Multifocal Oscillatory Potentials of the Human Retina

376 Kurtenbach and Jägle

213−1, resulting in a recording time of 3.7min split into 16 shorter sequences. The hexagon size was scaled with cone density for each stimulated area. For fixation, a dim red cross was presented at the center of the stimulus display.

The stimuli were presented on either an Iiyama or Sony color display at a refresh rate of 75 Hz. For black-white stimulation, each hexagon was alternated between white (81 or 100 cd/m2 depending on the study) and black (1.5 or 0.2 cd/m2) with an ambient room luminance of about 20 cd/m2. Stimuli aimed at isolating the response to either the L or M cones, using the silent substitution technique [5, 6], were calculated from the emission spectra of the monitor and the L- and M-cone spectral sensitivities [7]. The L-cone-isolating stimulus had an average luminance of 19.2 cd/m2, that of the M cone was 33.8 cd/m2. This corresponds to average quantal catches of approximately 4.46 log quanta/s/cone for the L-cone-isolating stimulus and 4.43 log quanta/s/cone for the M-cone-isolating stimulus, with a Michelson contrast of around 47% for both. To slow the stimulation frequency, three black frames were inserted between consecutive stimulus frames, producing a base interval for the pseudorandom stimulation of 53.33 ms.

The signal was recorded from both eyes simultaneously with DTL (Dawson-Trick-Litzkow) fiber electrodes (UniMed) that were positioned on the conjunctiva beneath the cornea. Reference and ground electrodes (Ag-AgCl) were attached to the ipsilateral temple and forehead, respectively. The signal was amplified using a Grass amplifier with a frequency bandpass of 100–1000 Hz. For recording, pupils were dilated to 8 mm or more with 0.5% tropicamide. Monitoring the raw signal controlled the quality of recordings, and segments contaminated by blink artifacts or saccades were discarded and rerecorded. Signals were analyzed after a single step of artifact rejection. Average response amplitudes (nV/deg−2) were calculated from retinal areas of equal eccentricity for the first-order and second-order first-slice kernel of the mfOPs. The first-order component is the mean response to all the “white” frames minus the mean response to all the “black” frames in the m-sequence, giving largely the linear response to the stimulation. However, interaction between flashes can also occur, which is considered by the second-order (first-slice) response component. This component is computed from the sum of responses to stimulation from two consecutive responses of the same sign (i.e., both black or both white), minus the sum of traces obtained when two consecutive hexagons have different signs. As we will see later, the first-order kernel from a normal subject typically shows two peaks at around 22 and 30ms, although a small additional potential at around 15ms is apparent in some cases. The first slice of the second-order kernel displays three potentials around 21, 27, and 32ms.

UNDERLYING MECHANISMS

The topography of the mfOP response, the response pattern over the retina, recorded from normal healthy subjects can provide information about the mechanisms underlying their formation. In Fig. 1 we show typical thee-dimensional representations of the topography of the firstand second-order (first-slice) kernel response amplitudes of mfOPs over the central 50° of the retina from a 27-year-old healthy subject. Striking is that the responses are by no means uniform throughout the retina. Both kernel analyses clearly show evidence of the blind spot, or optic nerve head (right of array), which lacks photoreceptors. A second major feature of these topographies is that there is no large peak in the central area, although the fovea has over ten times more cone photoreceptors than at

Fig. 1. Three-dimensional representation of the first-order (left) and second-order, firstslice (right) kernel amplitude analyses of mfOPs from a 27-year-old subject. The height of the plot represents the scalar product of the response waveform amplitude with a waveform template, representing the overall average of all the local responses, normalized to the areas of the stimulus element that generated it.

5° [8, 9]. This indicates that cone signals alone do not play a major role in the generation of mfOPs or that the signals are modified by other retinal processes.

The importance of postreceptoral activity in the generation of mfOPs has been shown in recordings from animals treated with pharmacological agents, which can elucidate cell interactions by blocking one or more of the neurotransmitters responsible for the propagation of the signal from one retinal cell type to another. When signals that arise in the inner retina of rabbits and rhesus monkey are selectively blocked, for example using the inhibitory neurotransmitters glycine or GABA (γ-aminobutyric acid), the oscillatory potentials extracted from photopic mfERGs (multifocal electroretinogram) are selectively diminished [10, 11]. This has also been shown for conventional fOP recordings from the mudpuppy retina [12], where, however, the early oscillatory potentials appear more sensitive to the treatment than the later. Treatment with pharmacological agents can also selectively block specific pathways in the retina. Cone circuitry involves two parallel pathways directly from the cone photoreceptor to the ganglion cell through the cone bipolar cell, one that hyperpolarizes to light stimuli (ON pathway) and one that depolarizes on stimulation (OFF pathway). In rhesus monkeys, the suppression of ON activity with APB (L-2-amino-4-phosphononbutyric acid) causes the selective extinction of almost all of the mfOPs, demonstrating that the ON pathway plays a crucial role in their generation [11]. There is evidence from clinical studies (see congenital stationary night blindness [CSNB] discussion), however, that the OFF pathway does contribute to some of the mfOP response in humans.

Although the amacrine cells of the inner retina are important for mfOP generation, only certain types among the 40 or so different morphological forms are important. Some are dependent on the neurotransmitter dopamine, but it has been demonstrated that this transmitter is not crucial for mfOP generation in humans: Patients with early

Age related changes

20 year old

First order

59 year old

kernel

Second order

kernel

0-2 deg

1.8-7 deg

5-13 deg

11-22 deg

17-30 deg

2 nV/deg2

10 msec

Fig. 2. Comparison of traces from a 20-year-old normal subject (dashed lines) with those of a 50-year-old normal subject (continuous lines), both grouped into rings of equal eccentricity (see upper right inset). The major waveform components from the older subject are generally smaller and are delayed compared to those of the younger subject.

Parkinson’s disease, who have reduced dopamine levels, have intact mfOPs extracted from mfERGs [13].

It is evident from Fig. 1 that in addition to depressed recordings from the fovea and optic nerve head, the responses are not uniform throughout the retina, and that there is a considerable amount of local variation in amplitude. In the first-order kernel analysis (left in Fig. 1), it will be seen that the maximum amplitude is found in a concentric ring between about 2° and 13° around the fovea (see also Fig. 2, left), indicating that both rods