Chapter 4. Auditory Event-related Potential of Subjects with Suspected
4.2.4 Statistical analyses
For demographic characteristics, we used analyses of variance and chi-square tests (or Fisher’s exact tests if necessary) to compare continuous and categorical variables
across different risk groups and normal controls, respectively. The correlations between ERP parameters were examined using the Spearman rank correlation tests. Analyses of variance with post-hoc analyses were used to examine differences in ERP parameters among these four groups. Treating the risk level as a continuous covariate, linear trends of ERP parameters across these four groups were checked by regression models. A subgroup analysis of participants within the UHR group was performed to determine factors associated with converting to full-blown psychosis or not. Demographic characteristics, SOPS symptom dimensions (i.e. positive, negative, disorganized and general symptoms) and ERP parameters were compared between converters and non-converters. Chi-square or Fisher’s exact tests were used for categorical variables, while nonparametric
Mann-Whitney U tests were used for continuous variables because of the small sample size for converters and non-converters in the UHR group. All tests were 2-sided with α = 0.05.
4.3 Results
In total, we recruited 99 clinical subjects, including 32 FEP, 30 UHR, 37 E-BARS, along with 56 normal controls (Table 4.1). There were no significant differences in age, gender, education, and smoking status. Only the UHR and FEP subjects were prescribed with antipsychotics.
Regarding the relationship between individual ERP indicators, the Spearman’s rank correlation coefficients are outlined in Table 4.2. The majority of P50 and N100
parameters were mutually correlated, except no correlation existed between N100 ratio and any P50 parameter. MMN was correlated with neither P50 nor N100 parameters.
With respect to the differences in ERPs between these four subgroups (Table 4.2), only MMN reached statistical significance (p = 0.019). In post hoc analyses, there were
significant differences in MMN in the E-BARS (p = 0.007), UHR (p = 0.035), and FEP (p
= 0.035) groups as compared to the controls.
Figure 4.1 demonstrates linear trends of P50 ratios (S2/S1) and the N100 differences across different risk groups (P50 ratios, p=0.060; N100 differences, p=0.018); that is, these two sensory gating indicators were largest in the FEP group followed by the UHR group, the E-BARS group and the normal controls in order. Grand average MMN waveforms for the FEP patients (in blue) and control subjects are shown in Figure 4.2.
The MMN waveform reversed in polarity at the mastoid electrodes.
Further analysis for participants within the UHR group showed no significant differences between converters and non-converters in either demographic profile or any of the four symptom dimensions (Table 4.4 and Figure 4.3). There was some evidence suggesting that the converters had a poorer performance than the non-converters in several P50 and N100 indicators including P50 gating ratio (p = 0.099), N100 gating ratio (p = 0.060), N100 difference (p = 0.088), and N100 S2 amplitude (p = 0.060), but not MMN.
4.4 Discussion
To the best of our knowledge, this study is one of the first to examine auditory ERPs (P50/N100/MMN) in not only subjects with first-episode psychosis (FEP) and
ultra-high-risk (UHR) subjects, but also in those with presumed early/broad at-risk mental states (E-BARS). In general, MMN was correlated with neither P50 nor N100, whereas many parameters of the latter two were inter-correlated with each other.
Specifically, as compared to healthy controls, all three clinical groups, i.e. E-BARS, UHR and FEP had significantly lower MMNs. On the other hand, the differences in P50 and N100 between control and clinical groups were not significant, while a linear trend
of more deviance from controls across different levels of clinical severity was noticed in P50 ratios (S2/S1) and N100 differences (Figure 4.1). For subjects within the UHR group, certain P50 and N100 indicators might be useful when attempting to discriminate converters from non-converters.
Examining subjects with a gradient of clinical severities spanning from normal control, early at-risk state, ultra-high risk state to first episode psychosis is helpful to delineate the pathophysiological mechanisms throughout the formation of psychosis. Our results suggest that MMN and P50/N100 represent quite different inferences in the pathological information processing of subjects with at-risk mental status. This is in agreement with current knowledge that MMN reflects deviance detection which might be mediated by glutamate(Korostenskaja et al., 2007; Leung et al., 2007; Javitt et al., 2008;
Korostenskaja & Kahkonen, 2009), while P50/N100 refers to sensory gating which is more likely related to dopamine and other neurotransmitters (Pekkonen et al., 2005; Hall, Schulze, Bramon, et al., 2006; Price et al., 2006; Turetsky et al., 2007; Javitt et al., 2008;
Keshavan et al., 2008; Turetsky et al., 2009). The high correlations between N100 difference and P50 ratio and P50 difference was compatible with previous studies(Fuerst et al., 2007; Brockhaus-Dumke et al., 2008), suggesting “both P50 and N100 reflect stimulus registration in similar ways but gating or habituation to repeated stimulation in different ways”(Brockhaus-Dumke et al., 2008).
Our findings in duration MMN suggest it to be a trait, or a very sensitive marker, for schizophrenia, which means reduced MMN could be detected at subjects presenting with symptoms suggesting a putatively pre-psychotic state (Green et al., 2009; Atkinson et al., 2012), yet such a reduction might not get much worse along with the increase of clinical severity, especially in terms of emergence of attenuated psychotic symptoms. Previous studies have demonstrated impaired duration MMN in nonpsychotic biological
first-degree relatives of patients with schizophrenia (Michie et al., 2002) and reduced MMN in subjects at ultra-high risk state (Michie et al., 2002; Shin et al., 2009; Atkinson et al., 2012), and glutamate system dysfunction has been noted in at-risk mental state subjects (Stone et al., 2009). This study further revealed that even people at early/broad risk states might already demonstrate detectable MMN reduction.
In contrast to MMN, the parallels between the extent of sensory gating problems manifested by P50 gating ratio and N100 differences and the gradient of clinical severity suggest these two ERP indices might be state-dependent markers for schizophrenia. This might violate the definition of an ideal endophenotype (state-independent or
symptom-independent). However, several studies have provided mixed results with regards to the relationship between P50 gating ratio and clinical presentations (Ringel et al., 2004; Louchart-de la Chapelle et al., 2005), between clinical high-risk and genetic high-risk (Myles-Worsley et al., 2004), as well as between different clinical stages (Brockhaus-Dumke et al., 2008). Nonetheless, our findings could provide new insights regarding the interpretation of such inconsistent findings. We conjecture that during pre-psychotic state when sensory-gating deficits are relatively mild, P50/N100 might be state-dependent markers as revealed by our findings; but once frank psychosis occurs and the sensory-gating problems become manifest, the severity of symptoms or duration of psychosis were less likely to have strong correlation with the extent of P50 deficits as revealed by a review of studies(Potter et al., 2006).
Based on our preliminary analysis, P50 and N100, rather than MMN, are potential candidates to differentiate converters and non-converters among subjects at ultra-high risk for schizophrenia, even though a recent study revealed reduced duration MMN associated with a higher risk of converting to first-episode psychosis among at-risk subjects (Bodatsch et al., 2011). Actually, among our UHR subjects, the mean MMN of
converters was indeed lower than non-converters (converter vs. non-converters = -.50 vs.
-1.06) but this was not statistically significant. This could merely be an issue of statistical power because of the small sample size in this subgroup analysis (converters, N=6;
non-converters, N=13). Further research about predicting conversion in UHR subjects by different indices of ERPs will be necessary to clarify this issue.
There are several limitations that are worth noting. The relatively small sample size limits our statistical power to detect smaller between-group differences. The validity of our clinical subgrouping of early/broad at-risk mental states is pending further follow-up and exploration. UHR and FEP subjects were not studied in an antipsychotic-free status;
while use of antipsychotic might diminish the magnitude of P50 gating deficit hence masks some potential findings. In addition, we used data collected by midline electrodes to analyze the ERPs for consistency with previous literature and protocols, while the German Research Network on Schizophrenia Group used lateral electrodes to yield positive findings on prodromal studies (Frommann et al., 2008), thus topographic maps and source localization are factors to be considered when studying the ERPs underlying these high-risk subjects.
By employing the concept of E-BARS, this study provides new inferences about pre-attentional auditory event-related potentials, i.e. P50, N100 and MMN, in subjects across different risk levels of psychotic disorders, from early/broad at-risk mental state, ultra-high risk state, and first episode psychosis. Impaired deviance detection already exists in people at pre-psychotic state, regardless of clinical severity. On the contrary, sensory gating varies depending on different risk levels. A preliminary analysis showed some promising results for predicting conversion to psychosis. Further longitudinal research monitoring neurobiological changes of the same subjects at different levels of clinical severity are necessary to explore the underpinning pathogenesis.
4.5 Tables and Figures
Table 4.1 Demographic data of the four subgroups.
Table 4.2 The Spearman's correlation coefficients among P50, N100 and MMN Parameters. Number of subjects for P50/N100 was 152 and for MMN was 130.
Table 4.3 P50, N100 and MMN parameters among the four Subgroups a
Table 4.4 The comparison of clinical characteristics and ERP parameters in converters versus non-converters among ultra-high risk group. (n=30)
Figure 4.1 P50 ratios and N100 differences. The left panel demonstrates P50 ratio (S2 amplitude / S1 amplitude) and the right one N100
difference (μV; S2 amplitude–S1 amplitude) of individual participants. Larger ratio (S2/S1) and smaller difference (S1–S2) indicate poorer gating.
The horizontal lines indicate the mean values within each risk group. CTL:control; E‐BARS: early/broad at-risk mental states; UHR: ultra-high risk group; FEP: first-episode psychosis.
Figure 4.2 Grand average mismatch negativity (MMN) waveforms for healthy control subjects (in blue) and (A) MRG, (B) IRG, (C) UHR, (D) FEP subjects (in red). Left and right columns indicate Fz and A1 (mastoid) electrodes. The MMN waveform reversed in polarity at the mastoid electrodes.
Figure 4.3 Three event-related potentials in non-converters versus converters within the ultra-high risk subgroup (UHR). The left panel shows P50 ratio (S2 amplitude/S1 amplitude), the middle one N100 difference (μV; S2 amplitude–S1 amplitude), and the right on MMN (μV) of UHR individuals. The horizontal lines denote the mean values.