This study aimed to explore the visual cortical processing among patients with schizophrenia compared to healthy controls and to explore its link with previous retinal electrophysiological processing and visual cognition such as visual hallucinations. First, we found a decreased P100 amplitude and an increased P100 latency in SZ compared to HC. Importantly, these results were independent of either SF or TF nature of the stimuli. In the SZ group, P100 latency was correlated with the VOSP object score and previous retina results such as N95 latency 1. Third, P100 amplitude was decreased in AHNH compared to both VH and HC and P100 latency was increased in VH compared to HC. In the VH group, P100 latency was correlated with the VOSP object score and previous retina results such as the N95 latency 1. Finally, a mediation analysis reported both causal and temporal relations between retinal and cortical measures as well as visual cognitive abnormalities in VH group, shown by a longer N95 latency, a longer P100 latency and weaker performances in the VOSP object score.
Our primary goal was to explore the visual cortical processing among patients with schizophrenia compared to healthy controls. Thereby, we found P100 alterations in the SZ group, which are consistent with the most studies which refers to impairments in the very first step of the early visual cortical processing and notably the primary visual cortex 69,70. Analysis also mentioned a higher P100 amplitude and shorter P100 latency for LSF stimuli and dynamic condition 19,21−23, consistent with the magnocellular sensitivity to LSF information and movements 64. Despite, our analysis failed to find any SF*Group interaction or TF*Group interaction. Such results hence undermine this potential magnocellular alteration in schizophrenia although well described in the literature. For instance, Butler et al. 22 reported a P100 reduction in response to magnocellular biased stimuli with gratings very similar to those in this present study. Nevertheless, and compared to our study, their patients were older (35.9 ± 2.2 years vs. 29 ± 8.15 years in our study), with a duration of illness twice as long (16.7 ± 1.8 years vs. 94.67 ± 93.21 months in our study) and chlorpromazine equivalent rates nearly three times higher (1365.4 ± 157.7 mg/day vs. 544.55 ± 241.29 mg/day in our study). In this regard, literature mentions that visual abnormalities in schizophrenia may be more pronounced with age, duration, stage, and severity of the illness (Joseph 2013). Similarly, the use of antipsychotic medications may impair visual performance, including contrast perception, to a greater extent than without the use of antipsychotic medication 8,71. In sum, several factors could have an influence on this potential magnocellular alteration listed in schizophrenia, which would be less visible here in our patients.
If our results are not conditioned by the use of visual stimuli preferentially oriented towards the magnocellular or parvocellular system, it is important to consider the interaction between retinal and cortical processing. In our paradigm, we found a positive correlation between the N95 latency at the retinal level and the P100 latency at the cortical level in the SZ group. Hence, one hypothesis would be that P100 cortical alterations rather reflect retinal dysfunctions than thalamic magnocellular dysfunctions. This hypothesis also fits with previous studies which have shown that retinal anomalies can have repercussions at the cortical level in ophthalmic and neurological disorders such as anisometropic amblyopia 47, Alzheimer disease 46 and migraine 48, as evidenced by multiple deficiencies on electrophysiological measurements of N95, P50 and P100 waves.
This study also aimed to determine the links between electrophysiological visual abnormalities, both at retinal and cortical level, and the presence or absence of visual symptoms such as VH in schizophrenia. Although electrophysiological abnormalities appear to be present in all subgroups, both at the retinal and cortical level and either in the VH or the AHNH group, the delay at the cortical level appears to follow the delay at the retina level in the VH group, as reflected by a greater N95 latency and a greater P100 latency. Moreover, the P100 delay in the VH group was associated with lower VOSP object-score which is also consistent with our recent retina results mentioning that greater N95 latency was associated with weaker performances at the VOSP score predicted the risk of having VH 49. In patients with VH, there would therefore be a potential association between the retinal abnormalities, the visual cortical abnormalities and processing in high-level visual cognition such as VH. On this purpose, our mediation analysis performed in the VH group also emphasize cognitive models of VH that hypothesized the role of impaired retinal input as a risk factor for developing VH 72. For instance, the Activation Input Modulation model proposed by Diederich, Goetz and Stebbins in 2005 73 hypothesizes that the retina may cause a dysregulation of the input dimension leading to release of internally generated visual images causing VH in Parkinson's disease 74. This may be consistent with our findings and previous studies linking retinal dysfunction, cerebral dysfunction, and visual symptoms 49,75 and also be applicable here in the VH group for schizophrenia. In sum, one hypothesis could be that patients with VH would be potentially more sensitive to changes in the retinal processing, which impact therefore the visual cortical measures in EEG and the visual clinical performances. However, no existing model are perfectly tailored for VH in schizophrenia as they aim to explain VH in neurological disorders 72 and further research are required to clarify the role of the early visual processing in VH.
The main strength of our study is that our findings are the first to raise potential arguments for a relation between retinal and visual cortical abnormalities in psychosis. More, such results concerning the symptomatology of VH coupled with high-level visual cognition has never been reported. First, this possible link would suggest that retinal anomalies could potentially have repercussions on the visual cortical processing in patients with schizophrenia and VH. This is especially important since our regression analyses integrated the VOSP object score, which assesses perceptual processes independent of potential disease-related cognitive and motor impairments 76,77. Moreover, electrophysiology provides good objectivity, reliability, and reproducibility in the results, especially during measurements involving low-level visual stimuli that are poorly sensitive to attentional factors. As VH are associated with a severe psychopathology, a low prognosis 78–80 and a high risk of mortality 81, electrophysiology could be used to study patients in clinic for better therapeutic management.
Our study has also certain limitations. First, our sample sizes were small and additional studies are needed to validate these results on a larger scale. Second, the AHNH group included patients with a history of auditory hallucinations or patients with no history of hallucinations at all. Thus, interpretation of the results on this group was very difficult and further studies are needed among patients with AH only. Third, although there were no differences between groups in terms of clinical characteristics, substance use must be considered. Indeed, Schwitzer et al. 82 found an increase in N95 latency among regular cannabis users. At the cortical level, we recently found a P100 impairment in regular cannabis users in response to the same stimuli to those described in this study 61. Similarly, smoking nicotine would affect the P100 latency in studies using visual modality tasks 83 and alcohol could also have effects by causing a P100 latency delay, even in healthy subjects 84. Nevertheless, we did not find any correlations between cannabis or cigarettes, or number of alcohol glasses consumed daily and the P100 results. Moreover, only two patients were cannabis users, and they were not alcohol dependent. Additional studies in patients with schizophrenia without substance use are now necessary. Fourth, although we had the same patients as in our previous retina study 1, stimulations were different for both electrophysiological methods and measures were decoupled. As the stimuli used tested low-level visual processing for both techniques, we assume to compare the retinal and brain measures. However, simultaneous ERG-EEG recordings are now necessary to confirm these results.