be consistent with RNFL swelling, (≥15μm above GCIPL) occurred in 21/36 (58%) patient with GCIPL ≥65μm vs 5/30 (17%) patients with GCIPL <65μm (p<0.001).

Conclusions:

Patients with severe optic atrophy may have limited optic nerve swelling during acute ON. Ganglion cell layer thickness can serve as a surrogate marker for optic atrophy during early acute ON. We identified quantitative thresholds (GCIPL [<65μm] or prior RNFL [<70μm]) at which optic nerve swelling is unlikely to occur during recurrent acute ON. This is important for objective assessment of ON in clinical practice and clinical trials.

Introduction

Optic neuritis (ON) is a common manifestation of central nervous system demyelinating diseases, including multiple sclerosis (MS), myelin oligodendrocyte glycoprotein antibody-associated disease (MOGAD), and aquaporin-4 antibody positive neuromyelitis optica spectrum disorder (NMOSD).^1-3 Acute ON may cause optic disc swelling and peripapillary retinal nerve fiber layer (RNFL) thickening. Optic disc swelling occurs in approximately one third of patients with ON and is most often observed in patients with MOGAD-related ON.^4-9 RNFL and macular ganglion cell-inner plexiform layer (GCIPL) thickness can be reliable quantified using optical coherence tomography (OCT) which is informative for the objective assessment of optic neuritis and demyelinating disease.^10-15 OCT may also detect subtle optic nerve swelling as measured by RNFL thickening that may not be observed clinically. ^10,16-19 These clinical and imaging findings are important as they may guide clinical evaluation and management. Acute RNFL thickening and subsequent thinning on OCT are objective components of the diagnostic criteria for ON while disc edema on examination should lead to consideration of autoantibody testing for MOGAD as well as potential consideration of infectious etiologies^1,17,20-22 Substantial optic nerve thinning and atrophy, however, may occur after severe or multiple episodes of ON.^23-29 Accumulation of RNFL thinning may preclude swelling in subsequent acute ON, and thus hinder application of the ON diagnostic criteria and potentially impact treatment decisions, but this has not been systematically assessed or quantified with OCT. The GCIPL does not swell during acute ON but also undergoes thinning after acute ON.^10,19,23,30 This suggests GCIPL thickness during acute ON may serve as a proxy for prior optic nerve atrophy and thinning. This study determined if eyes with preexisting RNFL and/or GCIP^{16, 17}L thinning demonstrated RNFL swelling during recurrent acute ON and if there is a cutoff beyond which an atrophic optic nerve is less likely to swell.

Materials and Methods

Study population and patient consent.

This study was approved by Mayo Clinic Institutional Review Board and site-specific relevant bodies. All patients consented to medical record review for research.

Patients with MOGAD, MS, and NMOSD who fulfilled respective diagnostic criteria were retrospectively identified.^1-3 Patients with recurrent acute ipsilateral ON (≥1 prior attack) per ON diagnostic criteria as determined by site investigators²⁰ and OCT imaging ≤2 weeks of visual symptom onset were included. Baseline OCT measures completed before the acute ON of interest but ≥6 months after prior episodes of ON (to allow thinning to plateau) were included if available.²³ Patients with high refractive error (+/−6 diopters) or comorbid ocular conditions were excluded.³¹ Patients received treatment as part of site-specific standard of care and the type, timing, and duration of acute and chronic treatment did not prevent inclusion in this study. MRI findings were incorporated as part of site-specific standard of care as needed for the diagnosis of optic neuritis. Given the anatomic connection of ganglion cells and their axons, the values of GCIPL and RNFL are closely related. A cohort of patients with demyelinating ON³² and a cohort of healthy volunteers³³ from prior studies were used to determine the relationship between RNFL and GCIPL. This analysis demonstrated that an RNFL ≥15μm thicker than GCIPL during acute ON suggests a potential for RNFL swelling and that an RNFL ≥25μm thicker than GCIPL during acute ON is highly suggestive of RNFL swelling (Supplemental Methods).

OCT Imaging.

Peripapillary RNFL and GCIPL thickness were obtained using spectral domain OCT including Cirrus HD-OCT (Zeiss; Dublin, CA, n = 86) or Spectralis OCT (Heidelberg Engineering, Germany, n = 26) as per center protocol. Average RNFL and GCIPL thicknesses measurements were devised by device segmentation software for the Cirrus machine and RNFL thickness was calculated by the Spectralis machine. For the Cirrus, scans were acquired with the Optic Disc Cube 200x200 protocol for RNFL thickness to measure RNFL thickness in a 3.4mm circle centered on the optic disc. Macular scans were acquired with the Macular Cube 200x200 or 512x128 to measure GCIPL thickness. For the Spectralis scans, average RNFL thickness was measured using a 3.4-mm peripapillary ring scan. Ganglion cell layer data was not included for Spectralis machines. Quality control of images was completed locally as per OSCAR-IB criteria.³⁴ RNFL data from the two machines were harmonized, utilizing the methods detailed as Kenney et al. ³⁵Baseline images were included if available and obtained on the same OCT system utilized during the acute ON scan.

Statistical Analysis.

RNFL thickness and change in RNFL thickness from baseline to acute recurrent ON were stratified by GCIPL thickness at acute recurrent ON and prior baseline RNFL thickness. Visual inspection of the distribution was used to determine cut-offs. An average pRNFL thickness of <70 μm and GCIPL thickness of <65μm were used as thresholds for optic atrophy. Statistical analysis was performed in GraphPad Prism (Boston, MA), with significance set at p ≤0.05. Continuous variables were assessed using paired t-tests or Kruskal-Wallis test (Dunn post-hoc test) and categorical variables with Fisher’s exact test or Kruskal-Wallis test. Correction for multiple comparisons was not completed for this exploratory study, though post-hoc testing was completed as above.

Results

We identified 112 patients (MOGAD, 45[40%]; MS, 36[32%], NMOSD, 20[18%], idiopathic ON, 11[10%]) with 137 episodes of recurrent ON (median age, 39 years [range 12-78]; female sex, 81 [72%], Table 1). Of the ON attacks included, 93 (68%) had one prior attack of ipsilateral ON and 44 (32%) had greater than one prior attack (range, 1-6). OCT evaluation was completed on a median of 4 days (interquartile range, 2-9) after onset of visual symptoms. RNFL thickness at time of recurrent ON was variable; MOGAD, mean 79μm (range, 28-368); MS, 82μm (49-129); NMOSD, 64μm (31-169); idiopathic 87μm (36-185, p=0.009, Figure 1A, Table 2).

Table 1.

Demographics and clinical features of the MOGAD, MS, NMOSD, and idiopathic cohorts

	Full	MOGAD	MS	NMOSD	Idiopathic	p-value
	n=112	n=45	n=36	n=20	n=11
Demographics
Median age at onset, years (range)	39 (12-78)	40 (12-75)	36 (20-68)	45 (19-64)	51 (24-78)	0.20
Children <18 years (%)	4 (4%)	4 (9%)	0 (0%)	0 (0%)	0 (0%)	0.1
Sex, female (%)	81 (72%)	26 (58%)	31 (86%)	16 (70%)	8 (72%)	0.03
Ethnicity, White (%)	83 (79%) b	37 (82%)	32 (89%)	12 (59%) b	5 (45%)	0.02
Ethnicity, Black (%)	11 (10%) b	3 (7%)	3 (8%)	5 (29%) b	1 (9%)	0.1
Clinical Information
Prior episodes of ipsilateral ON,	1 (1-6)	1 (1-5)	1 (1-2)	1 (1-6)	1 (1-4)	0.04 c
Episodes of recurrent ON meeting	1 (1-4)	1 (1-4)	1 (1-2)	1 (1-2)	1 (1-2)	0.59
Time from visual symptoms to OCT,	4 (0-14)	3 (0-14)	6 (0-14)	3 (0-13)	4 (1-12)	0.09

^a,Calculated across 4 disease cohorts;

^b,Data available for Full Cohort, n=109; NMOSD, n=17

^c,Dunn's multiple comparisons test, MS vs NMOSD, p=0.03, other comparisons not significant

Abbreviations: MOGAD, myelin oligodendrocyte glycoprotein antibody-associated disease; MS, multiple sclerosis; NMOSD, neuromyelitis optic spectrum disorder; ON, optic neuritis

Figure 1: Prior optic atrophy precludes RNFL swelling during recurrent ON.

A) Peripapillary RNFL at acute recurrent optic neuritis in full cohort and individual disease subgroups. Bars height represent mean with whiskers demonstrating standard deviation. B) Scatter plot of RNFL thickness in μm at time of acute attack as compared to macular GCIPL thickness in μm. C) Patients with GCIPL ≥65μm had significantly more RNFL swelling at acute ON compared to patients with GCIPL<65μm. Solid line for each group represents the mean. D) Scatterplot of difference between RNFL and GCIPL thickness by GCIPL thickness. Gray bar set at 15μm, threshold suggestive of the potential for RNFL thickening in the acute setting. E) Patients with GCIPL ≥65μm were more likely to have an RNFL ≥15μm (gray bar) thicker than GCIPL during acute ON suggesting RNFL swelling. F) Scatter plot of change in RNFL thickness from baseline to recurrent clinical attack in μm as compared to RNFL thickness at baseline in μm. G) Change in RNFL thickness from baseline to acute recurrent attack in patients with baseline RNFL <70 μm as compared to patients with ≥70μm. Solid line for each group represents the mean; ns, not significant, ***p<0.001.

Abbreviations: GCIPL, ganglion cell layer; ON, optic neuritis; RNFL, retinal nerve fiber layer

Table 2.

OCT values of the MOGAD, MS, NMOSD, and idiopathic cohorts

	Full	MOGAD	MS	NMOSD	Idiopathic	p-value
	n=112	n=45	n=36	n=20	n=11
OCT
Acute RNFL thickness available, n	137	59	39	24	15
Acute RNFL thickness, mean, μm	78 (33- 368)	79 (33- 368)	82 (49- 129)	64 (31- 169)	87 (36- 185)	0.002 b
Acute GCIPL thickness available, n	66	32	25	5	4
Acute GCIPL thickness, mean, μm	66 (48-88)	66 (52-86)	70 (54-88)	55 (48-70)	76 (56- 85)	0.003c
RNFL baseline available, n	47	19	15	6	7
RNFL baseline, mean, μm (range)	67 (27- 102)	60 (27-92)	77 (50- 102)	63 (31-81)	69 (37- 100)	0.04 d
OCT stratified by optic atrophy	Acute GCIPL<65	Acute GCIPL≥65	p-value	Baseline RNFL<70	Baseline RNFL≥70	p-value
Full Cohort, n	30	36		25	22
Full Cohort, RNFL thickness, mean	65 (50-84)	108 (65- 368)	<0.0001	53(27-72)	91 (64- 129)	<0.001
MOGAD, n	17	15		13	5
MOGAD, acute RNFL, mean in μm	62 (51-81)	125	0.004	54 (38-67)	90 (67- 124)	<0.001
μm, n	8	17		4	11
MS, acute RNFL, mean in μm	71 (62-84)	92 (68- 129)	0.003	60 (50-64)	88 (71- 129)	0.01
NMOSD, n	4	1		3	3
NMOSD, acute RNFL, mean, μm	62 (50-72)	77	N/A	51 (31-72)	74 (64-87)	0.14
Idiopathic, n	1	3		4	3
Idiopathic, acute RNFL, mean, μm	65	126 (94- 185)	N/A	53 (36-65)	92 (91-98)	0.005

^a,Calculated across 4 disease cohorts;

^b,Dunn's multiple comparisons test, MS vs NMOSD, p=0.003, other comparisons not significant;

^c,Dunn's multiple comparisons test, MS vs NMOSD, p=0.007; NMOSD vs Idiopathic p=0.006, other comparisons not significant;

^d,Dunn's multiple comparisons test, MOGAD vs MS, p=0.03, other comparisons not significant

Abbreviations: GCIPL, ganglion cell layer; MOGAD, myelin oligodendrocyte glycoprotein antibody-associated disease; MS, multiple sclerosis; NMOSD, neuromyelitis optic spectrum disorder; OCT, optical coherence tomography; RNFL, retinal nerve fiber layer;

Of the 66 patients with available GCIPL thickness during acute ON, patients with preserved GCIPL thickness (≥65μm) had higher RNFL thickness measurements which can be suggestive of RNFL swelling (n=36, mean 108μm[range, 65-368]) as compared to patients with prominent GCIPL atrophy (<65μm), who had lower RNFL thickness values (n=30, 65μm[50-84]; p<0.0001, Table 2, Figure 1BC and 2). Patients with MOGAD and MS with preserved GCIPL were more likely to have higher RNFL measurements which can be indicative of swelling using this cutoff (Table 2). Analysis of the NMOSD and idiopathic cohorts was limited by small sample size. Of patients with preserved GCIPL thickness (≥65μm), 21/36 (58%) had an RNFL ≥15μm thicker than GCIPL suggesting the potential for RNFL swelling as compared to 5/30 (17%, p<0.001) patients with prominent GCIPL atrophy (<65μm, Figure 1DE, Table 3). Similarly in patients with preserved GCIPL thickness (≥65μm), 10/36 (28%) had an RNFL ≥25μm thicker than GCIPL which was highly suggestive of RNFL swelling as compared to 0/30 (0%, p=0.001) patients with prominent GCIPL atrophy (Table 3).

Figure 2: Optical coherence tomography in recurrent ON.

A) A patient with MOGAD and three prior episodes of optic neuritis in the left eye and severe ganglion cell thinning (52 μm) at the time of recurrent acute ON had no RNFL swelling at the time of attack (52 μm). B) A patient with MOGAD and two prior episodes of optic neuritis in the left eye had mild ganglion cell layer thinning (74 μm) and clear swelling of the RNFL (119 μm) during acute recurrent ON. C.a) A baseline OCT evaluation of patient with MOGAD after one episode of acute ON in the right eye with resulting minimal thinning of the GCIPL (85 μm) and subtle RNFL thinning (88 μm). C.b) The same patient displayed in row C.a returned one year later with acute recurrent ON in the right eye. The GCIPL thickness at the time of acute ON is unchanged from the baseline evaluation, however, the RNFL has a 7 μm increase in thickness compared to prior baseline OCT consistent with acute ON.

Abbreviations: GCIPL, ganglion cell layer; MOGAD, myelin oligodendrocyte glycoprotein antibody-associated disease; OCT, optical coherence tomography; ON, optic neuritis; RNFL, retinal nerve fiber layer

Table 3:

Difference between RNFL and GCIPL predicts RNFL swelling in MOGAD, MS, NMOSD, and idiopathic cohorts

	GCIPL<65 μm	GCIPL≥65 μm	p-value
Full Cohort, n	30	36
Full Cohort, frequency of RNFL-GCIPL ≥15μm (%)	5 (17)	21 (58)	<0.001
Full Cohort, frequency of RNFL-GCIPL ≥25μm (%)	0 (0)	10 (28)	0.001
MOGAD, n	17	15
MOGAD, frequency of RNFL-GCIPL ≥15μm (%)	2 (11)	9 (60)	0.008
MOGAD, frequency of RNFL-GCIPL ≥25μm (%)	0 (0)	6 (40)	0.005
MS, n	8	17
MS, frequency of RNFL-GCIPL ≥15μm (%)	2 (25)	11 (65)	0.1
MS, frequency of RNFL-GCIPL ≥25μm (%)	0(0)	3 (17)	0.52
NMOSD, n	4	1
NMOSD, frequency of RNFL-GCIPL ≥15μm (%)	1 (25)	0 (0)	1.0
NMOSD, frequency of RNFL-GCIPL ≥25μm (%)	0 (0)	0 (0)	1.0
Idiopathic, n	1	3
Idiopathic, frequency of RNFL-GCIPL ≥15μm (%)	0 (0)	1 (33)	1.0
Idiopathic, frequency of RNFL-GCIPL ≥25μm (%)	0 (0)	1 (33)	1.0

Abbreviations: GCIPL, ganglion cell layer; MOGAD, myelin oligodendrocyte glycoprotein antibody-associated disease; NMOSD, neuromyelitis optic spectrum disorder; MS, multiple sclerosis; RNFL, retinal nerve fiber layer

In the 47 patients with baseline OCT prior to acute recurrent ON, those with baseline RNFL ≥70μm had a mean RNFL increase of 5.0μm (range −5 to +39, n=22) compared to those with baseline RNFL <70μm who had less swelling with mean RNFL increase of 0.7μm (−4 to +13, n=25) though this difference was not significant(p=0.07, Table 2, Figure 1FG). In MOGAD patients, where anterior optic nerve involvement often causes swelling, those with RNFL ≥70μm had an increase in RNFL from baseline during the time of acute recurrent ON compared to those with baseline RNFL <70μm (10.8 μm [−3 to 32], n=5 vs 0.7μm[−4 to 6], n=14; p=0.01). In MOGAD patients, patients with preserved RNFL thickness (baseline RNFL≥70μm) were also more likely to have clinically meaningful RNFL thickening (≥5μm change in RNFL thickness, 4/5[80%]; compared to RNFL<70μm, 1/14[7%], p=0.006). In the full cohort, there were 23 patients who had baseline GCIPL measurements. There were no differences in GCIPL from baseline to acute recurrent ON in this cohort (mean baseline GCIPL, 65.2μm±10.6; GCIPL at time of recurrent ON, 65.1μm±10.8; p=0.96; mean difference −0.13±1.5)

Discussion

This international multi-center study identifies several key findings that will impact the evaluation of patients with ON that will be broadly applicable to the practicing neurologist and neuro-ophthalmologist and for relapse assessment in ongoing clinical trials in demyelinating disease. We found patients with prior optic atrophy exhibited less RNFL swelling during acute recurrent ON episodes. Specifically, we identified that a prior baseline RNFL <70μm or GCIPL <65μm at the time of attack were thresholds at which RNFL swelling was less likely to occur. We also identified that an average RNFL ≥15μm thicker than GCIPL may reflect RNFL swelling in acute ON in the appropriate clinical context and an average RNFL ≥25μm thicker than GCIPL is highly suggestive of RNFL swelling. These findings assist in OCT interpretation in recurrent optic neuritis, where a true baseline is often unavailable and the fellow eye may no longer be a reliable comparator.³⁶ Taken together, these data establish practical, objective thresholds that can assist bedside decision-making.

Evaluation of patients with recurrent visual symptoms may present a diagnostic challenge, particularly when trying to differentiate between a true acute relapse and pseudorelapse, which has important treatment implications. Thus, objective metrics for relapse evaluation in demyelinating disease are critically important. OCT provides objective measures of optic nerve structures and is increasingly used in neurology and neuro-ophthalmology clinics to evaluate optic neuritis.^37,38 It is also incorporated into multiple ongoing clinical trials as objective evidence of clinical relapse for adjudication committees. In patients with demyelination ON, the contralateral eye has been proposed as surrogate measure for comparison as many patients do not have baseline OCT for comparison. In patients with recurrent ON or longstanding disease, however, the contralateral eye is no longer a suitable surrogate for baseline measurements because the amount of RNFL thinning in an eye with recurrent ON cannot be compared with an unaffected contralateral eye. Furthermore, there may be prior thinning from prior clinical events of ON in the contralateral eye or subclincial thinning in either eye, particularly in patients with MS prior to the widespread use of high-efficacy immunomodulatory therapies.^10,27 Our study identified GCIPL thickness as a surrogate for optic nerve atrophy during acute ON. GCIPL thickness decreases after ON and may be reduced in patients with demyelinating disease outside of ON attacks, particularly in patients with MS.^10,23,27 The GCIPL, however, does not swell during acute ON and thus can be used at the time of ON recurrence to provide a baseline assessment of optic atrophy with inclusion of the thresholds identified in this study.¹⁰ The strength of this approach is that it provides a rapid, reliable measurement which can be obtained at the time of recurrent ON to determine whether an optic nerve can swell during a recurrent ON attack.

In addition to atrophy, the ON etiology impacts expected disc edema and RNFL thickening. MOGAD-associated ON often involves the anterior optic nerve, with ~80% showing disc edema and ~50% showing moderate to severe edema and RNFL thickening in eyes without prior ON.^6,17,39-41 In contrast, only one-third of typical MS ON has disc edema, which is typically mild.^17,42 This likely explains the spread of RNFL values in our MOGAD cohort, where eyes with minimal GCIPL thinning had a propensity for severe RNFL thickening during recurrent ON, while other etiologies had less RNFL thickening even when the GCIPL was relatively intact. In all etiologies, a severely atrophic nerve with GCIPL thinning had minimal RNFL thickening during recurrent ON.

Our findings have implications for attack adjudication criteria in clinical trials, particularly for MOGAD, where ON is the dominant attack type.^6,23,43 There are multiple clinical trials that are currently ongoing where OCT RNFL thickening is included as objective evidence in the relapse assessment for patients with acute ON. In care of patients with severe optic atrophy, MRI orbit with contrast with fat saturation along with careful clinical examination including evaluation for new relative afferent pupillary defect, visual acuity, and color vision may play a larger role confirming a clinical event.²⁰

Our study is limited by the retrospective nature of the evaluation. Given the retrospective nature, there were limited patients with baseline evaluation and longitudinal assessment to best assess potential OCT temporal dynamics, however, this is often representative of real-world clinical care. ^44,45 We also included two OCT platforms which are commonly used for assessment and monitoring of optic neuritis and demyelinating disease throughout the world though used harmonization procedures to adjust for differences between machines. For longitudinal comparisons of RNFL, patients were scanned on the same platform. These data suggest that GCIPL may be used as a surrogate for patients without baseline assessment in clinical practice. The difference between RNFL and GCIPL thickness may be utilized, and RNFL swelling should be considered in patients where the difference between RNFL and GCIPL is greater than 15 μm in the appropriate clinical context. Furthermore, there were limited patients with NMOSD and idiopathic ON despite the multicenter approach. The limited number of patients with these conditions precluded detailed statistical analysis of all subgroups. In addition, acute treatments for ON were not standardized, including in the type, timing, and duration of acute treatment. There is a potential this may have impacted our findings, particularly in the MOGAD group, where there is some data that early steroids may impact clinical course.^44,45 Other imaging modalities, such as MRI, were not evaluated in this cohort, and future work could consider longitudinal studies of OCT with MRI in recurrent ON in a prospective manner.

In summary, patients with prior optic atrophy exhibit markedly reduced RNFL swelling during acute attacks and macular GCL thickness can serve as a robust surrogate for underlying optic atrophy in the acute setting. Using OCT, we identified quantitative thresholds (GCIPL [<65μm] or prior RNFL [<70μm]) at which optic nerve swelling is unlikely to occur during recurrent acute ON which can be easily utilized in clinical practice and clinical trials in care of patients with demyelinating disease.

Highlights:

• Optical coherence tomography can measure retinal nerve fiber layer (RNFL) thickness.

• In acute optic neuritis (ON), RNFL swelling can occur.

• RNFL swelling during acute ON is limited in patients with prior optic atrophy.

• Ganglion cell layer thickness may serve as a surrogate for optic atrophy during acute ON.

Data Availability:

Anonymized data not published within this article will be made available by request from any qualified investigator.