Collateral damage of radiation to normal tissues: What can the retina tell us?

Abstract

Purpose : Damage to normal tissue in an irradiated field (normal tissue injury, NTI) is a complication of radiation therapy believed to be caused by microvessel pathology. In the Collaborative Ocular Melanoma Study (COMS), vascular pathology was described in uveal melanoma (UM) patients 2 years post-radiation therapy. However, the COMS and other studies used subjective scoring systems, lacked a control group, did not assess neuronal function in detail, and used dosimetry methods simplified for clinical application.

Methods : Optical coherence tomography-angiography (OCT-A) was used to characterize vascular pathology by vessel density, fractal dimension (FD), lacunarity, foveal avascular zone (FAZ) size, parafoveal density (PVD), and by machine learning-derived density maps. OCT-derived retinal layer thickness and 10-2 visual field testing (VF) were used to quantify corresponding neuronal loss and dysfunction. Patient-specific 3D dosimetry models were constructed to spatially correlate dose to each outcome measure. Measurements were taken in both eyes of 31 UM patients at 0.5-150 mo post-¹²⁵I plaque brachytherapy, and 16 age-matched controls. Data were normalized to the contralateral eye to account for subject variation due to age or systemic disease.

Results : Abnormal vessel density, FD, lacunarity, FAZ size, and PVD were observed after 24 mo in 25 out of 31 subjects, whereas only 6 subjects showed any vascular pathology before 24 mo. Visual acuity was the least sensitive indicator of adverse outcome, with only 16% of the entire population showing a visual decline of two or more Snellen lines since diagnosis. Neuronal dysfunction by VF correlated strongly with most measures, especially vessel density, thinning of the superficial retina, and radiation dose (p<0.001). Abnormalities were observed only at doses above 20 Gy.

Conclusions : Using a series of quantitative approaches, radiation microvasculopathy was detected starting at 24 mo post-exposure in most patients. Strong correlations between loss of neuronal function, vascular pathology, and dose were observed. Prospective studies will help determine whether vascular pathology precedes neuronal loss.

This is a 2020 ARVO Annual Meeting abstract.