Abstract The association of P300 components with age, illness duration and gender were examined in schizophrenia patients and whether such variables indicate a progressive course. A total of 60 patients with schizophrenia and 70 healthy controls were studied utilizing standard auditory oddball tasks. Both healthy and schizophrenia groups had a significant positive correlation between age and P300 latency. There was also a significant positive correlation between illness duration and P300 latency in the schizophrenia group. The prolonged latency of P300, associated with age or illness duration, was more prominent in male than female schizophrenia subjects. These findings suggest gender differences in disease progression in schizophrenia.
The hypothesis that schizophrenia is a progressive disease has been proposed based on clinical observations that some schizophrenia patients display progressive deterioration of both symptoms and cognitive functions. Brain imaging studies have shown progressive changes in the brains of schizophrenia patients including enlargement of the ventricles,1–7 a decrease in cerebral volume,1,5,8,9 and a reduction in size of the frontal cortex10,11 and superior temporal gyrus.10,12 Based on studies using post-mortem tissue and brain imaging, a two-step model of pathogenesis has been proposed in which any neurodevelopmental abnormality present prior to disease development is accompanied by additional progressive changes after disease onset.13–15
Measures of P300 are representative components of event-related potentials (ERP) and are a useful method to non-invasively identify changes in brain function. It is reported that patients with schizophrenia have an attenuation of P300 amplitude16–18 and a prolongation of latency.19–21 Many studies have reported prolongation of P300 latency in neurodegenerative diseases such as Alzheimer's and Parkinson's disease.22,23 In healthy subjects it is known that P300 latency prolongs with age.24 Therefore, in the present study we examined whether abnormalities in P300 amplitude and latency among subjects with schizophrenia are progressive in nature.
There are a few reports of a correlation between age and P300 latency25–27 or amplitude28 in patients with schizophrenia. Mathalon et al. and Frangou et al. also reported a correlation between illness duration and P300 latency.25,27 However, in many of these reports the subjects belonged to very specific groups such as chronic male patients,25 elderly patients28 or familial onset patients.27 In the present study we therefore examined a much larger and more diverse group of patients ranging in disease course from the first episode to the chronic phase. Thus we hoped to determine whether abnormalities in P300 amplitude and latency are related to gender or illness duration.
We studied 60 subjects, 28 men and 32 women, with schizophrenia, who were patients at the Department of Neuropsychiatry of the Fukushima Medical University, and who fulfilled the DSM-IV and ICD-10 diagnostic criteria. Seventy healthy controls, 39 men and 31 women, were also studied. The healthy controls all fulfilled the appropriate measures of the Minnesota Multiphasic Personality Inventory (MMPI) and had no history of alcohol or drug abuse, psychiatric or neurological illnesses. The onset of illness in patients with schizophrenia was considered to be the time at which the patient first fulfilled the ICD-10 diagnostic criteria for schizophrenia. The subjects were provided with an explanation regarding the intent and details of the study, and provided written consent in accordance with the ethical committee of Fukushima Medical University.
The mean age of the control subjects was 39.0 ± 12.0 years (range 20–63 years), and that of the schizophrenia subjects was 33.0 ± 12.0 (range 17–64years). When divided by gender, the mean age of the normal male subjects was 39.4 ± 10.3 years (range 21–63 years) and that of the normal female subjects was 39.3 ± 13.8 years (range 20–63 years). The mean age of the male schizophrenia subjects was 31.0 ± 11.0years (range 18–64 years), while that of the female schizophrenia subjects was 35.0 ± 12.0 years (range 17–64 years). There were no significant differences in mean age between any of the groups.
The mean duration of illness in the patient group was 6.9 ± 8.0 years (range 0.1–43 years), and when divided by gender that of the male patients was 7.3 ± 9.0 years (range 0.1–43 years) and that of the female patients was 6.5 ± 7.3 years (range 0.1–31 years). There was no significant difference in illness duration between the patient groups.
The mean dose of neuroleptics (mg haloperidol daily) was 13.0 ± 9.2 mg (range 0–39.5 mg), and when divided by gender that of the male subjects was 16.0 ± 11.0 mg (range 0–39.5 mg), while that of the female subjects was 11.0 ± 7.3 mg (range 0.8–25.8 mg). The mean dose of neuroleptics tended to be higher in men (t = 1.958, d.f. = 58, P < 0.1).
Measurement of event-related potentials
Auditory oddball tasks were used. Pure tones of 1000 Hz and 2000 Hz were used as the stimuli and were provided to both ears at 20% target stimulus and 80% non-target stimulus. Either the 1000-Hz or the 2000-Hz tone was set as the target stimulus and counting was initiated with the target stimulus. Two target stimulus sessions were conducted. The stimulation intensity was 70 dBSL, the stimulation time was 100 ms and the stimulus interval was 0.5 Hz.
Electroencephalographic data (EEG) were detected using silver/silver chloride electrodes placed on both ear lobes and derived from the Fz, Cz and Pz using the international 10–20 system. For averaging, one session involving 30 events and two sessions involving 60 events were averaged. To measure eye movement, an electrode was placed on the lateral side of the left eye and responses > ±4 μV were excluded as artifacts. For EEG averaging, the grand averaged waveforms of each subject with a mean potential at 100-ms intervals prior to starting stimulation (taken as 0) were determined for both groups. The latency and amplitude of the maximum positive peak seen at 250–600-ms intervals with the low-frequency target stimulus were defined as P300 latency and amplitude and were determined for each subject.
For statistical analysis, P300 component data were categorized for both groups and means were compared by category using a t-test. For the P300 components, correlations of age with latency and amplitude were assessed using the Pearson correlation coefficient. An analysis of variance was conducted for both groups using diagnosis and age as independent variables and P300 latency as the dependent variable. In the schizophrenia group the correlation between illness duration and P300 components was also assessed. For statistical analysis we used SPSS (SPSS, Chicago, IL, USA). Significant level was set at P < 0.05 (two-tailed test).
Latencies and amplitudes of the P300 components
The grand averaged waveforms of the healthy control and schizophrenia groups are shown in Fig. 1, while means and standard deviations of the P300 components are shown in Table 1. P300 amplitudes for all electrode-positions (Fz, Cz, and Pz) were significantly attenuated in the schizophrenia group (Fz, t = 4.342, d.f. = 127, P < 0.01; Cz, t = 6.170, d.f. = 128, P < 0.01; Pz, t = 7.018, d.f. = 128, P < 0.01). At all electrode-positions the schizophrenia group had a significant prolongation of latency (Fz, t = −3.763, d.f. = 127, P < 0.01; Cz, t = −3.881, d.f. = 128, P < 0.01; Pz, t = −3.417, d.f. = 128, P < 0.01). When the healthy control and schizophrenia groups were divided by gender, there were no significant differences between male and female subjects in either group in regards to latency or amplitude of P300 at any of the electrode positions.
Table 1. P300 amplitude and latency
Normal control subjects
Total sample (n = 70)
Male (n = 39)
Female (n = 31)
Total sample (n = 60)
Male (n = 28)
Female (n = 32)
Correlations between age and P300 components
Table 2 shows the correlation of age with P300 components as assessed by Pearson correlation coefficient. In both healthy control and schizophrenia groups there was a significant positive correlation between age and P300 latency at all positions. We examined the correlation between age and P300 latency in the schizophrenia patients and normal subjects grouped by age. When we examined the younger group of schizophrenic subjects (<40 years of age) we observed significant positive correlations between age and P300 latency for both Fz and Cz (Fz, r = 0.346, P < 0.05, slope = 2.54; Cz, r = 0.337, P < 0.05, slope = 2.47) and a tendency to correlate at Pz (Pz, r = 0.230, P < 0.1, slope = 1.78). In contrast, there was no significant positive correlation at any position measured in the normal group (Fz, r = −0.219, P < 0.1; Cz, r = −0.293, P < 0.05; Pz, r = −0.179, P = 0.144). When we analyzed all younger subjects, a positive correlation between age and P300 latency was observed only in schizophrenia patients.
Table 2. Pearson correlations of P300 latency and amplitude with age
When schizophrenia and healthy control groups were analyzed by gender, men within both groups had a significant positive correlation between age and P300 latency at all positions. Women in the schizophrenia group also had a significant positive correlation between age and P300 latency at all positions, but women in the healthy control group had a significant positive correlation only at the Pz position.
Neither group had any significant correlation between P300 amplitude and age. Analysis of the groups by gender also failed to find any significant correlation between P300 amplitude and age.
Correlations between illness duration and P300 components
The correlation between illness duration and P300 components in the schizophrenia patients is shown in Table 3. There was a significant positive correlation between illness duration and P300 latency for all schizophrenia patients at all positions. However, when we examined the correlation between illness duration and P300 latency in schizophrenia patients grouped by age, in patients under 40 years of age there was no significant positive correlation between illness duration and P300 latency at any electrode positions (Fz, r = −0.205, P < 0.1; Cz, r = −0.196, P = 0.104; Pz, r = −0.035, P = 0.413).
Table 3. Pearson correlations of P300 latency and amplitude with illness duration
Difference in P300 latency from estimated mean of normal control subjects
When the schizophrenia group was subdivided by gender, male patients had a significant positive correlation between illness duration and P300 latency at all positions. There was no significant correlation between illness duration and P300 latency in female patients. There was also no significant correlation between amplitude and illness duration in patients with schizophrenia, and no correlation was observed when the patients were further subdivided by gender.
To further assess the correlation between illness duration and P300 latency in patients with schizophrenia, the difference calculated between the P300 latency predicted from the regression line in healthy controls versus the actual P300 latency in patients was correlated with illness duration. The results are shown in Table 3. Patients with schizophrenia had a significant correlation at all positions. In the schizophrenia group longer illness duration was associated with a prolongation of P300 latency beyond that predicted from the regression line of healthy controls for a given age. When the correlation was assessed based on gender, male patients had a significant positive correlation between illness duration and the difference in P300 latency as predicted from the regression line of the healthy controls at all positions. By contrast, among women there was no significant correlation between illness duration and differences in P300 latency as predicted from the regression of healthy controls.
Differences in the slopes of the regression lines for P300 latencies between schizophrenia patients and healthy controls
An analysis of variance was conducted for each position, with age and diagnosis used as independent variables and P300 latency as the dependent variable. The results indicated that interactions with age and diagnosis were significant for Fz (F = 3.042, d.f. = 3, 128, P < 0.05) and Cz (F = 4.117, d.f. = 3, 129, P < 0.01) and showed a tendency to correlate for Pz (F = 2.255, d.f. = 3, 129, P < 0.1). When differences in the slopes of the regression lines between the two groups were analyzed on t-test, the slopes of the schizophrenia group were found to be significantly steeper than those of the healthy control group at all positions (Fz, t = 2.478, d.f. = 125, P < 0.05; Cz, t = 2.658, d.f. = 126, P < 0.01; Pz, t = 2.032, d.f. = 126, P < 0.05). Therefore, age-associated prolongation of P300 latency was significantly greater in the schizophrenia group than the healthy control group.
A similar analysis of variance was conducted for the slopes of the regression lines for age and P300 latency with subjects divided by gender. The results indicated that in men the interaction with age and diagnosis was significant for Cz (F = 3.259, d.f. = 3, 66, P < 0.05) and a tendency to correlate was found for Pz (F = 2.566, d.f. = 3, 66, P < 0.1). When differences in the slopes of the regression lines between the two groups were analyzed on t-test, it was confirmed that for positions Cz and Pz the slope for men in the schizophrenia group was significantly greater than the slope for men in the healthy control group (Cz, t = 2.640, d.f. = 63, P < 0.05; Pz, t = 2.710, d.f. = 63, P < 0.01). By contrast, in women, only Pz had a significant correlation between age and P300 latency in the healthy female controls and Pz had no significant interaction for diagnosis and age (F = 1.432, d.f. = 3, 62, P = 0.243). When the slopes of the regression lines were analyzed on t-test, there was no significant difference between women in the schizophrenia group and those in the healthy control group (t = 0.408, d.f. = 59, P = 0.685). Therefore, male patients with schizophrenia had a greater age-associated prolongation of P300 latency than did healthy male controls; but this was not seen in women.
Prolongation of P300 latency in the schizophrenia group
This study clearly shows that in patients with schizophrenia the prolongation of P300 latency during aging is greater than in healthy controls. Also, the extent of the latency prolongation was related to illness duration. An earlier neuroimaging study indicated that among schizophrenia patients showing progressive changes in brain-imaging outcome, patients with a short duration of illness5 and young patients6,29 had a faster progression; for example, in the rate of ventricular enlargement. In this context, a longitudinal study reported that neuropathological changes based on magnetic resonance imaging occur early during the course of the disease,30 while a long-term follow-up study of the prognosis of schizophrenia patients reported that disease progression occurs within the first 5–10 years after onset.31 It has also been reported that patients with greater ventricular enlargement at follow up constitute a poor-prognosis group and show deterioration of their psychiatric status.7,14 This group with poor prognoses also had ventricular enlargement on their initial scans.6
The present study found a significant positive correlation between P300 latency and illness duration. Taking into account this result and that of the aforementioned studies that show that brain structural change may occur early in the course of the disease, we speculated that the functional change seen in P300 latency occurs later and as a consequence of early structural changes. However, it should be noted that the present study is a cross-sectional study using patients with various durations of illness, while the neuroimaging studies cited here were longitudinal and used standardized subjects. This speculation will therefore need to be confirmed in future studies.
We examined the correlation between age, illness duration, and P300 latency in subjects grouped by age. In the younger group (<40 years) no significant positive correlation between illness duration and P300 latency was found. The reason for the absence of a positive correlation in the younger group was due to the short mean illness duration (3.46 years) and small standard deviation (3.45). By contrast, we observed significant positive correlations between age and P300 latency in the same group of patients although there was no significant positive correlation in the normal group. When we analyzed all younger subjects, a positive correlation between age and P300 latency was observed only in schizophrenia patients. Therefore, along with disease progression, P300 prolongation must have already started in the younger group of patients.
There are only a few reports concerning the correlation between age and P300 latency of patients with schizophrenia.25–27 Mathalon et al. and Frangou et al. also found a correlation between illness duration and P300 latency.25,27 These reports are in agreement with the present study, that the prolongation of P300 latency during aging in a schizophrenia group is greater than that found in a control one. However, we think it noteworthy that when the present schizophrenia patients were divided on the basis of gender, that prolongation of P300 latency during aging was greater in the male schizophrenia group than the female one.
Possible cause of the observed gender difference
As described here, the prolongation of P300 latency associated with age and illness duration is more prominent in male than in female subjects, while there were no significant differences in age or illness duration between the male and female patient groups. This finding raises the possibility that the process of progression after onset of the disease differs between men and women.
There have been reports showing male–female differences in schizophrenia patients with respect to prognosis, morphological abnormalities in the brain, and results of neuropsychological testing. It has also been reported that female patients have a better social adjustment prior to diagnosis, a higher age at onset, a shorter duration of hospitalization, fewer re-hospitalizations, a better response to medication, and fewer frequent adverse effects as compared to their male counterparts.32 There are other reports showing that morphological brain abnormalities are more remarkable in male than female patients; for example, male patients have larger ventricles,33,34 a smaller temporal lobe,35 smaller thalamus,34 smaller left hippocampus,36 and a smaller left parahippocampal gyrus.37 It is also reported that male patients have an inferior attention function, verbal memory, and executive function compared to female patients.38 Thus, once male patients have developed schizophrenia, their symptoms and associated impairments tend to be more severe, respond less well to medication, and are associated with a poorer prognoses compared to female patients.
The estrogen hypothesis proposes that these gender differences are due to the neuroprotective effects of estrogen that begin during brain development in the fetus and continue into adulthood.39 Estrogen has also been reported to affect the dopaminergic system, where it decreases the sensitivity of dopamine receptors (downregulation) during brain development, resulting in the moderation of symptoms.40 It is also known that the incidence of schizophrenia increases in elderly women as postmenopausal estrogen secretion decreases. It is therefore possible that estrogen-mediated neuroprotective effects in the premenopausal period may resist the progression of schizophrenia in female subjects and that this may be the basis for the observed gender-related differences in correlations of age, illness duration, and P300 latency.
Absence of a correlation between age or illness duration and P300 amplitude
In the present study we found no significant correlation between age or illness duration and P300 amplitude. The only prior study that has reported a correlation between illness duration and P300 amplitude involved elderly patients with schizophrenia who had a negative correlation between these two variables.28 Many studies have reported P300 amplitude attenuation in schizophrenia patients and some report P300 latency prolongation in chronic ones. In patients with progressive neurodegenerative disease such as Alzheimer's, disease progression is better reflected in latency than in amplitude changes. We therefore speculated that the progression of a schizophrenic pathology, associated with illness duration, is more easily detected in P300 latency changes rather than in changes in amplitude. However, this speculation will need verification in future studies.
The present study clearly indicated that prolongation of P300 latency during aging in patients with schizophrenia is greater than in healthy controls and that the extent of latency prolongation is related to illness duration. Prolonged latency of P300, associated with age or illness duration, was more prominent in male than female schizophrenia subjects. This study is the first to demonstrate a gender difference in P300 prolongation associated with age and illness duration. Thus, P300 latency is prolonged as a function of illness duration and is more prominent in male compared to female schizophrenia subjects, suggesting gender-related differences in disease progression.