Correspondence: NoriakiSumi Department of Neuropsychiatry, Osaka Medical Center for Cancer and Cardiovascular Diseases, 1-3-3, Nakamichi, Higashinari-ku, Osaka 537-8511, Japan
Abstract Most studies on event-related potentials (ERP) in psychiatric illness or dementia have focused on the single-peak latency of ERP components. In the present study, not only peak latencies of ERP components (N1, P2, N2, and P3) but also interpeak latencies (IPL; N1–P2, P2–N2, and N2–P3) were analyzed using the auditory oddball task. Thirty-five senile depressed patients and 34 patients with dementia of the Alzheimer type (DAT) were compared to 39 age-matched healthy volunteers. The mean latencies of P2 and P3 were shorter in patients with senile depression than in controls. In DAT patients, the N2 and P3 latencies were longer. When the IPL was assessed, however, only the mean IPL of N1–P2 was shorter in patients with senile depression, while the P2–N2 IPL were longer in those with DAT. The IPL results suggest that in senile depression the early cognitive process is hastened and in DAT the middle process is disturbed. Based on these results, we conclude that IPL of the auditory ERP might be used to reveal the disturbed steps within the cognitive process.
Some components of event-related potential (ERP) appear to be correlated with such cognitive processes as sensory input, encoding or registration of sensory information, reference to the memory template, stimulus evaluation, response selection and cognitive context updating. A number of studies employing the oddball task (a task in which subjects are required to detect infrequent auditory stimuli) have indicated a probable correlation between the latency of ERP components and changes in information processing in the elderly, both of which are factors that are remarkably pronounced in mental illnesses generally and in dementia particularly.1–7 However, almost all of these reports have evaluated the single-peak latency of a single ERP component, without considering the interaction among several such components. If each peak of ERP is regarded as a ‘guidepost’ in the course of the cognitive process, then each single-peak latency can be regarded as time spent traveling from the start to the ‘guidepost’. Each interpeak latency (IPL), in turn, may show a passing speed through a step within the cognitive process. When the delayed latency of P3 is observed, a question arises: is the delayed P3 latency caused by the prolongation of N1–P2 IPL, the prolongation of P2–N2 IPL, the prolongation of N2–P3 IPL, or a complex prolongation of two or more IPL? The aim of the present study was to compare the efficacy of single-peak and interpeak evaluation, and to differentiate the cognitive difficulties in senile depression and DAT.
Three groups of patients were investigated: a depression group, a DAT group, and a control group. The depression group consisted of 35 patients (15 males and 20 females; mean age, 68.9 ± 5.0 years; range, 60–78 years; disease onset, over 55) with major depressive disorder or dysthymic disorder without organic brain illness. Subjects in this group were diagnosed according to DSM-IV criteria; all of them showed normal brain computed tomography or magnetic resonance imaging results and had no symptoms of dementia. The mean age of depression onset was 65.8 ± 5.9 years, and the mean score on the Hamilton Rating Scale for Depression (HRSD) was 21.7 ± 5.7 at the time of the present ERP examination. Ten subjects in the depression group were drug free, and the other subjects were receiving antidepressants or minor tranquilizers at the time of the study.
The DAT group consisted of 34 demented patients (16 males and 18 females; mean age, 70.0 ± 6.6 years; range, 60–84 years) diagnosed with probable Alzheimer's disease, based on NINCDS-ADRDA criteria. These were designated as cases of predementia or mild dementia, in keeping with the Hasegawa Dementia Scale (range, 10.5–27; mean 17.0 ± 4.0). Fifteen subjects in the DAT group were drug free, but the other subjects were receiving therapeutic drugs (e.g. minor tranquilizers and/or low-dose major tranquilizers) for their insomnia, anxiety or agitation. No statistically significant differences between subjects with and without therapeutic drugs were found on the ERP results in either the depression group or the DAT group. Finally, 39 healthy volunteers were investigated as normal controls (18 males and 21 females; mean age, 68.5 ± 4.9 years; range 60–77 years). All subjects were right handed and had no physical difficulties that interfered with their performance of the trials in this study. All subjects gave their informed consent to participate in this study.
Electroencephalogram (EEG) activity was recorded from midline scalp electrodes at Fz, Cz and Pz referenced to linked earlobes. Skin impedance for each electrode site was measured at below 5.0 k ohm. The EEG was amplified (1 × 105) with a bandpass of 0.1–100 Hz. Electro-oculogram (EOG) was recorded from electrodes placed above and below the right eye. Using a standard auditory oddball paradigm, the subject was instructed to press a button with his/her right thumb whenever rare target tones occurred among a series of frequent non-target tones. The stimulus was a pure tone delivered binaurally through light headphones at an intensity of 70 dB, with a duration of 50 ms sound pressure level (SPL) (rise and fall of 5 ms) and with an interstimulus interval of 1.6 s. The target rare tones and the non-target frequent tones, 1 kHz and 2 kHz, were presented at probabilities of 20% and 80%, respectively. The analysis time was 1000 ms beginning at 200 ms before stimulus onset with a sampling ratio of 256 Hz for each electrode site. EOG (over 100 μV) contaminated trials were automatically rejected from ERP averaging. Recording was continued until 30 ERP to target tones and 30 ERP to non-target tones had been averaged.
Peak latencies for N1, P2, N2 and P3 at the Pz electrode site were measured from stimulus onset to the point of maximum voltage and used in the following analysis. Interpeak latencies of N1–P2, P2–N2 and N2–P3 were computed by subtracting the former peak latency from the latter peak latency. Group differences in single-peak latencies and in IPL were analyzed using one-way analysis of variance (ANOVA) followed by the Scheffé's test. To evaluate the ERP measures in patient groups, a criterion was set at 2.0 standard deviations from the means of the normal control group. To assess the relationships between each ERP latency and the HRSD scores in the depression group, and between each ERP latency and the Hasegawa Dementia Scale (HDS) scores in the DAT group, the coefficient of product-moment correlation was calculated.
Mean latencies and standard deviations of ERP components at Pz for subjects from each group are given in Table 1. Significant differences among groups were found for each of the P2, N2 and P3 peak latencies (P2: F2,105 = 16.4, P < 0.001; N2: F2,105 = 42.6, P < 0.001; P3: F2,105 = 40.2, P < 0.001). The depression group had shorter mean latencies of P2 and P3 than the control group (P2: P < 0.01; P3: P < 0.05). The DAT group had longer mean latencies of N2 and P3 than the controls (N2: P < 0.01; P3: P < 0.01).
Table 1. The mean latencies of the single peaks (and SD) for each group
In Fig. 1, individual P3 latencies for each group are plotted. In the depression group, four of 35 (11%) depressed subjects had a shorter P3 latency than 2 SD under the mean of the control group (shorter basis), but no individual had a longer P3 latency than 2 SD above the mean of the control group (longer basis), and 14 of 34 (41%) demented subjects had P3 latencies longer than the longer basis.
Mean IPL and standard deviations in each group are given in Table 2. Significant differences among groups were found for each of the N1–P2 and P2–N2 IPL (N1–P2: F2,105 = 14.1, P < 0.001; P2–N2: F2,105 = 18.7, P < 0.001). The depression group had a shorter mean IPL of N1–P2 (P < 0.01) than the control group, but no significant differences were found in P2–N2 or N2–P3 IPL. In the DAT group the mean IPL of P2–N2 was longer than that of the control group (P < 0.01). Individual IPL (N1–P2, P2–N2, and N2–P3) in each group are plotted in Fig. 2. For N1–P2 IPL, four of 35 (11%) depressed subjects had a shorter IPL than the shorter basis, but no individual had a longer IPL than the longer basis. In the DAT group, three of 34 (9%) subjects had a longer IPL than the longer basis. For P2–N2 IPL, one of 35 (3%) depressed subjects had a shorter IPL than the shorter basis, while eight of 35 (23%) depressed subjects and 21 of 34 (62%) demented subjects had longer IPL than the longer basis. Finally, N2–P3 IPL were shorter than the shorter basis in five of 35 (14%) depressed subjects longer than the longer basis in one of 35 (3%) depressed subjects and five of 34 (15%) demented subjects.
Table 2. The mean interpeak latencies (and SD) for each group
Event-related potential components and clinical scores
Using the coefficient of product-moment correlation, relationships between each ERP latency and the HRSD scores in the depression group, and between each ERP latency and the HDS scores in the DAT group were assessed. In the depression group, no statistically significant correlation was found. In the DAT group, negative correlations were found between HDS and both the N2 and P3 latencies (N2: r = – 0.376, P < 0.05; P3: r = – 0.353, P < 0.05).
The results of the present study indicate that both the single-peak latency and the IPL measure were useful to differentiate the pathological groups from the normal control group, while the single-peak latency and IPL values varied in a peculiar manner within each pathological group. In the depression group, the means of P2 and P3 peak latencies were significantly shorter than those in normal controls; however, among IPL, only the mean of N1–P2 IPL was shorter, and the other IPL showed no significant differences. Recently, Patterson et al. found no significant differences in single-peak latencies between senile depressed subjects and normal controls; however, the reported means of P2, N2 and P3 latencies at Pz in senile depression (P2: 182.5; N2: 238.5; P3: 364.0 ms) tended to be shorter than the means in normal controls (P2: 191.5; N2: 256.8; P3: 377.9 ms).5 Kraiuhin et al. also reported a shorter mean P3 latency in senile depression (371 ms) than in normal controls (386 ms).8 These studies did not assess the IPL, but the results of single-peak latencies were consistent with those of the present study. The early components, N1 and P2, have been considered to consist of not only exogenous evoked potentials but also some endogenous cognitive potentials.9
In the present study, it remains unclear why N1–P2 IPL in the depression group was shortened, but it may be possible that such early cognitive processes as sensory input or initial encoding of sensory information were hastened in senile depression. However, no significant correlation was found between each ERP latency and HRSD score in the depression group. It is thus possible that these alterations of ERP latencies in depression, especially shortening of N1–P2 IPL, depend on the trait of depressive patients rather than the degree of depressive state.
In the DAT group, the mean latencies of N2 and P3 were longer than those in the normal controls. The N2 component consists of the mismatch negativity (MMN, N2a) and N2b.10,11 Näätänen and Picton described in their review that the MMN was regarded as an automatic cerebral response to a stimulus physically deviating from the stimuli of the immediate past, and that N2b was considered to be a component associated with controlled processes and elicited by unexpected stimuli.12 It has also been considered that the process manifested by the P3 component may be associated with the updating of the schema.13,14 For the single-peak latencies, it is generally accepted that N2 and P3 are later when the target stimulus is more difficult to discriminate and it has been reported by many authors that the N2 and P3 latencies are prolonged in dementia1,3,4,6,7 and in DAT with memory disturbance.15,16 These previous results suggest that both the ‘context updating’ process and stimulus discrimination process are delayed in dementia. In the present study, however, the results of the single-peak latencies differed from those of the IPL. Among IPL, only the mean IPL of P2–N2 was lengthened; the other IPL were not different from those of the controls. It is likely that the prolongation of the P3 latency in the DAT group was caused by the prolongation of the P2–N2 IPL. It is likely that the prolongation of the P3 latency in the DAT group was caused by prolongation of the P2–N2 IPL (i.e. by relative prolongation of the N2 latency to the P2 component). It would appear that, in the DAT studied, the period from the end of sensory input to stimulus discrimination was prolonged while sensory input and updating of the cognitive context may have been reserved. These findings may suggest that DAT patients have difficulty analyzing the meaning of stimuli. In the present study, both N2 and P3 single-peak latencies were negatively correlated with HDS, but no IPL was significantly correlated with HDS. P3 latency has been reported to show negative correlation with scores on the Wechsler Adult Intelligence Scale (WAIS)4 and Mini-Mental State (MMS).17 In the present study, DAT subjects probably found it increasingly difficult to discriminate the target tone according to their degree of dementia. However, because IPL were not correlated with HDS, further investigation will be required into the relationship between IPL and other scales (e.g. MMS or WAIS), as well as between IPL and degree of memory disturbance or clinical symptoms.
However, some studies have suggested that ERP is not a sufficiently sensitive index for use in clinical diagnosis of demented individuals.2,5 In the present study, the sensitivity was 41% by means of P3 single-peak latency. This sensitivity was considered inadequate to diagnose DAT. By means of P2–N2 IPL, the sensitivity was improved to 62%; however, 23% of patients in the depression group had longer P2–N2 IPL than controls. Thus, the specificity was not adequate for clinical use. In order to use the ERP technique for clinical diagnosis, further improvements will be needed in both sensitivity and specificity. Nonetheless, the ERP technique may be valuable for assessing cognitive disturbance in psychiatric illness or dementia, and IPL measurement in particular may be useful to investigate the disturbed steps in the cognitive process of these diseases.