Comparison of Fast and Slow Flash Multifocal Electroretinogram Techniques: Prediction of Nonproliferative Diabetic Retinopathy Development

Abstract

Abstract: : Purpose: Both the "standard" multifocal electroretinogram (s–mfERG) and the slow flash mfERG (sf–mfERG) can detect local retinal dysfunction in diabetic subjects, both with and without nonproliferative diabetic retinopathy (NPDR). Whereas abnormal s–mfERG implicit times are predictive of local NPDR development, it is not known whether sf–mfERG implicit times are predictive. In this study we compare the two techniques. Methods: S–mfERGs and sf–mfERGs were recorded from the central 45 deg of one eye of each of 12 diabetics with mild NPDR and 25 control subjects. S–mfERGs were evoked by 200 cd/m² pseudorandom flashes presented at the 75 Hz frame rate, and sf–mfERGs were evoked by 100 cd/m² pseudorandom flashes separated by 3 dark video frames. At each of the 103 stimulated locations, implicit times were measured using a template stretching method and Z–scores were calculated based on the control data. Fundus photos were taken within 2 months of the baseline mfERG testing and at 1–year follow–up. The tested retinal area was divided into 35 contiguous zones. Each retinal zone was assigned the maximum baseline implicit time Z–score within it, its baseline NPDR status, and its follow–up NPDR status. A Z–score > 2 was considered abnormal. Results: The diabetics' local s–mfERG and sf–mfERG implicit time Z–scores were correlated (R² = 0.53; P<0.001). However, their local sf–mfERG implicit times were more abnormal than their s–mfERG implicit times (P<0.001), and more implicit times were abnormal for the sf–mfERG (48%) than for the s–mfERG (36%). At baseline, 51 out of the 420 total zones (12%) had NPDR. At follow–up, NPDR remained in 21 zones, it resolved in 28 zones, lesion type changed in 2 zones, and NPDR developed in 49 new zones. For both types of mfERG, zones that remained free of NPDR or where NPDR resolved were more normal at baseline than zones that developed NPDR at follow–up (P<0.001). Baseline s–mfERGs were abnormal in 86% of the zones that subsequently developed NPDR and in 42% of the zones that remained free of NPDR (odds ratio = 8.4; P<0.001). Baseline sf–mfERGs were abnormal in 90% of the zones that developed NPDR and in 58% of the zones that remained NPDR–free (odds ratio = 6.3; P<0.001). Conclusions: The implicit times of both the s–mfERG and the sf–mfERG are predictive of local NPDR development one year later, with the s–mfERG having a slightly greater odds ratio. Retinal areas that remain free of NPDR, and areas where NPDR resolves, function more normally at baseline than areas where retinopathy develops over the following year.

Keywords: electroretinography: clinical • diabetic retinopathy • diabetes