Introduction
The pattern electroretinogram (PERG) is used to record ganglion cell responses in a variety of clinical and research contexts [
1,
2]. This includes not only testing of macular function [
3] and early detection of glaucoma [
4,
5], but also assessment of drug effects [
6‐
9]. In cases of suspected malingering, the combination of a normal PERG and an altered visual evoked potential (VEP) response may provide decisive evidence for an organic disorder [
10]. Recently, the PERG has received increased interest as a potential diagnostic biomarker for psychiatric disorders [
11‐
15].
In contrast to the flash ERG, the PERG uses pattern reversal stimuli, which are particularly suitable to target retinal ganglion cells via local contrast inversion without change in mean luminance [
16]. With every reversal of the pattern stimulus, local luminance responses on the retina cancel out, and only nonlinear components remain, which constitute the PERG [
16]. The steady-state variant (> 10 reversals per second (rps)) of the PERG (ssPERG) is frequently used if the response magnitude, rather than the shape of the response curve, is of primary interest. It allows for efficient recording and relatively simple frequency-space response detection and statistical assessment [
17].
In several fields of application, accurate central fixation is not always guaranteed. For instance, patients with central visual field defects may rely on eccentric fixation [
18]. Studies in patients with psychiatric disorders (e.g., schizophrenia or depression) showed differences in the ability to maintain proper fixation compared to controls [
19‐
22], and in cases of malingering, patients may choose to fixate improperly.
Effects of improper fixation have previously been assessed for different types of electrophysiological exams, such as multifocal ERGs [
23,
24] and acuity VEPs [
25]. For the transient PERG, Persson and Wagner [
26] did not find a sizable effect with fixation at 4° eccentricity and check sizes of 24 arcmin.
The present study was designed to quantify the effects of different amounts of deviation from central fixation on the ssPERG in order to provide a basis for judging the relevance of fixation inaccuracies in clinical practice and research applications. We performed two experiments.
In Experiment 1, we assessed purely horizontal misfixation. In this case, as the gaze direction changed along the edge between checks, luminance changes at the time of the checkerboard reversals continued to be balanced, as a switch from black to white above the eccentric fixation point was compensated by a switch from white to black below the fixation point. However, the location of vertical edges in the stimulus changed on the retina and the stimulus pattern as a whole was displaced.
In Experiment 2, fixation was varied along the
diagonal. This resulted in the gaze being directed at points that were located in the interior of a check. Thus, when considering the vicinity of the fovea, the reversals of the stimulus resulted in locally unbalanced luminance changes. Due to the eccentricity dependence of retinal circuitry [
27], we expected this effect not to be fully balanced across retinal locations, potentially resulting in an undesired luminance response in the ssPERG. The imbalance should be more pronounced for larger checks.
Discussion
The present data show that moderate fixation inaccuracies, in the order of 7 degrees, have a nearly negligible effect on the ssPERG amplitude. In particular, when the check size is large and the gaze does not fall directly in the center of a check, ssPERG amplitudes remain robust, for both tested directions of misfixation. The exact dependence of ssPERG amplitude and fixation eccentricity will obviously depend on the overall extent of the stimulus area, as the position of the outer boundary of the stimulus area can be assumed to be the main determinant of ssPERG amplitude loss. This is consistent with previous findings showing that stimulation at higher eccentricities has a relatively lower contribution to the generation of the PERG [
33]. The amount of fixation inaccuracy that is tolerable in a given research context or in diagnostic use will depend on the expected effect sizes. The observed minor ssPERG amplitude alterations (< 10%) with moderate misfixation are comparable with normal PERG amplitude variations due to inter-session (coefficient of variation 6⎼16%) or diurnal (coefficient of variation ≈ 10%) variability [
34].
Comparing both directions of horizontal deviation, only minor differential effects were observed on the ssPERG responses from the different eyes and thus from nasal and temporal sides. Two previous studies however reported larger PERG amplitudes with nasal compared to temporal hemifield stimulation [
35,
36] and suggested that this might be due to the higher number of nasally distributed retinal ganglion cells in the peripheral retina [
37]. In our case, naso-temporal differences were rather small but seemed to be stimulus-specific. With temporal fixation deviation (rightward deviation of the right eye or leftward deviation of the left eye), the ssPERG signal decreases similarly with both check sizes when misfixation increases. With nasal fixation deviation, however, ssPERG responses to the finer checks seemed to be somewhat more affected with increasing fixation deviation, compared to the responses to the larger checks. Considering the complex relationship between PERG responses and the various stimulus parameters, including check size, a definite interpretation of this effect is beyond the present study.
An inequality of the responses to both checkerboard polarities was observed in Experiment 2, particularly if the interior of a large check was fixated. This is most likely a luminance effect that originates at least partly from cell types other than ganglion cells and may need to be considered when interpreting ssPERG findings in terms of ganglion cell function. As the respective frequency of 7.5 Hz was relatively close to the lower cut-off frequency of the bandpass filter of our set-up (5 Hz), the luminance responses were possibly somewhat attenuated.
Using a relatively large stimulus extent, as in the present study (30° × 30°), can be useful when fixation problems are expected (as implicated by Sakaue et al. [
38] and Junghardt et al. [
39]), if the purpose of the recording does not require a smaller stimulated area. We estimate that the present results would in principle also hold for the standard stimulus extent (15° mean width and height) with the acceptable angle of misfixation scaled correspondingly (e.g., up to around 3° misfixation).
The present study addressed the question of static fixation inaccuracies. Clearly, eye movements might have additional undesired effects on ssPERG signal quality, arising from the electroretinographic response to the moving retinal image [
40] and the intrusion of electrooculographic artifacts [
41]. However, standard threshold-based artifact detection combined with frequency-domain response analysis should normally ensure that eye movements do not have a sizable effect on the test outcome.
In summary, the present study suggests that moderate fixation inaccuracies do not have a major impact on the outcome of ssPERG recordings.
Acknowledgements
The study was supported by the Deutsche Forschungsgemeinschaft (project# 462923710). Furthermore, we would like to thank Verena Gauggel for her support with data collection. Kathrin Nickel is funded by the Berta-Ottenstein-Program for Advanced Clinician Scientists, Faculty of Medicine, University of Freiburg.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.