Electroretinography in psychiatry: A systematic literature review

Abstract

This review aims to consolidate the available information on use of electroretinography as a diagnostic tool in psychiatry. The electroretinogram (ERG) has been found to have diagnostic utility in cocaine withdrawal (reduced light-adapted b-wave response), major depressive disorder (reduced contrast gain in pattern ERG), and schizophrenia (reduced a- and b-wave amplitudes). This review examines these findings as well as the applicability of ERG to substance use disorder, Alzheimer’s disease, autism spectrum disorder, panic disorder, eating disorders, attention deficit hyperactivity disorder, and medication use. While there have been promising results, current research suffers from a lack of specificity. Further research that quantifies anomalies in ERG present in psychiatric illness is needed.

1. Introduction

Diagnosis of mental illness is often a lengthy and involved process. As definitions and categorizations of diseases become increasingly complex, a search for objective biomarkers of mental illness is underway to facilitate swift and accurate clinical assessment [1]. Because the retina uses neurotransmitters for phototransduction, it is hypothesized that the deregulation of neurotransmitter physiology present in many mental illnesses may be detectable through assessment of retinal function.

The electroretinogram (ERG), is a non-invasive diagnostic test that measures electrical activity generated by neuronal and non-neuronal cells within the retina and has shown utility in the assessment of psychiatric illnesses [1–3]. It measures electrical impulses through a contact lens which can detect summation of retinal electrical activity at the corneal surface. In mouse models, it has been shown that alterations in retinal dopaminergic and serotonergic neurotransmission parallel abnormalities observed in ERG recordings [2].

The ERG outputs multiple waveforms, which represent different components of retinal electrophysiology [4]. Of these, two waveforms have been shown to be especially important in the setting of psychiatry: the a-wave and the b-wave. The a-wave is representative of photoreceptor function and is measured as the height of the baseline electrical activity before flash onset to the trough of the first wave. The b-wave reflects bipolar cell function and is measured from the trough of the a-wave to the peak of the second wave [4]. Examples of ERG waveforms are shown in Fig. 1.

Fig. 1. fERG, PERG, and mfERG waveforms and parameters. Waveforms based on information provided in included studies.

There exist different types of ERGs, which can assess varied components of the retina. The full-field flash ERG (fERG) assesses the electrophysiological response to a flash of light and is well-established and routinely used in ophthalmology [4]. It is also under investigation for applications in psychiatry. The fERG assesses retinal function under dark-adapted (scotopic) or light-adapted (photopic) settings. The scotopic fERG is usually indicative of rod function, while the photopic fERG commonly correlates with cone function [4]. Mixed cone/rod responses may also be elicited on scotopic ERG. In addition to fERG, pattern ERG (PERG) is also under evaluation in psychiatric populations. PERG provides information regarding macular, bipolar, and retinal ganglion cell function by stimulation with a rapidly reversing high-contrast black and white checkerboard or alternating horizontal and vertical lines, and therefore also allows for inferences of retinal contrast sensitivity [7, 8]. Moreover, some investigators have also begun to study the newer multifocal ERG (mfERG) in the setting of mental illness. This ERG measures the response from hundreds of points on the retina simultaneously and creates a ‘topographic map’ of retinal functioning [9].

Although there remains contention with regards to the neurochemical basis of ERG waves, retinal physiology is known to be primarily dependent on glutamate [7]. Thus, it is hypothesized that changes to the glutamate neurotransmitter system, as are observed in many psychiatric illnesses, may manifest in different ERG profiles in varied disease states. Some groups hypothesize that the retinal endocannabinoid system, dopamine system, or the GABAergic system may also be at play, allowing for insight into psychiatric conditions resulting from aberrant functioning of these signaling pathways [2, 5, 6].

This review explores variations in ERG waveforms across several different psychiatric populations and highlights potential applications for ERG in the detection and assessment of mental illness.

2. Material and methods

2.1 General methods

The present review was conducted in line with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [10].

2.2 Literature search

We performed a systematic search of the Web of Science electronic database using the following keywords: ‘electroretinogram’, ‘electroretinography’, and [‘electroretinography’ + psychiatry (topic)]. To ensure we captured all available evidence in the field, our sole eligibility criteria was that ERG was used on human subjects for diagnosis or assessment of mental illness. We excluded articles not investigating psychiatric conditions, articles with the keyword ‘ophthalmology’, as well as wet lab research and animal studies. Our electronic search was supplemented by reviewing the references of included articles. For the purposes of our review, Alzheimer’s disease was considered a psychiatric condition and was thus included, while epilepsy was considered a neurological condition and was excluded. We included both full reports and conference proceedings, provided that a subsequent article had not been published.

2.3 Data collection

We downloaded the complete articles of all included studies. We created data extraction forms with fields for: population studied, type of ERG (fERG, PERG, mfERG), study design, and ERG results.

3. Results

3.1 General results

Our literature search yielded 9,195 results, from which 55 articles met our inclusion criteria (Fig. 2). Articles covered a spectrum of psychiatric illnesses including substance use, depressive disorders, schizophrenia, panic disorder, and eating disorders. Most articles focused on the applicability of ERG to substance use, depressive disorders, and schizophrenia, with a minority investigating panic disorder and eating disorders. Distribution of psychiatric illnesses amongst the articles is shown in Fig. 3.

Fig. 2. PRISMA flow diagram. The selection process for identifying eligible studies is shown.

Fig. 3. Number of articles by specific psychiatric illness and topic.

Included articles covered over 30 years of research and a timeline of how the landscape of the field has evolved is shown in Fig. 4. Notably, there has been increased interest in the applicability of ERG to psychiatric illness. Use of ERG as a diagnostic tool has had an early start in the scientific literature, however, there has been little success in establishing it for clinical use. As shown, there are also periods with little research progress, concordant with periods of plateauing in technological innovation.

Fig. 4. Timeline of the literature on applications of ERG in psychiatry stratified by specific illness and topic.

3.2 Substance use

We found 15 articles evaluating the effects of various substances on ERG [6, 11–24]. Of these, nine studied the effects of cocaine, three studied cannabis, two studied alcohol, and one studied mixed drug use. A complete summary of the included studies and their findings is shown in Table 1.

Table 1 Articles on substance use.

Notes: CI = confidence interval; CSF = cerebrospinal fluid; ERG = electroretinogram; fERG = flash electroretinogram; mfERG = multifocal electroretinogram; PERG = pattern electroretinogram.

In the setting of cocaine use, it was noted that there was decreased light-adapted b-wave amplitude on fERG. Moreover, cocaine craving was observed to correlate with a depression in the b-wave amplitude [12, 14, 17, 18]. Variable levels of correlation were noted across studies, with calculated r-values ranging from 0.35 to 0.9.

Studies reporting on cannabis use presented varied parameters. Two articles [5, 23] reported an increase in implicit time, with Schwitzer et al. [5] reporting a sensitivity of 78.6% (95% CI 60.5–89.8) and a specificity of 75.0% (95% CI 55.1–88.1) for an N95 implicit time longer than 93.15 ms for PERG.

For alcohol consumption, one study [6] found a reduced dark-adapted b-wave amplitude on fERG.

3.3 Alzheimer’s disease

Four studies evaluating differences in ERG waveforms in Alzheimer’s disease met our inclusion criteria [25–28], all of which used PERG. No studies were able to determine a statistically significant difference in PERG in patients with Alzheimer’s disease, however, a trend was noted towards a reduced b-wave amplitude. Details of studies investigating Alzheimer’s disease are shown in Table 2.

Table 2 Articles on Alzheimer’s disease.

Notes: PERG = pattern electroretinogram.

3.4 Autism spectrum disorder

Similar to Alzheimer’s disease, four articles studying ERG in the context of autism spectrum disorder were included in our review [29–32]. Three articles [29, 30, 32] used a fERG while one used a PERG [31]. Articles using a fERG found a decreased b-wave amplitude (rod function) in patients with autism spectrum disorder, however, quantitative analyses were not provided. Realmuto et al. [30] showed a similar reduction in a families of patients with four siblings and two fathers affected, congruent with the neurodevelopmental origin of autism. Results from studies in this group are shown in Table 3.

Table 3 Articles on autism spectrum disorder.

Notes: fERG = flash electroretinogram; PERG = pattern electroretinogram.

3.5 Depressive disorders

12 articles evaluating ERG in depressive disorders were eligible [33–44]. Seven of these studied seasonal affective disorder (SAD) and the remaining five major depressive disorder (MDD). There was little consistency in protocol across studies, making cross-comparisons difficult. For MDD, a trend was noted towards decreased contrast gain on PERG, especially for moderate-to-severe depression, although Fam et al. [42], were unable to replicate these findings. Notably, studies also evaluated the impact of interventions on MDD and found that ERG abnormalities resolved with treatment. Similarly, studies evaluating SAD showed variable results on ERG, although there was normalization of irregularities with light therapy or during the summer months [36]. Details of the studies investigating depressive disorders are shown in Table 4.

Table 4 Articles on depressive disorders.

Notes: BDI = Beck Depression Inventory; ERG = electroretinogram; fERG = flash electroretinogram; HAM-D = Hamilton Depression Rating Scale; MDD = major depressive disorder; PERG = pattern electroretinogram; SAD = seasonal affective disorder; SSAD = sub-syndromal seasonal affective disorder.

3.6 Schizophrenia

In total, 11 articles investigating ERG in the context of schizophrenia were eligible [45–55]. All but one of the articles used the fERG [45–53, 55], with the remaining article implementing PERG [54]. Hébert et al. [50] noted an increase in implicit time in patients with schizophrenia. Across articles, there was also a trend of reduced a- and b-wave amplitudes in schizophrenia. Interestingly, Gerbaldo et al. [46] failed to find this reduction, although their study population differed from that of other articles as it included individuals with a history of sungazing. The most powered study in this group, by Hébert et al. [48], showed reductions in both a- and b-wave amplitudes. Notably, Demmin et al. [55] also provided data on photopic negative response, noting that the schizophrenia group demonstrated attenuated negativity in comparison to healthy controls. Full details of articles investigating ERG in schizophrenia are available in Table 5.

Table 5 Articles on schizophrenia.

Notes: ERG = electroretinogram; fERG = flash electroretinogram; PANSS = Positive and Negative Syndrome Scale; PERG = pattern electroretinogram.

3.7 Panic disorder

Two articles evaluating the diagnostic ability of fERG in participants with panic disorder were included in our review [56, 57]. Both articles found decreased b-wave amplitudes in patients, and reduced differences between right and left eyes in the b-wave amplitudes. The details of these two studies are discussed in Table 6.

Table 6 Articles on panic disorder.

Notes. fERG= flash electroretinogram.

3.8 Eating disorders

Only two studies evaluating ERG in the context of eating disorders met our inclusion criteria [58, 59]. Nasser et al. [58] discussed oral food stimulation in the setting of binge eating, while the study by Moschos et al. [59] discussed anorexia nervosa. An increase in b-wave amplitude in response to brownie consumption and a correlation between the Gormally Binge Eating Scale and increased b-wave amplitude (r = 0.68) was noted on fERG for binge eating. Moreover, the authors noted that there was a similar increase in b-wave amplitude following administration of methylphenidate [58]. In anorexia nervosa, a decreased P1 retinal response density amplitude was found for ring 1 using mfERG [59]. Complete details of studies in this group are available in Table 7.

Table 7 Articles on eating disorders.

Notes. fERG= flash electroretinogram; mfERG= multifocal electroretinogram.

3.9 Attention deficit hyperactivity disorder

One article discussing attention deficit hyperactivity disorder (ADHD) was eligible, and it utilized the PERG [60]. Treatment with methylphenidate was also discussed. The authors found that there was increased background noise on PERG in patients and that this normalized with treatment. Table 8 includes details of the article investigating ERG in ADHD.

Table 8 Articles on ADHD.

Notes. ADHD= attention deficit hyperactivity disorder; PERG= pattern electroretinogram.

3.10 Medication

Five articles assessed the effects of medication on healthy participants by ERG [61–65]. Two articles found that perphenazine decreased b-wave amplitudes, while bromocriptine increased them [62, 63]. Several dopamine antagonists such as haloperidol, chlorpromazine, fluphenazine, and metoclopramide, reduced b-wave amplitudes, although these decreases were not quantified [63, 64]. Details of studies in this group are discussed in Table 9.

Table 9 Articles on medication use.

Notes: fERG = flash electroretinogram; NDRI = norepinephrine-dopamine reuptake inhibitor; TCA = tricyclic antidepressant.

4. Discussion

The present review aimed to evaluate the utility of ERG as a diagnostic tool within a psychiatric setting in a systematic manner. Because of the lack of consistency amongst protocols, small sample sizes, and lack of replicable findings, there remains much work that needs to be done before ERG may be used reliably in psychiatry. The major findings of our review are summarized in Table 10.

Table 10 Summary of ERG findings across psychiatric illnesses and groups.

Notes: ADHD = attention deficit hyperactivity disorder; ERG = electroretinogram; MDD = major depressive disorder; PERG = pattern electroretinogram; SAD = seasonal affective disorder.

In performing our review, we found it exceedingly difficult to make comparisons across studies and draw conclusions for specific disease groups. There was little consistency between protocols in included articles, and studies within disease groups often examined distinct forms of therapy and were conducted over drastically different timelines, leading to variable results. Moreover, many studies did not quantify their results, making it challenging to draw comparisons and eliminating any possibility of aggregating results meaningfully through a meta-analysis. While qualitative observations of waveform reductions and increases are insightful at an exploratory level, they do not provide an objective assessment of diagnostic utility.

In order for ERG to become a diagnostic modality in psychiatry, there is a need for quantifiable data that allows for determination of important measures such as sensitivity, specificity, and positive and negative predictive values in comparison to gold standard diagnostic methods. To facilitate this, we suggest that future studies implement the International Society for Clinical Electrophysiology of Vision (ISCEV) guidelines for the measurement of ERG waveforms and should quantify anomalies and calculate the magnitude of correlations with disease states. Prior studies were largely underpowered due to a small sample size. This may have been a result of requiring trained technicians and nurses to carry out pharmacological pupil dilation prior to ERG. Newer ERG technology permits gathering of data without the need for pupil dilation. Thus, authors should ensure that their studies have sufficient sample sizes and are adequately powered to assess outcomes of interest. To enrich results and minimize bias, future studies should be multi-center. At present, most of the research on a particular condition is conducted by the same authors. For instance, of 12 articles on depressive disorders, seven were from the same research group, and out of 15 articles on substance use, nine were from the same group of authors. Having multiple centers participate in studies would serve to both internally and externally validate study protocol and improve the reliability of findings.

Future studies should also consider expanding the use of ERG to yet unexplored psychiatric conditions. For example, our search did not identify any eligible studies examining bipolar disorder, indicating a gap in the present evidence.

An inherent difficulty of using ERG as a diagnostic tool in psychiatry is the limited number of detectable anomalies, restricting diagnostic specificity. For example, a reduction in b-wave amplitude on fERG may be indicative of cocaine withdrawal, autism spectrum disorder, panic disorder, perphenazine use, or intake of dopamine blockers. From a neurobiological perspective, this is perhaps due to the high degree of interconnectedness between many neurotransmitter systems [66]. Emerging research in animal suggests that ERG responses can also be affected by dopamine, accounting for observed anomalies in relevant psychiatric illnesses and medication use [2]. To be used reliably as a diagnostic tool, ERG will have to demonstrate more particular anomalies for each condition. If this proves to be unfeasible, however, then the diagnostic utility of ERG may be limited and its use in psychiatry may be better suited to measuring the effect of treatments or predicting relapse in patients.

5. Limitations

Our review is not without its limitations. We do not present quantitative analyses or a meta-analysis summarizing the results of ERG anomalies in specific psychiatric conditions as included studies did not provide such data.

Moreover, our review is at risk for publication bias as many of the studies included were conducted by a limited number of research groups and were largely single-center in design. Our review is also limited by its broad research question. There is a paucity of literature when examining the applicability of ERG to specific psychiatric illnesses, and thus we chose to look at a myriad of pathologies in order to provide the most comprehensive summary of the current evidence.

As the pace of research in this field accelerates, we anticipate future reviews will have well-refined and focused research questions and will allow for quantitative analyses.

6. Conclusions

This review systematically examined the literature on the use of ERG as a diagnostic tool in psychiatry. Although it was difficult to draw quantitative conclusions, consistent trends in ERG waveform anomalies in specific psychiatric conditions were observed across included studies. ERG is a non-invasive, quickly administered, and well-characterized test that has the potential to become an objective tool for the diagnosis of psychiatric illness. Further investigation through adequately powered multi-center studies, in concordance with the rapid pace of technological advancement, will allow for thorough evaluation of ERG in comparison to existing gold standard modalities and permit its successful integration into the diagnostic repertoire of modern psychiatry.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.