Evaluación de la capacidad diagnóstica del software de polo posterior de Spectralis SD-OCT con la cuadrícula 8x8 inclinada a 7º y horizontalizada en ojos glaucomatosos

Abstract

Introduction: The Spectralis SD-OCT contains a posterior pole software (PP) that provides the total value of retinal thickness in an 8x8 grid (with 64 cubes or macular cells) and allows independent quantification of all retinal layers in the macula. Purpose: To evaluate the diagnostic ability of Spectralis SD-OCT PP software, with the 8x8 grid tilted at 7º and horizontalized, to distinguish between healthy eyes and those with primary open-angle glaucoma (POAG). Methods: This is an observational, cross-sectional, retrospective, and descriptive study. A total of 299 eyes were studied (136 of 81 healthy subjects and 163 of 104 with POAG). The following retinal layers and retinal layer groupings were studied mRNFL, mGCL, mIRL, mORL, mGCLIPL, the entire macular retina, the macular ganglion cell complex (mRNFL + mGCL + mIPL) and mOPL complex + mONL of healthy and glaucomatous subjects. The analysis data have been obtained from each eye and in each layer of the 64-cell (8x8) thickness maps of the Spectralis SD-OCT PP pattern automatically aligned with the disc-fovea axis at 7°. Subsequently, all grids were horizontalised for each eye and the 64-cell thickness values were obtained again in the same way. The diagnostic ability of the thickness of each of the layers has been evaluated by the value of the area under the receiver operating characteristic curve (AUROC) and represented in heat maps. The thickness of the cells with ≥0.70 AUROC values was combined in each inclination to obtain a global index for each of the grids, inclined at 7º and horizontalised. A cut-off point for classifying healthy or sick patients was established. The diagnostic validity indices and 95% confidence intervals were computed for each of the layers. The statistically significant differences considered are those with p < 0.05. Results: The heat map patterns and the results of the comparison of the global index have shown a similar diagnostic ability of the studied segmentations when considering the macular cube 8X8 inclined at 7° or horizontalised. The segmentation with the highest diagnostic yield is mRNFL with an AUROC (95% CI) of 0,796 (0,746 – 0,845), followed respectively from the highest to the lowest by macular ganglion cell complex 0,785 (0,734 – 0,835), mGCL 0,784 (0,732 – 0,836), mGCLIPL 0,770 (0,717 – 0,824), mIRL 0,755 (0,701 – 0,809) and macular retina as a whole 0,752 (0,698 – 0,806). Conclusions: The 8x8 macular cube of the posterior pole software in Spectralis SD-OCT offers good diagnostic ability in different layers and combinations of layers of the inner retina as a complementary method, to other measurements, for POAG diagnosis.