Peripapillary hyper-reflective ovoid mass-like structures (PHOMS) in AQP4-IgG-positive neuromyelitis optica spectrum disease (NMOSD) and MOG-IgG-associated disease (MOGAD)

Background Peripapillary hyperreflective ovoid mass-like structures (PHOMS) have recently been described as new optical coherence tomography (OCT) marker. It is not yet clear whether the occurrence of PHOMS is disease-specific or disease-spanning. PHOMS have been described in 16–18% of patients with multiple sclerosis (MS). Currently, no data on the prevalence of PHOMS in other demyelinating diseases including aquaporine-4-IgG-positive neuromyelitis optica spectrum disease (AQP4 + NMOSD) or myelin oligodendrocyte glycoprotein-IgG-associated disease (MOGAD) are reported. Methods We performed a cross-sectional, retrospective spectral domain OCT study evaluating the frequency of PHOMS in AQP4 + NMOSD (n = 47) and MOGAD (n = 44) patients. To test the association with retinal neuroaxonal damage, we compared demographic and clinical data as well as retinal layer thicknesses between eyes with vs. eyes without PHOMS. Results PHOMS were detected in 17% of AQP4 + NMOSD and 14% of MOGAD patients. Intra-cohort analysis revealed that AQP4 + NMOSD patients with PHOMS were significantly older [mean (years): 57.5 vs. 50.0; p value = 0.04]. We found no association of PHOMS with retinal neuroaxonal degeneration. In addition, in subjects with only one eye affected by PHOMS compared with the unaffected fellow eye, no differences in retinal parameters were observed (n = 4). Conclusions In summary, we found PHOMS in 17% of AQP4 + NMOSD and 14% of MOGAD patients. This is comparable to the prevalence of published MS PHOMS data. Therefore, a disease-specific occurrence of PHOMS is unlikely. Interestingly, PHOMS do not seem to depend on retinal neuroaxonal degeneration.


Introduction
Peripapillary hyper-reflective ovoid mass-like structures (PHOMS) are a recent finding in optical coherence tomography (OCT). PHOMS are thought to be associated with axoplasmic stasis and/or congestion in the glymphatic translaminar pressure system [1]. However, pathophysiology and significance of PHOMS remain unclear. Nevertheless, as a very new OCT marker, PHOMS have already been investigated in various neurological diseases including multiple sclerosis (MS). Here, a prevalence of PHOMS positivity was shown in 16-18% of patients [2,3]. However, data on the prevalence of PHOMS in other inflammatory cerebral nervous system (CNS) disorders are lacking and little is known about the PHOMS frequency in adult healthy controls (HC). Therefore, we aimed to investigate the prevalence of PHOMS in patients with (i) aquaporine-4-IgG-positive (AQP4) Jonathan  neuromyelitis optica spectrum disorders (AQP4 + NMOSD) and (ii) myelin oligodendrocyte glycoprotein-IgG (MOG)associated disease (MOGAD).

Study design
In this cross-sectional, retrospective cohort study, patients with AQP4 + NMOSD and MOGAD were evaluated for the prevalence of PHOMS. All NMOSD and MOGAD patients from two university hospitals (Institute of Clinical Neuroimmunology, NeuroVisionLab, LMU Hospital; Department of Neurology, Klinikum rechts der Isar, TUM school of medicine, Technical University of Munich; recruitment: 2013 until 2021) with availability of clinical data and at least one OCT scan were included in the analysis. Diagnosis of AQP4 + NMOSD was defined by Wingerchuk et al. [4] and MOGAD by Jarius et al. [5]. Age-and sex-matched HC were included. Exclusion criteria were systemic or ophthalmologic diseases that might affect OCT data (arterial hypertension; diabetes mellitus; refraction error of > 5 dioptres; history of any known eye disease; eye surgery). All eyes with a (anamnestic and/or clinical) history of optic neuritis were excluded from OCT retinal layer analysis. The study was approved by the ethics committee of LMU and TUM and performed according to the Declaration of Helsinki. All individuals gave written consent.

Assessment of PHOMS
Radial scans were examined for presence of PHOMS according to the multirater consensus of 2020 [7] by two independent, experienced raters (RW; JAG) blinded to clinical information: PHOMS present (PHOMS +) vs. no PHOMS present (PHOMS−). Incoherent ratings were categorized as PHOMS + or PHOMS− in an open discussion (all authors).

Statistical analysis
For statistical analysis, a paired eye approach was used to account for inter-eye correlations in each patient. For this purpose, mean values of both eyes were calculated if both eyes were available and assigned to the same group. Graph-Pad Prism (v9.1.1) was used for statistical analysis. Interrater agreement on rating of PHOMS per patient as well as eyes was calculated using Cohen´s kappa [8]. Patients with unilateral PHOMS were classified as PHOMS patients for further analysis. In case of unilateral PHOMS+, the PHOMS− eye was excluded from further analysis. For statistical analysis, we used the Fisher's exact test for categorical data, the unpaired t test for normally distributed quantitative parameters and the Mann-Whitney U test for non-parametric data. Data are represented as median with corresponding 25-75% interquartile range. Statistical significance was set at p < 0.05.

Study cohorts
In total, 131 subjects were screened, consisting of 81 AQP4 + NMOSD and 50 MOGAD patients. Of the patients screened, 47 AQP4 + NMOSD patients with 89 eyes, 44 MOGAD patients with 84 eyes met the stated inclusion criteria. Due to a previous ON, 35 eyes each had to be excluded in both the AQP4 + NMOSD (bilateral ON: n = 22 eyes, unilateral ON: n = 13 eyes) and MOGAD group (bilateral ON: n = 18 eyes, unilateral ON: n = 17 eyes) for further retinal layer analysis, but not for PHOMS analysis. A pool of 55 HC with 110 eyes was selected, of which 47 subjects could be analysed age-and sex-matched to the AQP4 + NMOSD cohort (HC.1) and 36 subjects to the MOGAD cohort (HC.2) (Tables 1, 2).

PHOMS rating
When evaluating all included eyes for occurrence of PHOMS an inter-rater agreement of 0.86 per patient (κ = 0.69, good inter-rater agreement), 0.91 per eye (κ = 0.55, moderate agreement) was reached. Radial scan images with PHOMS detected within the HC are presented in the Fig. 1.

OCT characteristics in subjects with unilateral PHOMS
A total of 4 individuals (25% of all PHOMS + subjects) showed PHOMS detection unilaterally only (1 AQP4 + NMOSD, 2 MOGAD, 1 HC). This cohort also showed no significant difference in retinal neuroaxonal degeneration of the PHOMS + eye compared with the PHOMS-eye (data not shown).

Discussion
We found PHOMS in 17% of patients wit h AQP4 + NMOSD and in 14% with MOGAD. In both cohorts, the proportion of patients with PHOMS was comparable to previously published data on PHOMS prevalence in MS [2,3]. Occurrence of PHOMS in AQP4 + NMOSD and MOGAD patients was not associated with sex, disease duration, disability and retinal neuroaxonal degeneration. Previous studies have shown that retinal neuroaxonal degeneration is prominent in nON eyes in MS patients, whereas ON-independent retinal neuroaxonal degeneration in AQP4 + NMOSD and MOGAD currently remains under debate [9][10][11]. If PHOMS occur in relation to the extent of neuroaxonal degeneration, a different prevalence between the diseases could be suspected. Based on our data, no association between retinal neuroaxonal degeneration and PHOMS prevalence is apparent. Interestingly, the only factor we found was an age-dependent association of PHOMS prevalence in AQP4 + NMOSD patients. In addition to a possible association with retinal neuroaxonal degeneration, it is speculated that an impaired retinal glymphatic system or axoplasmatic congestion might be linked to PHOMS [2]. The glympathic system has been studied in the context of neurodegenerative diseases [12], whereas little is known in MS, NMOSD or MOGAD [13]. One could speculate that there is a direct link to the glymphatic system in AQP4 + NMOSD through the presence of AQP4-IgG. The functional integrity of aquaporin-4 water channels is essential for retinal homeostasis [12,14]. Binding of AQP4-IgG might disrupt the retinal glymphatic system and thus lead to an increased frequency of PHOMS in AQP4 + NMOSD [14]. In addition, the age-dependent increase in PHOMS frequency within our AQP4 + NMOSD cohort might support the hypothesis that PHOMS could  [2]; (ii) Hamann et al. described 53 patients with unilateral non-arteritic anterior ischaemic optic neuropathy (NA-AION) and found 5 PHOMS in 30 unaffected (no NA-AION) and optic drusenfree eyes [15]. We found evidence of PHOMS in 4% of HCs.
However, for our PHOMS + HCs, we cannot definitively exclude a possible subclinical ophthalmologic, neurologic, or vascular disease, because the information on ophthalmologic comorbidities was based solely on the medical history of the included HCs. Thus, it is necessary to independently investigate the prevalence of PHOMS in larger HC cohorts. Nevertheless, our data show a higher prevalence of PHOMS in AQP4 + NMOSD and MOGAD, suggesting a general disease dependency of the occurrence of PHOMS.
The power of our study is limited due to the retrospective cross-sectional nature, the sample sizes of very rare diseases and the ON classification mainly based on anamnestic information. An international, multicentre design with longitudinal assessment of PHOMS could provide further clarity on this issue.

Conclusions
In summary, we found PHOMS in 17% of AQP4 + NMOSD and 14% of MOGAD patients comparable to published MS PHOMS data. A disease-specific occurrence is unlikely, but a disease-dependent occurrence can be assumed with a higher prevalence of PHOMS compared to HCs. Therefore, a dedicated examination for PHOMS in different disease groups is useful, especially since the presence of PHOMS could also be mistakenly classified as papilledema in the ophthalmological examination.