Towards an objective structural grading scheme of foveal hypoplasia in human albinism

Abstract

Purpose : Human albinism is characterized by foveal hypoplasia (FH). FH severity is often subjectively graded using a categorical scale based on the presence/absence of inner retinal layer extrusion, outer segment (OS) elongation, and outer nuclear layer (ONL) thickening. Here we quantify each of these structural features to develop objective criteria for FH grading.

Methods : Genetic testing and optical coherence tomography (OCT) imaging were performed in 67 subjects with albinism (33 M, 34 F, ages 5-69 years). OCT line scans were acquired at the fovea and graded using a previously published FH grading scheme.^a The ratio of maximum OS length to parafoveal OS length (OS ratio) was measured using longitudinal reflectivity profiles,^b,c and the foveal ganglion cell-inner plexiform layer (GCIPL) thickness was measured. The ratio of foveal:parafoveal ONL thickness (ONL ratio) was also measured in 22 subjects using directional OCT scans.^d Receiver operating characteristic (ROC) analysis was used to determine the optimal cutoff value of GCIPL thickness and OS ratio to differentiate between grade 1, 2, and 3-4. FH grade was reassessed objectively using these cutoff values.

Results : Subjects had oculocutaneous albinism type 1 (OCA1, n=25), OCA2 (n=13), OCA4 (n=2), OCA6 (n=1), ocular albinism (n=2), or Hermansky-Pudlak syndrome (n=2). Twenty-two subjects had absent, inconclusive, or pending genetic testing. Based on subjective FH grading, 29 subjects had grade 1, 10 had grade 2, 23 had grade 3, and 5 had grade 4. Foveal GCIPL thickness ranged from 24.95-81.49 µm (mean±SD: 66.74±12.09), OS ratio ranged from 1.13-2.10 (mean±SD: 1.51±0.24), and ONL ratio ranged from 1.08-2.88 (mean±SD: 1.52±0.44). The area under the ROC curve (AUC) for GCIPL thickness was 0.84 (p<0.0001) and the optimal cutoff was 71.95 µm. The ROC AUC for OS ratio was 0.81 (p<0.0001) and the optimal cutoff was 1.435. Using these cutoff values, objective FH grades agreed with subjective grades in 48/67 subjects.

Conclusions : Objective FH grading does not always agree with subjective grading, as quantitative foveal metrics vary widely in albinism and overlap between subjective FH grades. This indicates that the chronological development of foveal anatomical specializations may be more nuanced than the existing categorical grading scheme suggests.

^aPMC5648335
^bPMC2838989
^cPMC5186335
^dPMC6219160

This is a 2020 ARVO Annual Meeting abstract.