This study demonstrates that most children with optic pathway gliomas and decreased RNFL thickness have abnormal VA, visual field loss or both. Some children with decreased RNFL thickness had normal high-contrast VA, but abnormal visual fields. Interestingly, other children were found to have normal VA and normal visual fields despite a significantly decreased RNFL—possibly suggesting that visual acuity or visual field loss may not have been detected with current techniques or the child has yet to manifest symptoms. On the other hand, if our vision testing was accurate, a normal RNFL thickness shows a strong relation to normal VA and visual fields. Therefore, in children who are not cooperative enough to provide a reliable sensory vision examination, a normal RFNL thickness might be reassuring to the clinician that the patient likely has a normal VA and visual field. In contrast, the finding of a reduced RNFL thickness may or may not have a strong relation to clinical visual sensory function and thus should not be taken as absolute evidence of clinically significant visual sensory dysfunction.
Our findings are similar to other studies that use OCT to correlate RNFL axonal loss with visual function.8–12
Studies of patients with optic neuritis and multiple sclerosis have demonstrated a strong association between reduced RNFL thickness and reduction of high and low contrast visual acuity.10,12
VA loss and VF loss associated with RNFL thinning in multiple sclerosis are indicators of axonal loss in the anterior (pregeniculate) visual pathway (i.e., optic nerve, optic chiasm or optic tract).11,12
Danesh-Meyer and colleagues8
have demonstrated a correlation between RNFL thickness and visual field deficits in adults with tumors compressing the optic chiasm.
Comparable to studies of optic neuritis in multiple sclerosis,10,12,14
we found that low-contrast letter acuity measures had a strong relation with RNFL using the identical OCT acquisition protocol. Although not formally analyzed, we frequently noticed subtle inter-eye differences (i.e., 0.1 logMAR) in high-contrast VA testing, but discovered more robust differences using low-contrast letter acuity. One possible explanation is that our high-contrast VA protocol using the EVA can overestimate visual acuity, especially in eyes with decreased visual acuity, as compared to ETDRS letters in the clinic.18
Chang and colleagues have found that children with optic pathway gliomas and normal visual acuity had normal grating acuity but impaired contrast sensitivity using sweep visual evoked potentials.19
These findings suggest low contrast acuity measures, as in multiple sclerosis, may be more sensitive to visual pathway damage as compared to high contrast VA measures. Comparison of the discrimination ability between high-contrast VA and low-contrast letter acuity requires further study.
Visual evoked potentials (VEP), proposed as an objective measure of visual pathway integrity, have been used in an attempt to detect vision loss20,21
or to correlate VEP with visual acuity22
in children with optic pathway gliomas. The accuracy and utility of VEPs are limited since most optic pathway gliomas involve the optic chiasm, eliminating the ability to make meaningful inter-eye comparisons as the visual pathway of both eyes are affected. Also, very subtle changes (i.e., < 5%) in VEP results can profoundly alter the sensitivity of the test and suggests that VEP is an inadequate management tool in this population.23
An expert review has determined that insufficient evidence exists to use VEP for the detection or progression of visual loss in children with optic pathway gliomas.3
Our cross-sectional study design contains some degree of selection bias because we obtained a convenience sample and this may limit our interpretation of the relation between RNFL thickness and vision in optic pathway gliomas. However, one author (RAA) followed a pre-established study protocol for VA and low-contrast letter acuity testing to minimize the possibility that the examiner’s knowledge of clinical diagnosis could influence the patient’s performance on the tests of visual function.15
Also, without knowing the within subject change in RNFL over time, we cannot exclude other factors contributing to our findings, such as the effect of chemotherapy. Longitudinal within subject measures of RNFL before and after vision loss are needed to better establish this relation.
Our secondary analysis that included data from NF1 patients without optic pathway gliomas revealed nearly identical results as the primary analysis. This secondary analysis was performed since children with NF1 frequently have cognitive delays and attention-deficit hyperactivity disorder that could influence their ability to accurately complete the high contrast VA and low-contrast letter acuity tests. Also, children with NF1 demonstrate brain MRI findings such as tortuous optic nerves that are not formally considered an optic pathway glioma, but could conceivably influence both VA and RNFL. If RNFL is to be used as a structural marker or screening tool for NF1 associated optic pathway glioma related vision loss in the future, NF1 children without optic pathway gliomas will undoubtedly be part of the screening population.
The interpretation of the RNFL values is problematic since normative data for children have not been established and our study relied on our own control subjects. We observed a wide range of RNFL, high-contrast VA and low-contrast letter acuity values in children with and without abnormal visual acuity and visual fields. This may be a result of performance variability, especially in children with NF1 and developmental delay. It is also possible that some subjects with decreased visual acuity but normal RNFL may have unrecognized amblyopia. We performed an undilated refraction to determine best corrected visual acuity rather than a cycloplegic refraction. However, if our refraction was incorrect, thus causing the patient to have worse VA, this error would only weaken, rather than strengthen the observed association between VA and RNFL. However, this is unlikely since nearly all children in this cohort benefit from a thorough neuro-ophthalmologic exam on a regular basis by one investigator (GTL), reducing the chance of overlooking a new and clinically meaningful refractive error.
Six patient eyes with normal VA and normal visual fields were found to have decreased RNFL, all of which had fellow eyes with vision loss from an optic pathway glioma affecting the optic chiasm. Two of these patient eyes were myopic, which is known to be associated with a lower RNFL.24
Also, it is possible that some of these patients had an abnormal visual field that was not detected because of poor patient cooperation.
Our interpretation of the relation between RNFL and high-contrast VA is further complicated by the finding of normal high-contrast VA in children with abnormal visual fields. In most cases, this can be explained by attributing the visual field loss to chiasmal or optic tract damage. Interestingly, all but one subject with normal VA and abnormal visual field had an RNFL below 80 microns (see ). This suggests that VA might not be the most sensitive measure in children with optic pathway gliomas, especially in younger children where VA and VF testing is difficult due to patient cooperation.
This study demonstrates that most children with an optic pathway glioma and decreased RNFL thickness have abnormal VA or visual fields from their tumors. This study suggests RNFL thickness is a rapid, non-invasive, objective quantitative measure of visual pathway integrity in children with optic pathway gliomas. If our results can be replicated in younger children, who are at the greatest risk for vision loss from their optic pathway gliomas (i.e., less than six years old), then RNFL thickness may be useful as a marker of VA in patients who are unable to cooperate with standard VA testing. OCT performed while the child is sedated, could possibly be used in this young population. Future longitudinal studies comparing serial measures of RNFL thickness need to examine whether decreasing RNFL might precede VA loss and to determine if OCT could possibly augment the clinical management of children with optic pathway gliomas.