Optic pathway glioma (OPG) is a childhood tumour of the visual pathways, sometimes associated with neurofibromatosis type 1. Despite being characterized as low-grade, OPGs can be difficult to manage, as they tend to follow a highly unpredictable clinical course, with poor correlation between conventional imaging features and visual loss ¹. As such, similarly appearing tumours can range from stable, asymptomatic lesions, to aggressive lesions with relentless progression resulting in loss of vision. As diffusion-weighted imaging (DWI) is now performed routinely in many hospitals, we investigated if DWI can help to predict visual acuity in OPG patients.

Data Acquisition

We collated longitudinal DWI and visual assessment data from OPG patients treated at our institution between 2009-2014. The exclusion criteria were patients with less than 3 concurrent imaging and visual assessment measurements, and patients in which visual acuity (VA) was not measured using the standardised logMAR chart ². The remaining cohort consisted of 11 patients (4 male), each with imaging and visual assessments acquired at 7 time points on average (range 3-15). The mean age (±SD) at first MRI was 4.0 ± 2.5 years, and the mean period of assessment was 2.4 ± 1.5 years. MRI was performed on a 1.5 T Siemens Avanto (Erlangen, Germany), and the DWI protocol consisted of a 2D EPI readout with b-values of 0 and 1000 s/mm2 and resolution of 1.8 mm in-plane (interpolated to 0.90 mm), with 5.0 mm thick slices.

Post Processing

The clinical T2w images acquired during each imaging session were used to draw a region of interest (ROI) over the entire tumour volume. These images were also used to measure the thickness of the optic nerves (t_ON), which was defined as the maximum measureable width of the optic nerve anterior to the optic chiasm. The ADC maps acquired in the same imaging session were registered to the T2w images to calculate the median ADC in the entire tumour ROI (ADC_tumour). Median ADC was also measured in each optic nerve (ADC_ON), by placing ROIs over the optic nerves in each ADC map (Figure 1).

Data Analysis

MRI and visual assessment data for all time points, in all patients, were pooled for analysis. A linear regression model was used to determine the relationship between VA in the eye with the highest logMAR score and ADC_tumour (higher logMAR scores indicate poorer vision). In addition, as VA was assessed independently in each eye, linear regression was used to model VA vs. both ADC_ON and t_ON for the corresponding eye. Patients for which no quantitative logMAR score could be measured, due to the patient either being too young or having lost vision completely (i.e. no perception of light), were excluded from this part of the analysis. Additionally, for eyes with no perception of light, ADC_ON and t_ON measurements were compared (group-wise) to eyes with measureable logMAR values, using an un-paired t test.

No correlation was found between VA (logMAR score) and ADC_tumour (N=42 measurements, p=0.85). However, a significant positive correlation was found between VA and both ADC_ON (p=0.00043) and t_ON (p=0.014) in the corresponding eye (N=80 measurements). The product of the two regressors (ADC_ON ·Ÿ t_ON) also produced a significant positive correlation with VA (p=0.0017, Figure 2a). For the group-wise comparison between eyes with no perception of light vs. eyes with measureable logMAR values, the ADC_ON ·Ÿ t_ON parameter produced the most significant difference between the two groups (p=4.2x10^-10, Figure 2b).Our results indicate that both increased ADC and thickness in the optic nerve predict poorer vision in the corresponding eye, and the product of these measurements provides a metric that correlates with logMAR assessment of vision, and is significantly higher in patients who have lost their vision. A previous study found higher ADC in the entire tumour volume may predict visual decline in OPG patients ³, however in this study we found increased ADC in the optic nerve is a better predictor of VA. Increased ADC and nerve thickness in this region is likely to be due to tumour invasion causing disruption of the nerve fibres, and previous studies have shown reduced fractional anisotropy (using DTI) in the optic radiations in OPG patients ⁴. However, ADC maps are more widely available than DTI data in many institutions, and our results indicate that the combined ADC_ON ·Ÿ t_ON parameter may provide a useful metric for identifying tumours which are likely to cause visual impairment.