Effects of Myopia on Rates of Change in Optical Coherence Tomography Measured Retinal Layer Thicknesses in People with Multiple Sclerosis and Healthy Controls

Abstract Purpose To quantify the associations of myopia with longitudinal changes in retinal layer thicknesses in people with multiple sclerosis (PwMS) and healthy controls (HC). Methods A cohort of PwMS and HC with recorded refractive error (RE) prospectively scanned on Cirrus HD-OCT at the Johns Hopkins MS Center was assessed for inclusion. Exclusion criteria included OCT follow-up < 6 months, ocular comorbidities, incidental OCT pathologies, and inadequate scan quality. Eyes were classified as having high myopia (HM) (RE≤ −6 diopters), low myopia (LM) (RE> −6 and ≤ −3 diopters), or no myopia (NM) (RE> −3 and ≤ +2.75). Linear mixed-effects regression models were used in analyses. Results A total of 213 PwMS (eyes: 67 HM, 98 LM, 207 NM) and 80 HC (eyes: 26 HM, 37 LM, 93 NM) were included. Baseline average ganglion cell/inner plexiform (GCIPL) and peri-papillary retinal nerve fiber layer (pRNFL) thicknesses were lower in MS HM compared with MS NM (diff: −3.2 µm, 95% CI: −5.5 to −0.8, p = 0.008 and −5.3 µm, 95% CI: −9.0 to −1.7, p = 0.004, respectively), and similarly in HC HM, as compared with HC NM. Baseline superior, inferior, and nasal pRNFL thicknesses were lower in HM compared with NM, while temporal pRNFL thickness was higher, both in MS and HC (MS: 7.1 µm, 95% CI: 2.7–11.6, p = 0.002; HC: 4.7 µm, 95% CI: −0.3 to 9.7, p = 0.07). No longitudinal differences in rates of GCIPL change were noted between HM and LM vs. NM, either in MS or HC. Conclusion Cross-sectional differences in average GCIPL and pRNFL thicknesses are commonly seen in people with HM as compared to reference normative values from people with NM and can lead to false attribution of pathology if RE is not taken into account. However, our study suggests that longitudinal changes in average GCIPL thickness in PwMS with myopia are similar in magnitude to PwMS with NM, and therefore are appropriate for monitoring disease-related pathology.


Introduction
Myopia is an extremely common condition, with an estimated global prevalence of 22.9%, while high myopia's (HM) estimated global prevalence is 2.7%. Notably, both myopia and HM's prevalence are expected to increase significantly in the future, with estimates of 49.8% and 9.8%, respectively, in 2050. 1 Myopia has been shown to be associated with lower average macular ganglion cell/inner plexiform layer (GCIPL), and peri-papillary retinal nerve fiber layer (pRNFL) thicknesses in healthy individuals, as measured by optical coherence tomography (OCT), a non-invasive, reproducible, and high-resolution imaging technique of the retina. [2][3][4] In addition, there is also a suggestion of accelerated rates of average GCIPL and pRNFL atrophy in otherwise healthy individuals with HM in a single-center study performed in Korea. 5,6 The retina is an unmyelinated structure of the central nervous system that can be studied using OCT to monitor neurodegeneration in multiple sclerosis (MS). People with MS (PwMS) exhibit accelerated rates of average GCIPL and pRNFL thinning. 7 Moreover, rates of GCIPL atrophy are associated with brain atrophy, especially grey matter atrophy, and are differentially affected by disease-modifying treatments in relapsing-remitting MS. 8,9 However, because of the known thinning effect of myopia on retinal layer thicknesses in healthy individuals, people with HM have been largely excluded from MS studies that use OCTderived thicknesses as outcome measures.
In this observational study, we sought to assess the crosssectional and longitudinal effects of varying degrees of myopia on OCT-derived retinal layer thicknesses and thickness changes respectively, with the overarching goal of determining whether HM are valid to include in longitudinal OCT MS studies/trials.

Standard protocol approvals, registrations, and patient consents
This study was approved by the Johns Hopkins Institutional Review Board. All study participants provided written informed consent.

Study design and study participants
A total of 287 PwMS and 167 HC who were willing to participate in our study were identified from a prospective cohort followed at the Johns Hopkins MS Center (Baltimore, Maryland) with serial retinal imaging with OCT, and were retrospectively assessed for inclusion. Participants were studied between September 2008 and July 2021. Diagnosis in PwMS was confirmed by the treating neurologist based on the 2005 revised McDonald criteria. 10 MS subtype was classified according to the Lublin classification as relapsing-remitting (RRMS), primary progressive (PPMS), or secondary progressive MS (SPMS). 11 In order to define refractive error (RE) for the included participants, we retrospectively reviewed the ophthalmological electronic medical records (EMR) to record the subjective, non-cycloplegic RE. The ophthalmological EMR were reviewed to also record ophthalmological comorbidities and/or history of laser assisted in situ keratomileusis (LASIK) for RE correction. The EMR were reviewed to record demographic and clinical data, including diseasemodifying treatments (DMT) and their time periods of use. DMT were classified as low-potency (glatiramer acetate and interferon-beta), intermediate-potency (dimethyl fumarate, fingolimod, siponimod, and teriflunomide), and highpotency (alemtuzumab, daclizumab, natalizumab, ocrelizumab, and rituximab).
The inclusion criteria included: (1) available record of RE assessment, and (2) OCT follow-up ! 6 months. The exclusion criteria included: (1) acute optic neuritis (ON) within 6 months from the baseline visit, (2) ophthalmological and/ or neurological comorbidities that can affect retinal thickness, (3) incidental retinal pathologies, (4) post-LASIK only available RE, and (5) hyperopia, with RE > 2.75. Follow-up data of eyes that developed an acute ON were censored at the time of the last available OCT before the ON event.

Group classification according to RE
The eyes of the participants were classified as HM eyes, when RE À6 diopters, low myopia (LM) eyes, when À6 diopters < RE À3 diopters, and no myopia (NM) eyes, when À3 diopters < RE þ2.75 diopters. As a result, the eyes were classified into the following subgroups: HC NM, HC LM, HC HM, MS NM, MS LM, and MS HM. 1,12,13 Procedures Imaging of the retina was performed using spectral-domain OCT (Cirrus HD-OCT, Model 5000, software version 11.5; Carl Zeiss Meditec, Dublin, CA), as previously described. 14 Macular scans and peri-papillary scans were acquired from each participant. Macular volume scans were acquired with the Macular Cube 512 Â 128 protocol, and peri-papillary scans were obtained with the Optic Disc Cube 200 Â 200 protocol. The OSCAR-IB quality control criteria were applied to exclude scans of inadequate quality. 15,16 Measurement of macular layer thicknesses was performed with the use of automated macular segmentation, as described elsewhere. 17 The average thickness of each macular layer was obtained within a fovea-centered annulus, with an internal diameter of 1 mm and an external diameter of 5 mm. The average macular layer thicknesses estimated included the GCIPL, the inner nuclear layer (INL), and the outer nuclear layer (ONL). The average macular thickness was also estimated using the same method. Segmented macular scans were assessed in a blinded manner for segmentation accuracy, as well as incidental retinal pathology identification.
Because improper nasal displacement of the optic nerve head scan circle in HM with tilted optic discs may lead to falsely higher estimated temporal pRNFL thickness, we also estimated the nasal macular retinal nerve fiber layer (nasal mRNFL) thickness to confirm any observed differences in temporal pRNFL thickness between the myopia groups. 18,19 Measurement of pRNFL thickness was estimated using the incorporated software of the Cirrus HD-OCT device. The pRNFL measures included the average pRNFL and the quadrantic pRNFL thicknesses (superior, nasal, inferior, and temporal).

Statistical methods
Baseline characteristics were compared between patient groups using the following tests: Wilcoxon rank-sum test (age, disease duration, and follow-up), and Chi-squared test (sex, race, MS subtype, baseline DMT, and patient-years on DMT during follow-up).
Baseline OCT measures were compared between groups of myopia in HC and MS using mixed-effects linear regression models with random subject-specific intercepts. For comparison between groups of HC eyes, models were adjusted for age, sex, and race (Black/African American [AA] vs non-AA), while for comparison of groups of MS eyes, models were additionally adjusted for ON history, MS subtype, disease duration, and baseline DMT potency.
For comparison of longitudinal rates of change in OCT measures between groups of myopia, we fit mixed-effects linear regression models with random subject-and eye-specific intercepts and random slopes in time from baseline OCT measurement. Time (in years) was used as a continuous variable. These models account for intra-participant, inter-eye correlations. To more readily interpret model coefficients obtained from the longitudinal analyses, we used log-linear models as an approximation of the mean annualized percent changes in OCT-measured retinal layer thicknesses. 20 For comparison between HC myopia groups, models were adjusted for age at baseline, sex, and race. For the comparison between MS myopia groups, the models were additionally adjusted for ON history, MS subtype, and disease duration at baseline measurement. A cross-product term (i.e. myopia group Ã time) was used in each model; the coefficient for this term can be interpreted as the difference in the rate of change of the OCT measure between the myopia groups.
In order to further characterize the effects of myopia on rates of change in average GCIPL and pRNFL thicknesses, we fitted additional mixed effects models including RE as a continuous variable and its interaction with time of followup (RE Ã time). In addition, we modeled RE using restricted cubic splines to assess potential nonlinear relationships between RE and rates of change.
Statistical analyses were performed using STATA version 16 (StataCorp, College Station, TX). Statistical significance was defined as p < 0.05.

Study population
A total of 213 PwMS and 80 HC (372 MS eyes and 156 HC eyes), followed for a median duration of 4.4 years (interquartile range ¼ 2.3-7.4 years) were eligible for inclusion in this study (Supplementary Figure 1). Demographics of participants are shown in Table 1. Eye characteristics and baseline OCT measures, are summarized and presented in Table 2.

Baseline cross-sectional comparisons of OCT measures between myopia groups
Comparisons of baseline OCT measures between groups of myopia in HC and MS are presented in Figures 1 and 2 Figure 2). A similar numerical pattern was noted in LM compared with NM for both HC and MS, although the differences were less pronounced and not statistically significant.

Comparisons of longitudinal rates of change of OCT measures between myopia groups
Comparisons of longitudinal rates of OCT measures between myopia groups in MS are presented in Table 3. No differences were noted in rates of change of OCT measures between myopia groups in HC (data not shown).
No differences were noted in rates of change of GCIPL thickness between MS HM and MS NM, or between MS LM and MS NM. The rate of atrophy of average pRNFL thickness was slower in MS LM compared with MS NM (À0.50% [95% CI: À0.67% to À0.33%] vs.À0.74% [95% CI: À0.86% to À0.62%] per year, p ¼ 0.03). The rate of change of average pRNFL thickness was also numerically lower in MS HM compared with MS NM, but was not statistically significant.

Effects of RE on longitudinal rates of OCT measures
No associations were noted between RE as a continuous variable and rates of average GCIPL atrophy either in MS, or in HC. Each additional one diopter of myopia was associated with slower rates of average pRNFL atrophy in MS (þ0.03% [95% CI: þ0.00% to þ0.06%] per year, p ¼ 0.05), but no association was noted in HC.

Baseline and longitudinal comparison of nasal macular RNFL between myopia groups
The optic nerve head scan circle may be improperly nasally displaced in HM with tilted optic discs, leading to falsely higher estimated temporal pRNFL thickness. 18,19 Thus, to confirm the temporal pRNFL differences noted in our study, we estimated the nasal mRNFL thickness and performed baseline and longitudinal comparisons between myopia groups in MS and HC (Supplemental Figure 2). Comparing with HC NM, baseline nasal mRNFL was thicker in HC HM

Discussion
In this study, we aimed to assess the effects of myopia on baseline OCT-derived retinal layer thickness measures, as well as longitudinal rates of change in OCT measures, both in HC and MS. We found that average GCIPL and pRNFL thicknesses were lower in MS and HC HM than NM, which is important to be aware of when using NM HC reference databases to determine if OCT retinal measures are indeed normal. Superior, inferior, and nasal pRNFL quadrant thicknesses were lower in myopia, but temporal pRNFL thicknesses were higher. Importantly, rates of GCIPL atrophy did not differ with varying degrees of myopia in MS, suggesting that once a proper baseline has been established for an individual, the change in this measure can be used to monitor disease-related pathology, as is done in NM PwMS. However, we show that rates of pRNFL atrophy may be slower in people with higher degrees of myopia, and therefore monitoring change in GCIPL thickness is recommended.
As shown in a recent meta-analysis, myopia has been associated with lower average GCIPL and pRNFL thicknesses in otherwise healthy individuals; 2 our findings in HC and PwMS are in accordance with this study. A prior study reported accelerated rates of GCIPL and average pRNFL atrophy in HC HM eyes. That study was performed in Korea and compared high myopes to low/intermediate myopes in a cohort of 80 eyes per group with 60% males. 5,6 In our study, we had mostly Caucasians and some black Americans with a higher proportion of females, which reflects the known higher incidence of myopia in women. 21 We did not find an association of myopia with rates of change of GCIPL and average pRNFL thickness in HC. Importantly, we did not find an association between myopia and rates of average GCIPL thickness change in the MS cohort either, which was followed for an even longer time.
However, there seemed to be an association between myopia and slower rates of average pRNFL change in MS.
Our study findings suggest that the rate of GCIPL atrophy as measured by OCT can be potentially used in the monitoring of PwMS with HM or LM. In addition, crosssectional GCIPL thicknesses may be used in the assessment of LM PwMS. It is widely accepted that GCIPL thickness is an important biomarker in MS, and it has been used as an outcome measure in multiple MS studies and clinical trials. 22 These studies have largely excluded MS HM eyes, due to the potential confounding effect of HM on retinal layer thicknesses. Regarding pRNFL thickness, myopia exerts an effect on average pRNFL thickness both at baseline, as well as longitudinally in MS, with potential implications in regards to its clinical usefulness in HM PwMS.
There are many potential explanations of the effects of myopia on GCIPL and pRNFL thickness. First, the increased axial length in myopia may exert stretching forces on the retina. Because the peripheral retina (including the outer macula) lacks large blood vessels and retinal nerve fibers, it might be less resistant to mechanical traction. Peripheral retinal thinning may compensate for tractional retinal forces. [23][24][25][26] The hypothesis of mechanical stress is also supported by our finding that all OCT-measured layers were thinner at baseline in HM. Besides, tensile stress may also damage the retinal nerve fibers. 27 Second, there is some evidence of vascular compromise in myopia that could result in a net loss of ganglion cells. 28 Nevertheless, a compromised vascular density due to retinal thinning-mediated lower oxygen demand cannot be excluded. In addition, magnification effects may have affected our measurement in HM, especially since Cirrus HD-OCT does not provide an option for automated magnification correction. [29][30][31] Similar to the findings of the meta-analysis discussed above, we found lower cross-sectional pRNFL thicknesses in all quadrants both in HC and MS with HM, with the exception of temporal pRNFL thickness, which was higher in people with HM. The paradoxically higher temporal pRNFL thickness in myopia may be the result of the converging tendency of the superotemporal and inferotemporal RNFL bundles to the temporal side, which is more resistant to traction. 2,32,33 It is also worth mentioning that improper nasal displacement of the optic disc ring on optic disc OCT scans has been described in myopic eyes with tilted optic discs, leading to an overestimated temporal pRNFL thickness. 18,19 However, we do not believe this was the case in our study, since we found similarly higher baseline thickness for the nasal macular RNFL, the axons in which give rise to the temporal pRNFL.
Our study has a number of limitations. First, we did not measure axial length and RE prospectively, but instead, we used the available ophthalmological EMR at available dates. However, myopia progression rate starts decreasing early in adolescence, and remains slow during adulthood. 34, 35 Second, we used the subjective RE. While subjective RE measurement might be less accurate compared to objective RE techniques, the mean differences between the two seem to be small. 36 Third, the groups compared were not perfectly matched and thus had minor differences in demographic characteristics, and DMT use during follow-up. Additionally, Cirrus HD-OCT does not account for rotation on scan acquisition, which could have affected our study's measured rates of quadrantic pRNFL thicknesses. Finally, Cirrus HD-OCT does not provide an automated correction of magnification error, which could have again affected our longitudinal analyses, as the technicians may have acquired scans by manually correcting for different magnification errors over visits. Even though the technical aspects of scan acquisition could have arguably limited our study, OCTmeasured retinal layer thicknesses have excellent reproducibility, both between sessions and between observers. 14,[37][38][39][40] Given the increasingly recognized utility of OCT in monitoring neurological diseases, and the high prevalence of myopia, this study provides valuable information regarding expected baseline OCT retinal thickness values in myopic PwMS and may help avoid attribution of pathology to layers that are known to be thinner in HM vs. NM due to the eccentric spherical shape of the eye. Furthermore, these data validate the utility of assessing longitudinal change in GCIPL thickness to monitor disease-related thinning in people with myopia, who may have previously been excluded from research analyses and clinical monitoring. ESS has served on scientific advisory boards for Viela Bio, Alexion, and Genentech, and has received speaker fees from Viela Bio and Biogen.
SS has received consulting fees from Medical Logix for the development of CME programs in neurology, and has served on scientific advisory boards for Biogen, Genzyme, Genentech Corporation, EMD Serono, and Celgene. He is the PI of investigator-initiated studies funded by Genentech and Biogen, was the site investigator of a trial sponsored by MedDay Pharmaceuticals and received support from the Race to Erase MS foundation. He has consulted for Carl Zeiss Meditec, and has received equity compensation for consulting from JuneBrain LLC, a retinal imaging device developer.
PC has received consulting fees from Disarm and Biogen and is PI on grants to JHU from Biogen and Annexon.

Data availability statement
The data that support the findings of this study are available from the corresponding author, PAC, upon reasonable request.