Edinburgh Imaging

PhD projects 2019 003

Multi-modal retinal scanning for diagnostic & therapeutic biomarker discovery in neurodegenerative disease.


Purpose: Investigate changes in retinal structure and function caused by multiple sclerosis (MS) using established and novel modes of image acquisition and analysis, in order to identify candidate biomarkers of disease. Use longitudinal analysis to determine the rate at which changes to the retinal anatomy occur and whether this rate-of-change alters throughout the MS disease course.

Method: Within the Anne Rowling Regenerative Neurology Clinic (Royal Infirmary of Edinburgh), 72 MS participants (53 female, 19 male, age range 20-79, median age 44) and 80 healthy volunteers (HV) (61 female, 19 male, age range 23–73, median age 38) received spectral domain optical coherence tomography (SD-OCT) and ultra-wide field scanning laser ophthalmoscope (UWF-SLO) scans at baseline. Within the MS group, 16 (9 right eyes and 10 left eyes) had at least one clinically diagnosed episode of optic neuritis (ON). Follow up scans were taken at approximately 3, 6 and 12-month intervals over the course of two years. Visual acuity (VA) as an indicator of retinal function was also measured at each visit using Sloan letter charts at 100% and 2.5% illumination. Through proprietary software, automatic segmentation of SD-OCT images quantified changes to retinal nerve fibre layer (RNFL) thickness and macular volume. Novel image processing algorithms specifically adapted to semi-automatically segment the vasculature in UWF-SLO images (using multi-scale filtering and supervised classification) allowed for measurement of vessel diameter. Data from the 3 participant groups (MS (no ON), MS (ON) and HV) was first tested statistically for significant differences. Linear regression modelling with the inclusion of disease, sex, length-of-disease and age as predictors was then used to estimate the true effects of disease on the retinal anatomy without confounders.

Results: Baseline data showed that individuals with MS (no ON) have a reduced RNFL thickness in all peripapillary regions and a reduced macular volume when compared with HVs. The reduction in RNFL thickness was seen globally in both eyes (OS: -10μm, OD: -11μm, p<0.001) but was greatest in the inferior temporal region (OS: -15μm, OD: -13μm, p<0.01). In the MS (ON) group the reduction in RNFL thickness was even greater, again being largest in the inferior temporal region (OS: -29μm, OD: -24μm). Linear regression models confirmed the effect of disease on RNFL thickness and macular volume with significant coefficients in all regions but the inferior nasal. Follow-up analysis showed that the MS (no ON) group had a larger annual rate-of-thinning of the RNFL than HVs in the majority of regions, the largest difference being in the inferior nasal (OS: -1.32μm/year, OD: -1.06μm/year, p<0.05) and superior nasal (OS: - 1.11μm/year, OD: -1.13μm/year, p<0.05) regions. Retinal arteries were shown to have a significantly reduced diameter in only the inferior nasal quadrant, where the greatest reduction in vessel width was found (OS: -0.005mm, OD: - 0.007mm, p<0.05). For mean venous width, the inferior nasal quadrant of the right eye also showed a statistically significant difference between the two groups (OS: -0.005mm, OD: - 0.005mm). Linear regression models were more frequently significant for arterial rather than venous data, despite both vessel types having significant disease coefficients in the inferior nasal quadrant. Follow-up data was inconclusive on the effect that disease has on the annual rate-of-change of vessel thinning. Median baseline high contrast VA (OS: -6, OD: -3, p< 0.05) and low contrast VA (OS: -6, OD: -6, p<0.01) scores were significantly reduced in the MS (no ON) group when compared to HVs. In this study, MS seemed responsible for a 1 line drop in both high contrast and low contrast VA scores in the MS (no ON) group. A significant difference between the HV group and MS (ON) group was only found in the median low contrast VA score of the left eye (OS: -21, OD: -4, p< 0.05). As with vessel analysis, follow-up data showed no significant rates-of-change in VA score.

Conclusions: At baseline, the RNFL and blood vessels of the retina were significantly thinner in the MS (no ON) group when compared to HVs in particular regions. Consistent with previous reports in the literature, both RNFL and macular volume showed a significant decrease in individuals with MS (no ON). However, this is the first study to show that both arteries and veins in the retinal periphery also have a reduced width in the inferior nasal quadrant. Such changes in the retinal anatomy correlate with a drop in visual function as measured by VA. An increased annual rate of RNFL thinning was also seen in participants with MS (no ON) indicating that anatomical changes occur over time rather than acutely, however this rate-of-change is largest early in the disease course. A similar pattern was observed with the rate at which macular volume decreases in individuals with MS (no ON). Whether changes to the retinal blood vessels occur over time was not determinable within this study, due to the effect size being too small for the number of participants recruited. Future studies should focus on replicating findings pertaining to retinal blood vessels and determining when in the MS disease course these changes occur.

  • Multiple sclerosis (MS)
  • Optic neuritis (ON)
  • Retinal anatomy
  • Retinal nerve fibre layer (RNFL)
  • Spectral domain optical coherence tomography (SD-OCT)
  • Ultra-wide field scanning laser ophthalmoscope (UWF-SLO)
  • Thomas Pearson
  • PhD
Link to University of Edinburgh holdings: