Discriminating early-stage diabetic retinopathy with subjective and objective perimetry

Abstract

Introduction: To prevent progression of early-stage diabetic retinopathy, we need functional tests that can distinguish multiple levels of neural damage before classical vasculopathy. To that end, we compared multifocal pupillographic objective perimetry (mfPOP), and two types of subjective automated perimetry (SAP), in persons with type 2 diabetes (PwT2D) with either no retinopathy (noDR) or mild to-moderate non-proliferative retinopathy (mmDR).

Methods: Both eyes were assessed by two mfPOP test methods that present stimuli within either the central ±15° (OFA15) or ±30° (OFA30), each producing per-region sensitivities and response delays. The SAP tests were 24-2 Short Wavelength Automated Perimetry and 24-2 Matrix perimetry.

Results: Five of eight mfPOP global indices were significantly different between noDR and mmDR eyes, but none of the equivalent measures differed for SAP. Per-region mfPOP identified significant hypersensitivity and longer delays in the peripheral visual field, verifying earlier findings. Diagnostic power for discrimination of noDR vs. mmDR, and normal controls vs. PwT2D, was much higher for mfPOP than SAP. The mfPOP per-region delays provided the best discrimination. The presence of localized rather than global changes in delay ruled out iris neuropathy as a major factor.

Discussion: mfPOP response delays may provide new surrogate endpoints for studies of interventions for early-stage diabetic eye damage.

Keywords: diabetic retinopathy; multifocal; multifocal methods; objective perimetry; subjective perimetry; type 2 diabetes.

Figure 1

(A) Low luminance contours of the OFA30 stimuli showing their slightly overlapping five rings. (B) The left and right halves of rings 1,3,5 and 2,4 to show the luminance balancing of the test stimuli. (C) OFA response data, sensitivity, and delay TDs and PDs were mapped onto a 30-2 pattern with and extra four central regions to create six rings as a function of eccentric (color calibration bar at right). (D) Is similar to B but shows stimuli of the OFA15 method illustrating its test regions are scaled by a factor of 0.5 compared with OFA30.

Figure 2

The mean sensitivity and delay TD data for OFA15 (A-D) and OFA30 (E-H). The means were computed at each 30-2 field location across the ETDRS 10 eyes (noDR, A, C, E, G), and ETDRS 34 and 45 eyes (mmDR, B, D, F, H). From the top down, the rows alternate: sensitivities, delays, sensitivities, delays (n.b. the calibration bar units of dB and ms). Before taking the means, right-eye data were flipped to left eye format; hence, the figures have the nasal field on the right. The four small central locations are magnified and presented at the bottom left of each panel. Generally, mmDR eyes (B, D, F, H) showed more severe changes than those with noDR. Peripheral damage appeared to be more evident for OFA30 fields.

Figure 3

Areas under receiver operating characteristic (AUROCs) expressed as percentages for different comparisons of Total Deviations outputs. Blue bars are for delay data, and yellow are for sensitivity data (legend). AUROC of 50% represents chance classification, and an AUC of 100% represents perfect discrimination. The x-axis labels of each plot give the test: SWAP 24-2, Matrix 24-2, OFA15, or OFA30 for Sensitivities (Sens) or Times-to-peak (Delay). The leftmost four comparisons were for discriminating ETDRS 10 eyes from the Mild to Moderate NPDR eyes (ETDRS 34 and 45): (noDR cf mmDR). The rightmost four comparisons were between normal control eyes (Cont) and ETDRS 34 and 45 eyes (mmDR), or Control cf. ETDRS 10 (noDR). The leftmost pair of yellow and blue bars give AUROCs for SWAP and Matrix 24-2 tests. The SAP versus OFA15 (A) and OFA30 (B) methods performed similarly, with delays performing better except for the (Cont cf noDR) comparison where sensitivity was better. The error bars represent SEM.