Deficits of perceived spatial separation induced prepulse inhibition in patients with schizophrenia: relationships to symptoms and neurocognition

Eighty-seven patients and 60 healthy controls were enrolled with right-handed and did not show any pure-tone hearing impairment at the frequency of 1000 Hz in each ear. Diagnosis of schizophrenia was confirmed by administering the Structured Clinical Interview for DSM-IV (SCID) by research psychiatrists. Enrollment criteria for schizophrenia patients included: 1) clinically stable subjects without neurologic diseases or known history of head trauma, 2) no electroconvulsive therapy within the past 6 months, and 3) no history of drug or substance abuse and dependence (except tobacco). Patients were excluded if they had unstable medical conditions or IQ below 70. All patients received antipsychotic medications as usual during the period of this study.

The control group included mentally stable, healthy individuals. All control subjects answered semi-structured questions to exclude participants with a history of drug or alcohol abuse or dependence within 6 months before testing, major head trauma involving loss of consciousness for more than 5 min, epilepsy or other neurological dysfunction or first-degree family history of psychosis. Following a detailed description of the procedures involved and overall aims of this study, all participants provided informed written consent. This study was reviewed and approved by the Independent Ethics Committee (IEC) of the Beijing Anding Hospital, Capital Medical University, China.

Basic socio-demographic characteristics and clinical data were collected by a questionnaire specifically designed for this study. The psychopathology of each patients was evaluated by the Positive and Negative Syndrome Scale (PANSS, Chinese version) [36]. Patients and healthy controls were additionally assessed on Repeatable Battery for the Assessment of Neuropsychological Status (RBANS, Chinese version) [37] and the traditional card version of the Stroop Color-Word Test (Chinese version) [38]. There was good agreement on ratings performed by physicians and trained psychologists (Cohen’s = 0.80).

Each subject sat comfortably in a recliner chair in a sound-attenuated room. Two 4 mm Ag/AgCl electrodes were positioned below and lateral to the right eye, directly over the orbicularis oculi. A ground electrode was then placed behind the right ear. Electrode resistances were <5 kΩ. The eye blink component of the acoustic startle system was evaluated using a human electromyography (EMG) startle reflex system (EMG XEYE human Startle Reflex, Tian Ming Hong Yuan Instruments Company, Beijing, China). This system recorded 300, 1-ms epochs for digitization and analysis, beginning with the startle stimulus onset. In addition, EMG activity was band-pass filtered (100-1000 Hz) and amplified by 10,000. Acoustic startle stimuli were presented binaurally through two headphones (HD 265 linear, SENNHEISER, Germany) (Fig. 1a). Acoustic signals were calibrated by a sound-level meter (AUDit and System 824, Larson Davis, USA).

The test began with a 2-min adaptation period of 60 dB SPL broadband background noise, during which 4 startling sounds (broadband white noise, 40 ms, 104 dB SPL) were presented. Next, we conducted 2-blocks of PPI testing. In each block, 7 trials contained the startling sound alone delivered by two headphones, as well as 20 trials containing the prepulse (broadband white noise, 150 ms, 65 dB SPL) preceding the startling noise (120 ms between the prepulse offset and the startling-sound onset) (Fig. 2). The prepulse was presented from each headphone with the inter-loudspeaker onset delay being either +3 ms (left leading) or −3 ms (right leading). Due to the precedence effect, a single perceptually fused image of the prepulse would be perceived at the left loudspeaker location in the first block (when the left loudspeaker led) and at the right loudspeaker location in the second block (when the right loudspeaker led) [17]. In addition to the prepulse, background, wideband noise (0–10 kHz, 60 dB SPL) was continuously delivered from each of the two headphones as a masker. The inter-headphone onset delay for the masker was also either +3 or −3 ms, leading to a perceptually-fused, continuous noise-masker image that was presented either through the left headphone in one block or through the right in the other block. Thus, there were 4 (2×2) combinations of perceived locations between the prepulse and masker images across the two experimental blocks. Two types of perceived spatial relations between the prepulse and masker were created: perceptual separation (when the prepulse and masker came from a different leading headphone, Fig. 1b) and perceptual co-location (when the prepulse and masker shared the same leading headphone, Fig. 1c). Trials in each block were presented randomly with the inter-trial interval of approximately 20s (varying between 15 s and 25 s).

Startle eyeblink responses were recorded as electromyographic activity. Each trial was visually inspected for spontaneous and voluntary blinks. Trials with artifacts were excluded from analysis (total trial exclusion <4%). Electromyographic activity under either the perceived spatial co-location (PSC) or perceived spatial separation (PSS) conditions were recorded.

Percent habituation (100 × [average of first habituation block − average of second habituation block]/average of first habituation block) was calculated for each subject. PPI% = 100×(1 − [mean magnitude on prepulse trials/mean magnitude on pulse alone trials]).

Data analysis was performed by R Development Environment and R Studio Desktop (Open Source Edition) [39]. The normality of the distributions for continuous variables was checked using a one-sample Kolmogorov–Smirnov test. Comparisons between the two groups of subjects about socio-demographic characteristics and mean scores on PPI and cognitive tests were performed using independent sample t-test and chi-square tests. Binomial logistic regression was applied for discriminating patients from controls. Associations of performance on PPI tasks with socio-demographic and clinical characteristics and other neurocognitive tests were analyzed using Pearson correlation analysis if the data followed a normal distribution; otherwise, Spearman rank correlation analysis was performed. Stepwise multiple linear regression analyses were used to identify factors that were independently associated with performance on the two types of PPI. In the stepwise regression analyses, the perceived spatial co-location PPI (PSC-PPI) and perceived spatial separation PPI (PSS-PPI) scores were entered as the dependent variable, and all variables were entered as independent variables. Two tailed tests were used in all analyses with the significance level set at 0.05.

A total of 22 participants (12 patients and 10 control subjects) were excluded for their failing to blink to the startling stimulus. Demographic characters as well as the medication status for the remaining participants are summarized in Table 1. There were no significant differences in age, years of education, smoking status and gender between the two groups. Medication status revealed that patients had the following pharmacotherapy regimens: typical antipsychotics:18.7% (n = 14); atypical antipsychotics: 53.3% (n = 40); and combination: 18% (n = 21), respectively.

Results from between-group differences on attention modulated auditory startle reflex are shown in Table 2 (Fig. 3.) Startle magnitude and habituation revealed no differences across groups. In addition, we detected significant differences between healthy control and patients for PSC-PPI and PSS-PPI. Effect sizes of the two PPI values were all large (Cohen’s d: PSC-PPI = 0.84, PSS-PPI = 1.27). The main effect of perceived spatial separation in relation to PPI was significant in healthy controls (t = − 10.57, p < 0.01), but not in schizophrenics (t = − 0.47, p = 0.64).

The univariate logistic regression was used to generate models that maximally separated schizophrenia patients and healthy controls. Several parameters from Table 3 were used to evaluate the model. The goodness-of-fit test (model χ²: 29.3), and effect size estimate (Nagelkerke’s R²: 0.46) all indicated the PSS-PPI model’s greater appropriateness. The AUC of PSS-PPI obtained boosts from 72% to 85.2% for statistical significances comparing to that of PSC-PPI.

Between-group differences on cognitive variables are shown in Table 4 (Fig. 4). Patients showed significantly poorer performance than healthy controls on all cognitive tests (all values were significant at p < 0.05).

Table 5 presents the relationships between PPI and sociodemographic variables, other cognitive tests, and clinical variables. In the HC group, gender, higher scores on the attention score (ATT) and total composite score (TOT) were significantly associated with better PPI performance, and higher color interference time (INT-C) scores were associated with reflex startle, whereas less word interference time (INT-W) associated with better PSS-PPI performance. No variable was associated with any PPI tasks in patients except PANSS General negatively correlated with PSS-PPI.

In order to further discover the correlation between PPI and the symptoms severity of schizophrenia, we performed Pearson’s analysis using all potential confounding variables (age, gender, education, duration, recurrence times, onset, smoke amount and drug) as covariates [40]. We did not observe any significant association between the mean PPI and PANSS scores for all 75 patients and patients whose symptoms were classed as “remission” (total PANSS score < 60). However, in patients who were “non-remission” (total PANSS score ≥ 60) [41], we observed a significant correlation between the PSC-PPI, PSS-PPI with P1, P6, P7, PANSS Positive, N5, N7, G9 and PANSS Total, as well as a significant correlation between the PSS-PPI with PANSS Negative (see Table 6).

Table 7 shows the independent correlates of performance on PSC-PPI and PSS-PPI in healthy controls and patients. In the healthy group, male and higher ATT scores contributed to better PSC-PPI and PSS-PPI, while high nicotine dependent and longer INT-W contributed to poorer PSS-PPI performance. In patients’ group, ATT and TOT scores accounted for 9% of the variation in PSC-PPI, and did not reach statistical significance. But higher education levels, ATT and TOT scores contributed to PSS-PPI, and recurrent episodes contributed to poorer PSS-PPI.