Abstract
Purpose: :
The implicit time of the first order mfERG kernel (K1 IT) is a sensitive measure of retinal function, and predictive of retinopathy development, in diabetes. It has been proposed that prolonged K1 IT is in part due to an abnormality of the fast adaptive mechanisms reflected in the amplitude of the second order kernel (K2). Our purpose was to examine the relationships between K1 IT and the signal–to–noise ratio (SNR) of K2 in local retinal patches of diabetic patients.
Methods: :
K1 and K2 mfERGs were derived from 103 patches within the central 45 degrees of the retina in 20 diabetic patients without retinopathy in either eye, 6 patients with nonproliferative diabetic retinopathy and 30 normal subjects. The responses were recorded using 10 to 300Hz and then digitally low–pass filtered at 100Hz. The 103 responses were then averaged within 35 zones to improve their SNR. K1 IT and amplitude (AMP) were measured with a template stretching method (Hood & Li, 1997). K2 SNR was measured by first calculating the root mean square (RMS) amplitude of the signal epoch (10–70ms) and a noise epoch (120–180ms). Abnormalities were defined as values beyond the 95th percentile of the normal sample. Spearman's ranked correlations were used to analyze relationships.
Results: :
In patients without retinopathy, 277 (40%) of the zones had abnormal K1 IT, 88 (13%) had abnormal K1 AMP, and 110 (16%) had abnormal K2 SNR. In patients with retinopathy, 184 (88%) of the zones had abnormal K1 IT, 35 (17%) had abnormal K1 AMP, and 46 (22%) had abnormal K2 SNR. In both the with– and without–retinopathy groups there was a small but statistically significant relationship between overall K1 IT and K2 SNR (R2 = 0.24, N = 210 and R2 = 0.02, N = 700, respectively; both p < 0.001), with decreasing K2 SNR associated with increasing K1 IT. However, in zones with normal K1 AMP and abnormal K1 IT (N = 194; 28%), there was no relationship between K1 IT and K2 SNR (R2 = 0.018; p > 0.05) in patients without retinopathy. In patients with retinopathy there was a statistically significant relationship (R2 = 0.13; p < 0.001) but only 17% (N = 26) of the zones had abnormal K2 SNRs.
Conclusions: :
These results suggest that first order implicit time delays are independent of second order signal–to–noise ratios in diabetics without retinopathy when first order amplitudes are normal. It is not until after fundus signs of diabetic retinopathy are present that there is a statistically significant relationship between the two. Thus, it appears that abnormalities of fast adaptive processes may be associated with some but not all mfERG delays in diabetes.
Keywords: electrophysiology: clinical • diabetes • diabetic retinopathy