The recommended standard recording parameters of the ISCEV (14) for clinical PERG assessment have been adapted for ophthalmologic patients. However, it has not yet been investigated whether they are equally suitable for PERG recordings in psychiatric patients.

The aim of the current study was (1) to replicate the findings of reduced PERG responses in patients with depression (9, 10) in an independent sample and (2) to improve the PERG protocols for this patient group with a specific focus on check size and stimulus frequency.

The study was approved by the ethics committee of the University Medical Center Freiburg (Approval ID: 93/04) and was conducted in accordance with the Declaration of Helsinki. All participants gave their written informed consent. Patients were recruited at the Department of Psychiatry and Psychotherapy, University of Freiburg. The diagnosis of a major depressive episode was established by an experienced specialist in psychiatry according to DSM-5 criteria. A depressive episode in the context of bipolar disorder, the presence of psychotic symptoms, and comorbid alcohol abuse were defined as exclusion criteria. Initially, 17 patients with a diagnosis of major depressive disorder (MDD) were recruited. PERG measurement was performed within the first few weeks after starting antidepressant medication, without clinical response. The intake of neuroleptics, methylphenidate, or the antidepressant bupropion were defined as exclusion criteria.

In addition, 17 healthy controls without current or a history of psychiatric or neurological diseases were recruited. They had to score within the normal range of the Beck Depression Inventory [BDI; (29)] and the Hamilton Depression Rating Scale [HDRS; (30)]. The matching procedure controlled for effects of sex and age.

Exclusion criteria for both groups were defined as an age > 65 years, the presence of neurological or ophthalmological diseases or an uncorrectable low visual acuity (< 0.8).

The following questionnaires were collected from both patients and control participants: the Beck Depression Inventory [BDI; (29)] to assess the severity of depressive symptoms and the Wender-Utah Rating Scale [WURS-k; (31)] for ADHD symptoms in childhood. In addition, the Hamilton Depression Rating Scale [HDRS; (30)] was applied as third-party assessment questionnaire.

Before examination, visual acuity of each participant was assessed monocularly with the Freiburg Visual Acuity and Contrast Test [FrACT; (32)] and, if necessary, corrected with refraction. A minimum of 0.8 decimal visual acuity was required for each eye.

DTL (Dawson, Trick, and Litzkow)-like electrodes (33), placed at the lower limbus of each eye were used for PERG recordings. Gold-cup electrodes positioned at each ipsilateral eye-canthus served as reference, an ear-clip as ground.

The EP2000 system was used for stimulation and initial data collection (https://michaelbach.de/sci/stim/ep2000/index.html; retrieval date 15.10.2021). Pattern stimuli were presented at an observer distance of 57 cm on a CRT monitor with 75 Hz frame rate in 800 x 600 pixel resolution, covering a field size of 32° × 27°. Symmetrical black/white reversal checkerboards with a mean luminance of 45 cd/m² and a Michelson contrast of 80% served as pattern stimuli. Four different check sizes (0.8°, 1.6°, 3.2°, 16°) were presented using two different reversal frequencies (12.5 and 18.75 rps) in the steady state range. Every check size was presented for a duration of 10 sweeps with a constant sweep length for both frequencies (960 ms), starting with the lower reversal rate, followed by the higher one. Blocks for the different pattern sizes were shown in ascending manner (stepwise increasing check size: 0.8°, 1.6°, 3.4°, 16°). This sequence was repeated in 10 equal cycles, with a short break in between. Responses exceeding a threshold of 120 μV were automatically rejected as artifacts. A minimum of 100 artifact free sweeps were recorded per condition and submitted to stimulus-synchronized averaging.

First off-line data processing was performed in Igor Pro 7 (Wave Metrics) with the “EP2000” module. To eliminate mains hum artifacts, averaged response traces were digitally low pass filtered (40 Hz). A Fourier analysis was performed after any linear trend (e.g., due to baseline drifts) had been removed (34). PERG amplitudes were extracted from the Fourier spectra at the respective stimulation frequencies (12.5 and 18.75 Hz) and noise-corrected [as described in (34)]. The average magnitude from the direct adjacent frequencies served as noise estimate (35). Additionally, phases were extracted from the Fourier transformation.

Statistical analysis was carried out in “R” (36) with RStudio (37) using the “tidyverse” package (38) for data handling. For the 8 stimulus conditions (4 check sizes and 2 frequencies) PERG amplitudes from both eyes were averaged for every participant separately. Psychometric data comparisons and initial testing for differences between the groups or stimulation parameters were established using Wilcoxon rank sum tests [“rstatix” package (39)]. Response times (in ms) were calculated from the extracted phase values (40). The fully crossed factorial design was analyzed with a mixed analysis of variance (ANOVA) for repeated measures [“afex” package (41)]. The factors group, check size and stimulation frequency, as well as their interactions, were evaluated for their impact on PERG amplitudes or response times. The factors check size and frequency were considered as repeated measures factors for each subject. Post-hoc analysis was limited to group comparisons [“emmeans” package (42)] with equal variance assumed. Hedge corrected (43) Cohen's d was calculated as effect size estimation for unpaired samples [“rstatix” package (39)]. Significance levels were determined by applying the Bonferroni-Holm procedure for a familywise α of 0.05 (44).

Of the originally measured 17 patients, five had to be excluded. The reasons for exclusion were intolerance of the electrodes, the intake of neuroleptic medication, regular somatic medication, subsequently diagnosed psychiatric comorbidity and substance abuse. Finally, 12 patients between 19 and 51 years of age could be included in the final analysis. Four patients suffered from a first severe depressive episode, while 8 patients had a recurrent severe depressive episode. Of the 17 control participants measured, 12 were matched by sex and age to the included patients and considered in the final analysis. The psychometric data of the patient and the control groups are presented in Table 2.

Overall, we discerned smaller PERG amplitudes in patients with MDD compared to matched healthy control participants, which replicates earlier findings with an independent sample of patients and controls (9). After correction for multiple testing, statistically reliable reductions in the PERG in MDD were only indicated with the smallest check size (0.8°).

This effect was not only present for the lower (12.5 rps), as previously reported (9, 10), but also for the higher stimulation frequency (18.75 rps) with a comparable signal-to-noise ratio for both frequencies. A follow-up study should address, whether a higher frequency can provide the opportunity to reduce total recording time. A shorter measurement time would be particularly advantageous for psychiatric patients with depressive symptoms and limited ability to uphold attention.

The higher rate for checkerboard reversals (18.75 rps) is additionally beneficial by fine-tuning group differences between MDD and control subjects through check size dependent modulations, which allow for a normalization of PERG responses via the calculation of check size ratios, similar to those used for the detection of early glaucoma (17).

Besides, our observations are not in line with the results of Fam et al. (24), who reported normal PERG contrast gain in MDD patients applying a stimulation frequency of 12 rps and the check size of 0.8°. This could possibly be due to differences in the technical implementation of the measurement protocol.

Since the PERG amplitude reduction in patients with MDD was only significant with the smallest check size (0.8°), we cannot recommend a recording paradigm which uses exclusively larger check sizes (1.6°, 3.2°, or 16°) in the context of psychiatric disorders, which would render the correction of refractive anomalies unnecessary.

Considering the dopamine-dependent regulation of the receptive field sizes in the retina (48) and the assumption of a disturbed dopamine homeostasis in MDD (49, 50), check size specific PERG alterations in MDD patients seem convincing. Particularly, dopamine is known for its modulatory role in the light adaptation of the retina, favoring daylight vision with high acuity, a mechanism provided by the decoupling of horizontal cells in the retina, thereby shrinking the antagonistic surround structures of the receptive fields (48). A disturbed dopamine homeostasis probably results in alterations in the PERG signals in response to different check sizes as it was similarly described for patients suffering from Parkinson's disease (4, 5).

Moreover, shortened response times were observed in patients with MDD for all stimulus conditions. As described in transient stimulations (51), the effect of a shorter response time with larger check sizes is observed in both groups. In patients with glaucoma, not only reduced amplitudes but also shorter response times have been similarly described by Bode and colleagues (40). At this point, the authors discuss an effect observed by Viswanathan et al. (52) that leads to a shortening of the P50 peak time when N95 is eliminated. Whether such a differential change between the N95 and the P50, which both contribute to the PERG signal, occurs in MDD and can explain the observed changes in the steady-state response time would need to be addressed in future studies.

While it would have been useful to find strong effects of depression with the very large check sizes, which would have obviated refraction, one can turn the relative constancy of these amplitudes into our favor by using them for individual normalization. Inspired by the PERG ratio protocol in early glaucoma (16, 17, 45), we compared the standardized PERG amplitudes, i.e., the amplitude ratio over the two check sizes (0.8° and 16°), between groups for both stimulation frequencies. The advantage of this “PERG ratio” approach is that it reduces inter-individual variability. We observed a significantly reduced PERG ratio in MDD patients compared to healthy controls for the high stimulation frequency (18.75 rps), but not for the low stimulation frequency (12.5 rps). This alternative analysis approach is promising since it increases interpersonal comparability and statistical power, which is particularly important for an objective biomarker. Moreover, higher stimulation frequencies could possibly reduce the time required for recording, which should be addressed in a follow-up study.

The present study provides promising perspectives for the optimization of PERG recording procedures in psychiatric settings. However, some limitations have to be mentioned.

Due to the small number of patients, the results of the current study must be considered preliminary. Follow-up studies with larger samples sizes could yield further information about the adaptation of stimulation frequency and check size for PERG recordings in psychiatric patients. It should be noted, however, that the PERG ratio is feasible to minimize inter-individual differences.

At the time of measurement, patients had already been taking antidepressant medication for a few days or weeks. In a previous study by Bubl et al. (9), however, a reduction in contrast gain was detected in both medicated and un-medicated depressed patients. Another limitation is that smoking status was not considered as a matching factor between patients and controls. This could also have a confounding effect on results, as it could have an impact on dopamine neurotransmission (53). Lastly, the MDD group also exhibited elevated ADHD symptoms in childhood according to the WURS-k questionnaire compared to the control group. Since the PERG amplitudes from patients suffering from ADHD have been reported to be unaffected (54), we regard influences from ADHD symptoms as rather unlikely. Particularly since the so-called PERG noise, which has been shown to be elevated in ADHD patients (55), was not affected in our MDD patients (p = 0.94, one-sided Wilcoxon test comparing PERG noise between groups, data pooled across stimulus parameters).

In summary, in this methodological pilot study we could reproduce earlier findings of reduced PERG amplitudes in patients with depression as a potentially objective biomarker signal of the essentially subjective state of depression. In addition, we were able to methodologically improve the recording procedure by demonstrating the suitability of a higher stimulation frequency for recordings along with the introduction of an interpersonal normalization approach for the PERG signals, which further enhances the sensitivity of the method.