Abstract
Objective:
To assess retinal ganglion cell (RGC) injury and sex differences in axon loss in pediatric multiple sclerosis (MS).
Methods:
This is a cross-sectional evaluation of consecutive pediatric MS subjects and controls. Eyes with acute optic neuritis (ON) within 6 months of visit were excluded. Spectral domain optical coherence tomography (OCT) included peripapillary ring and macular scans with post-acquisition segmentation of retinal layers using automated software (Heidelberg v1.8.6.0). Generalized estimating equations (GEEs) measured associations of sex, history of ON, disease duration, and age with OCT outcomes.
Results:
In all, 53 MS subjects (100 eyes, median disease duration = 1.0 years, interquartile range (IQR) = 0.3, 2.5) were compared to 19 control subjects (38 eyes). Eyes with history of ON showed reduced retinal nerve fiber layer (RNFL: −26.8 µm, 95% confidence interval (CI) = −38.9, −14.8, p < 0.001) and 26% lower ganglion cell layer (GCL) volumes (−0.12 mm3, 95% CI = −0.16, −0.072, p < 0.001) compared to control eyes. Non-ON MS eyes had lower temporal RNFL (−11.9 µm, 95% CI = −18.6, −5.3, p < 0.001) and GCL volumes (−0.036 mm3, 95% CI = −0.06, −0.011, p = 0.004) than control eyes. In MS eyes, males versus females had lower global RNFL (−9.4 µm, 95% CI = −17.4, −1.33, p = 0.022) and in ON eyes had lower temporal quadrant RNFL (−9.6 µm, 95% CI = −15.1, −4.15, p = 0.001).
Conclusion:
Subclinical retinal injury occurs in pediatric-onset MS patients without a history of ON. As in adult-onset MS, substantial GCL thinning is present in eyes with prior ON. Finally, greater retinal axonal injury occurs in boys compared to girls.
Introduction
Pediatric-onset multiple sclerosis (MS) patients share similar features with adult patients, but also demonstrate notable differences including higher relapse rates, less short-term disability progression, and, among the youngest patients (age < 11 years), distinct cerebrospinal fluid (CSF) and magnetic resonance imaging (MRI) features.1–5 It has been suggested that children remyelinate lesions and recover more readily from injury than adult-onset patients with MS.4,5 However, while children may initially have better outcomes than adults with similar disease duration, they will likely spend more of their lives with disability than those with adult-onset MS. Tools for measuring neuroprotection are therefore as critical, if not more so, than in adults. 6
Given the importance of gray matter injury for disability in MS,7–9 establishing the timing of neuronal injury in pediatric-onset MS is essential for development of neuroprotective strategies in children. Standard measures of gray matter injury obtained from MRI include assessment of a complex metric that includes the cell soma, axonal fibers, and glial supportive cells including cortical myelin (“gray matter fraction”) that make up “cortex.” Utilizing MRI to assess the degree of injury in children also poses significant challenges based on logistical and safety issues created by requirements for anesthesia as well as additional cost burdens. Optical coherence tomography (OCT), a non-invasive technique for measuring retinal tissue integrity and thickness, has been demonstrated to have significant utility as an outcome marker in adult MS patients. 10 Measurements can be made of the peripapillary retinal ganglion cell (RGC) axons that form the optic nerve as well as the cell bodies of these neurons, which reside in the retinal ganglion cell layer (GCL) of the macula. With modern spectral domain technology, high-resolution scans may be performed reliably, enabling rapid assessment of neuronal and axonal loss in the visual pathway.11,12 The role of OCT in pediatric-onset MS is not well established and demonstration that high-quality retinal layer segmentation is feasible in young children is central to its use in this population. Given the advantages of very low risk and significantly less cost than MRI, we undertook analysis of spectral domain imaging of the retina in pediatric-onset MS patients.
Sex differences in adult optic neuritis (ON) outcomes have been reported, including higher degree of cumulative recovery in women versus men at 3 months; 13 however, the influence of hormonal factors is unknown. It is also unknown whether in childhood there are sex differences in retinal injury from ON.
We aimed to document RGC injury in a pediatric-onset MS cohort and to evaluate differences in neuronal levels of injury by sex in eyes with or without a history of ON.
Methods
Subjects
Consecutive subjects with pediatric-onset MS (<18 years at first symptom) who presented for visual evaluation to the University of California—San Francisco (UCSF) Pediatric MS Center (2009–2014) and completed spectral domain OCT evaluations before their 18th birthday were included in this Institutional Review Board (IRB)-approved study. Patients met published criteria for pediatric-onset MS.14,15 Eyes with acute ON within 6 months of visit or other ophthalmologic disease that would affect OCT measurements were excluded.
Control subjects included pediatric patients presenting for neuro-ophthalmic evaluation without evidence of MS or afferent visual dysfunction. Diagnoses for these controls included migraine (n = 6), conversion disorder (n = 1), ataxia (n = 1), Meniere’s disease (n = 1), sinusitis (n = 1), neurological disorder not otherwise specified (n = 3), isolated transverse myelitis (n = 2), isolated sixth nerve palsy (n = 1), and acute disseminated encephalomyelitis (ADEM) (n = 3) without clinical or radiographic evidence of visual involvement.
OCT
Spectral domain OCT was performed (SPECTRALIS, Heidelberg), including high-resolution macular scans. The OSCAR-IB quality criteria (based on recognition of obvious problems, poor signal strength, centration of scan, algorithm failure, retinal pathology other than MS related, illumination and beam placement) were met. 16 Contemporary studies of OCT measurements in MS patients have demonstrated predominance of temporal patterns of retinal nerve fiber layer (RNFL) loss,17,18 and in addition to mean global peripapillary RNFL, we also examined temporal quadrant measurements for analysis.
Segmentation of retinal layers to determine individual layer thickness was performed using automated software algorithm (Heidelberg v1.8.6.0). Laboratory quality control of segmentation was performed to evaluate any algorithm failure before further processing.
Statistics
Generalized estimating equations (GEEs) were used to measure associations of sex, ON, disease duration, and age with RNFL thickness, total macular volume, GCL volume, averaged combined GCL and inner plexiform layer (IPL), and inner nuclear layer (INL) thickness. These GEE models adjusted for within patient inter-eye correlations. While it was beyond the scope of this project to rigorously compare different reproducibility measures of the GCL provided by the Heidelberg software, we used a Pearson correlation coefficient to compare GCL volumes and GCL/IPL combined thickness measurements to demonstrate the similarity of these outcome measures in our cohort.
Results
In total, 53 MS subjects (n = 100 eyes) and 19 control subjects (n = 38 eyes) were included. Demographic features are similar to other published cohorts of pediatric-onset MS patients and summarized in Table 1. High-quality scans were obtained even in the youngest patients (Figure 1). Twenty-seven eyes had a history of clinical acute ON greater than 6 months prior to the date of OCT evaluation.
Subject characteristics.
IQR: interquartile range; SD: standard deviation; ON: optic neuritis; RNFL: retinal nerve fiber layer; GCL: ganglion cell layer; IPL: inner plexiform layer; SD: standard deviation; INL: inner nuclear layer.
GCL volume calculation generated by the Heidelberg 1.8.6.0 software.
High-resolution macular scan and segmentation from eye of a 7-year-old boy with MS.
Peripapillary RNFL measurements
After adjustment for age and sex, eyes of subjects with MS had significantly lower mean global RNFL thickness (−12.3 µm; 95% confidence interval (CI) = −21.4, −3.22; p = 0.008) and temporal quadrant thickness (−15.5 µm, 95% CI = −22.1, −8.83, p < 0.001) than controls. Notably, non-ON MS eyes had lower temporal RNFL thickness than control eyes. (−11.9 µm, 95% CI = −18.60, −5.29, p < 0.001). History of ON was associated with lower global and temporal quadrant RNFL thickness compared to controls (−26.8 µm, 95% CI = −38.9, −14.8, p < 0.001; −25.3 µm, 95% CI = −33.1, −17.6, p < 0.001, respectively) and MS non-ON eyes (Tables 1 and 2). Longer disease duration was associated with thinner nerve fiber layer (1.5 µm per year, Table 2).
Multivariable models for RNFL thickness in pediatric MS eyes.
RNFL: retinal nerve fiber layer; CI: confidence interval; MS: multiple sclerosis; ON: optic neuritis; M/F: male/female.
Models for global and temporal thickness with mutual adjustment for all variables listed.
Coefficient is for 1 year older in age or 1 year longer in duration, respectively.
There was modest evidence of an interaction between sex and history of ON (p = 0.052). See Table 3 for stratified result of eyes with history of ON.
Within MS eyes, after adjustment for age, disease duration, and history of ON, male sex was associated with −9.4 µm thinner global RNFL when compared with females (95% CI = −17.4, −1.33, p = 0.022, Table 2). In control subjects, there was no association of male sex with mean global RNFL thickness (p = 0.70), and the mean RNFL in male eyes was 112 µm and in females 105 µm.
In temporal quadrant measurements of these pediatric subjects, sex was more strongly associated with RNFL thickness in eyes with history of ON (pinteraction = 0.052, Tables 2 and 3). In these eyes (n = 27, male eyes = 12), the boys had nearly 10-µm thinner values of temporal quadrant RNFL than girls (95% CI = −15.1, −4.15, p < 0.001). This difference represents 16.0% of mean observed temporal thickness. In control eyes, there was no significant difference in temporal RNFL thickness between boys (mean: 77.1 µm) and girls (mean: 75.8 µm).
Multivariable models of association of sex with RNFL thickness in eyes with history of ON.
RNFL: retinal nerve fiber layer; CI: confidence interval; ON: optic neuritis; M/F: male/female.
Models for eyes with history of ON (12/27 male eyes) with mutual adjustment for the three variables listed.
Macular volume and segmentation
It has recently been reported that healthy boys have larger macular volumes than girls. 19 In our MS cases, we observed that male sex was associated with larger total macular volumes (0.075 mm3, 95% CI = 0.008, 0.14, p = 0.028). In our smaller set of control subjects, we observed a similar point estimate for greater volumes in males (0.077 mm3, 95% CI = −0.021, 0.17, p = 0.13) although this did not reach statistical significance. Older age was associated with greater volumes in control eyes (0.018 mm3, 95% CI = 0.0068, 0.029, p = 0.002). Within MS eyes, both history of ON and disease duration were strongly associated (p < 0.001) with smaller total macular volume, after mutual adjustment and adjusting for age and sex (Table 4).
Multivariable models of association of ON with total macular volume and GCL metrics in MS eyes.
ON: optic neuritis; CI: confidence interval; MS: multiple sclerosis; mac.: macular; vol.: volume; GCL: ganglion cell layer; IPL: inner plexiform layer; M/F: male/female.
Models for macular metrics with mutual adjustment for the four variables listed: age, history of ON, sex and disease duration.
Macular volumes have been reported to be larger in healthy boys versus girls. 19
The segmentation software provides GCL volume estimates (from which GCL thickness can be calculated). In addition, we evaluated associations with combined GCL/IPL thickness given controversy in the field as to whether GCL can be reliably measured separate from the IPL. Although this was not a study designed to rigorously compare the two GCL outcomes, we found strong correlation between the GCL volumes generated by the software and GCL/IPL thickness (r2 = 0.995). Eyes with history of ON were associated with near 16.4% smaller GCL volumes (−0.069 mm3, 95% CI = −0.10, −0.039, p < 0.001) compared to non-ON MS eyes and 26% smaller GCL volumes compared to control eyes (−0.12 mm3, 95% CI = −0.16, −0.072, p < 0.001). Eyes of MS subjects without clinical history of ON had lower GCL volumes compared to control eyes (−0.036 mm3, 95% CI = −0.06, −0.011, p = 0.004). Disease duration was also associated with lower GCL volumes, and similar results were observed for the outcome of the combined GCL/IPL thickness (Table 4).
Sex was not associated with either GCL metric (Table 4) in analyses within MS patients. However, as macular volumes were larger in male control eyes than females, we also evaluated whether in the control eyes boys had larger GCL volumes or GCL/IPL thickness. While the analyses did not reach nominal statistical significance, the point estimates were consistent with this hypothesis and GCL volumes were 0.031 mm3 greater in male control eyes (95% CI = −0.001, 0.064, p = 0.061) and GCL/IPL thickness 3.36 µm greater (95% CI = −0.15, 6.88, p = 0.061) than in female control eyes.
INL and microcystic macular edema
As a previous pathological study in adults demonstrated damage to the INL in eyes of MS subjects 20 and more recently greater INL thicknesses have been reported to be associated with MS disease activity21,22 and history of ON, 23 we investigated associations of patient factors with the INL thickness. Age at scan was not associated with INL thickness (−0.014 µm, 95% CI = −0.25, 0.22, p = 0.90). Compared to control eyes with adjustment for sex, eyes of MS subjects had thinner INL (−11.1 µm, 95% CI = −13.1, −9.02, p < 0.001; Table 1).
Within MS eyes, the range of INL thickness was 26.7–44.9 µm, with the exception of one strong outlier, a 13-year-old male with no history of clinical ON, with very thin GCL (25.7 µm right eye, 18.2 µm left eye) whose INL thicknesses were 81.3 µm right eye and 79.2 µm left eye. This young man had evidence of substantial bilateral microcystic macular edema (MME; Figure 2). While this outlier is of interest as an extreme case of injury, he had to be removed for multivariable analyses within cases due to significant departure from normal distribution of the rest of the values. Disease duration (0.099 years, 95% CI = −0.31, 0.50, p = 0.63) was not associated with INL thickness. Even without the outlier mentioned above, male sex was associated with 2.21 µm greater INL thickness (95% CI = 0.42, 4.01, p = 0.015), but males also had greater INL thickness in control eyes (β = 3.38, 95% CI = −0.0024, 6.75, p = 0.050). ON was not associated with INL thickness (1.02 µm, 95% CI = −0.21, 2.25, p = 0.10), compared to non-ON MS eyes.
Microcystic macular edema (MME) in the inner nuclear layer (INL) of the retina of a boy with history of MS.
In addition to the two eyes from the boy described above, two additional male eyes from two different subjects (aged 15 and 16) had evidence of MME. Both had severe GCL thinning (20.3 µm and 19.3 µm), but only one had history of clinical ON. However, for 17 eyes studied while the macular segmentation was successful, the imaging was not sufficient to distinguish MME from blood vessels or other focal artifacts. Thus, other cases of MME may have been missed.
Discussion
In a uniquely large dataset of pediatric MS retinal imaging, we have demonstrated that substantial GCL loss occurs in children with or without history of ON. There was a trend for all tissue measurements to be less in MS non-ON eyes than control eyes, strongly suggesting that subclinical retinal injury occurs in pediatric-onset disease. We further report greater degree of retinal injury in boys when compared to girls.
Previous studies have suggested that pediatric subjects may have less brain tissue destruction or better repair4,5 and that the visual system may have better recovery following ON than in adults. 24 We herein report that these patients have similar decrease in the retinal GCL as in adult-onset MS, and thus, RGC neuronal preservation may not be the mechanism for the better visual recovery previously reported in children. Our result differs from a prior study in children, which reported no difference in GCL measurements between ON and non-ON eyes, 25 but in the prior work fewer eyes were analyzed, a mixed population of demyelinating disorders were included, and differences in GCL segmentation methodology may also have contributed.
We have demonstrated that as has been recently reported in adults, 13 male eyes may suffer more axonal injury than female eyes following ON. Our study is limited by cross-sectional design, but these results are consistent with the prior adult longitudinal data. 13 We did not see lower RNFL thicknesses in male control eyes and prior studies have also demonstrated that males tend to have equivalent RNFL thickness compared to females in the healthy state.26–29 The mechanism of the sex difference in axonal injury is not known, but may be related to either differences in immune-mediated injury or recovery from injury in males versus females. Sex steroid hormones are putative candidates to drive these sex-related differences, and an additional study is needed to know whether the observed differences depend on pubertal status.
We observed reduced RNFL thickness in boys compared to girls, but no sex differences in GCL or IPL thickness. The contrast of these results may be related to the recently reported and our observed larger macular volumes in males. 19 A lack of difference in a macular layer such as GCL may actually reflect a greater loss in males, because in the normal state the GCL should be larger in boys than girls. Our data in control eyes did demonstrate point estimates of greater GCL thickness or volume in control male eyes than female eyes.
It has been recently demonstrated that there may be changes in the INL in eyes with history of ON 23 and that increases in this layer may be associated with disease activity and microcystic changes to the retina.21,22 INL changes may be related to transynaptic injury. Microcystic changes in this layer likely contribute to increased thickness measurement and have been associated with optic nerve atrophy from multiple diseases including genetic, compressive, and inflammatory etiologies. 30 We observed thinner INL in MS eyes than control eyes. This result is consistent with previous results in adults with MS and overall macular thinning. 31 However, within MS eyes, we observed an extreme case of a boy with bilateral microcystic changes and enlarged INL, without known history of ON but with significant GCL thinning. Prior enlargement of the INL has been seen in adults with MME and history of ON. 22 An explanation for the differences in INL thinning versus thickening in MS patients may be that in the absence of MME there is thinning and in the presence there is apparent thicker measurement of this layer. We observed greater INL thickness in boys with MS and healthy controls. This observation may be related to the overall larger macular volumes in male subjects. Evaluations of associations of MME with INL thickness or ON were limited by few observations of these changes.
Other limitations of the study include the retrospective and cross-sectional design. Longitudinal studies will help to confirm that subclinical tissue loss occurs in non-ON eyes. A prospective study will provide the opportunity for rigorous structure–-function correlations. It is possible that in this young age group clinical episodes of less severe ON may have not been reported or were otherwise missed. In addition, the comparison to a reference population with other neurological diseases from the same base clinic while reducing the impact of unmeasured confounders has the capacity to lead to underestimation of the effect of sizes measured. From a technical perspective, while this study demonstrates feasibility of high-quality retinal segmentation in children, without prospectively acquired repeated measurement from the same patient we cannot formally calculate reproducibility measures. Finally, a potential confounding factor with associations of age and OCT metrics in children is axial length,19,27,29 which was not measured in this study.
Strengths of the study are the large sample size for a pediatric MS cohort seen shortly after disease onset, exclusion of eyes with history of acute ON in the last 6 months, and use of high-resolution OCT and standard segmentation of the macular layers in a well-defined pediatric population meeting published criteria for a single disease entity. In addition, rigorous multivariable modeling was applied, including adjustments for inter-eye correlations.
Larger, prospective studies are in progress to better characterize retinal changes in pediatric neuro-immunological diseases. Multicenter collaboration will provide the samples sizes needed to further understand retinal injury in pediatric MS and ON patients and differences from healthy control and adult populations.
Footnotes
Acknowledgements
J.G. contributed to the study design, data preparation, statistical analyses, drafting, and revising the manuscript; H.C., B.C., S.A., and H.Y. contributed to data acquisition and preparation, and revising manuscript; E.W. contributed to the study design, data interpretation, and revising the manuscript; A.G. contributed to the study design, data interpretation, oversight of data acquisition and preparation, and revising the manuscript.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Dr Graves was supported by the Foundation for the CMSC, NIH BIRCWH program, and the Race to Erase MS during this work. She has current grant support from Biogen and Genentech. Hardeep Chohan, Benjamin Cedars, Sam Arnow, and Hao Yiu have no financial support to disclose. Dr Waubant is funded by the National MS Society, the NIH, UCSF, and the Race to Erase MS. She is a site PI for clinical trials with Roche, Biogen Idec and Novartis. She volunteers on an advisory board for a clinical trial of Novartis. Dr Green reports personal fees from Inception Sciences and Mylan Pharmaceuticals. He has served on end point adjudication committees for Biogen and Medimmune. He serves on trial steering committees for Novartis and Scientific Advisory Board for Bionure. He has grants/awards from the National Multiple Sclerosis Society, Novartis, UCSF CTSI, and That Man May See as well as philanthropic support from the Rachleff Family and the Robert Dale Family.