A method for comparing psychophysical and multifocal electroretinographic increment thresholds

The multifocal electroretinogram (mfERG) has been commonly used as a method for obtaining objective visual fields. Although qualitative comparisons have been good, quantitative comparisons between the results from mfERG and the results from Humphrey Visual Field Analyser (HVFA) have found variable degrees of agreement depending upon the mfERG response parameter examined and/or the disease studied. Lack of agreement may be due to differences in methodology, differences in the sites of response generation, and/or differences derived from comparing suprathreshold versus threshold responses. In addition, the two procedures are performed at different levels of adaptation. We developed an approach for matching stimulus parameters and compared mfERG and psychophysical thresholds to assess the effects of technique and level of adaptation on the two responses. Psychophysical and mfERG thresholds were obtained as a function of the adaptation level (1.5-4.0 log td) and retinal location. The derived increment threshold-versus-intensity functions for both measures were fitted using the equation logT=logT(0)+log((A+A(0))/A(0))(n). We found that the values of A(0) for the mfERG data were one log unit higher than those for the psychophysical data. In addition, the value of the slope (n) for the mfERG data was shallower (0.8) than that of the psychophysical data (1.0). Predictions were made about comparisons of HVFA threshold and mfERG amplitude data in patients with retinal disease based upon a two-site model of adaptation. The data for some groups of patients could be best-fitted with a model of a disease acting at a site distal to all gain changes, whereas data from other patients were best fitted with a model of a disease acting at a site proximal to all retinal gain. The relationship between the Humphrey visual field threshold losses and mfERG amplitude reductions depends upon the site and mechanism of a particular disease process and the model of retinal gain assumed. In no case is a one-to-one relationship between the losses in the two measures predicted.