181

Psychiatry Research, 44:181-191

Elsevier

Intensity Dependence of Auditory Evoked Potentials and Clinical Response to Prophylactic Lithium Medication: A Replication Study Ulrich Hegerl, Hilde Wulff, Received

and Bruno Miiller-Oerlinghausen

March 31.1992; revised version received July 24, 1992; accepted September

6, 1992.

Abstract. A predictor of clinical response to prophylactic lithium treatment in affective psychoses would be of considerable importance. In a pilot study, responders to prophylactic lithium medication, as compared with nonresponders, were characterized by a steeper slope of the amplitude/stimulus-intensity function (ASF slope) of the Nl/ P2 component of the auditory evoked potential. We tried to replicate this finding in 34 stabilized outpatients with affective illness who had been treated with lithium for at least 3 years. As in the pilot study, responders were again characterized by steeper ASF slopes than nonresponders. Since a steep ASF slope seems to indicate low central serotonergic function, it is speculated that a steep ASF slope characterizes those patients with a serotonin deficit who respond to serotonin agonists like lithium. Key Words. Affective disorder, electrophysiology, reducing, major depression, average evoked response,

serotonin, augmenting/ psychopharmacology.

It is estimated that 20-45s of patients with recurrent affective psychoses fail to respond to prophylactic lithium therapy (Greil and Schiiderle, 1986). Since the prophylactic response can be evaluated only after a long medication period, predictors of the individual response would be of great value. The intensity dependence of sensory evoked potentials (EPs) appeared to show promise in this context, because studies using the augmenting/reducing paradigm proposed by Buchsbaum and his associates (Buchsbaum and Pfefferbaum, 1971; Buchsbaum et al., 1983) found that patients whose EPs showed a pronounced intensity dependence (i.e., augmenters) were responders to acute antidepressant treatment with lithium (Buchsbaum et al., 1971, 1979; Baron et al., 1975; Nurnberger et al., 1979; Zerbi et al., 1984). Although these early studies were heterogeneous in their methodology, the results were intriguing enough to raise the question as to whether the intensity dependence of sensory EPs might also be related to the response to prophylactic lithium therapy.

Ulrich Hegerl, M.D., Privatdozent, is Lecturer in Psychiatry and Clinical Psychophysiology, and Supervising Psychiatrist at the Berlin Lithium Clinic, Department of Psychiatry, Freie Universitit Berlin. Bruno Miiller-Oerlinghausen, M.D., is Professor, Laboratory of Clinical Psychophysiology (Chief), Department of Psychiatry, Freie Universitst Berlin. Hilde Wulff is Research Fellow, Berlin Lithium Clinic. (Reprint requests to PD Dr. med. U. Hegerl, Psychiatrische Klinik der Freien Universitgt Berlin, Labor fiir Klinische Psychophysiologie, Eschenallee 3, D-1000 Berlin 19, Germany.) 0165-1781/92/%05.00

@ 1992 Elsevier Scientific

Publishers

Ireland

Ltd.

182 In a pilot study, we found that the slope of the amplitude/stimulus-intensity function (ASF slope), indicating the intensity dependence of auditory EPs (AEPs, NI /P2-component), was steeper in responders to lithium prophylaxis than in nonresponders (Hegerl et al., 1987). Subsequent studies on healthy subjects revealed that the ASF slope shows a satisfactory long-term stability and is not influenced in a relevant manner by variables like age and sex (Hegerl et al., 1988). These results indicate that the ASF slope may be clinically useful as a predictor of individual response to prophylactic lithium medication. We therefore performed a replication study to confirm the hypothesis that a steep ASF slope of the AEP Nl/P2 component is related to a favorable outcome of lithium prophylaxis.

Methods Thirty-four outpatients without affective symptoms (Beth-Rafaelsen score for mania or melancholia < 4) at the time of recording were investigated. All patients had been

Subjects.

treated continuously with lithium for at least 3 years. These patients were divided into responders and nonresponders (Table 1). Responders (n = 19) were those without any recurrence in the preceding 3 years, and nonresponders (n = 15) were those with one or more recurrences leading to hospitalization (n = 8) or psychopharmacological intervention in the preceding 3 years (n = 7). Recurrences and psychopathology are monitored by a standardized assessment protocol in the Berlin Lithium Clinic. Fifteen of the 34 patients had also participated in the pilot study. AEP recordings were performed after the patient made a routine visit to the clinic. Venous blood was drawn for the assessment of lithium plasma concentration.

Procedure.

lithium

Table 1. Clinical characteristics of the responder and the nonresponder group Nonresponders Reponders i-l

19

Men Women

15

8

3

11

12

50.9 * 14.5

51.6 + 13.8

0.75 k 0.12

0.65 -t 0.14

Neuroleptics

1

2

Antidepressants

1

1

Neuroleptics

0

1

8

5

15

10

Age (mean + SD; yr) Lithium plasma concentration Patients with additional

(mean f SD; mmolil)

medication

plus antidepressants

Others Diagnoses Bipolar depression Unipolar depression

4

2

Schizoaffective

0

3

disorder

Patients with residual affective symptomatology (score < 4) Beth-Rafaelsen Derxession

Scale Mania

2

1

1

2

183 The patient’s physician at the lithium clinic used the Beth-Rafaelsen Scale to assess affective symptomatology. Sensation thresholds to click stimuli were monitored before AEP recording to rule out deficits in auditory acuity. AEP Recording and Data Processing. Recordings were performed in a sound-attenuated room adjacent to the recording apparatus. Patients were sitting with their eyes open in a slightly reclining chair with a head rest. A black disk on the wall, 3 meters in front of the patients, was presented as a point of orientation, although no strict fixation was demanded. Binaural clicks (stimulus duration = 0.9 msec, interstimulus interval = 2.1 set) at four levels of intensity (52, 62, 72, and 82 dB HL) were presented through headphones in random order (Pathfinder II, Nicolet). We recorded with gold-plated cup electrodes from Cz, C3, and C4, and from 1 cm above the external canthus of the left eye, referenced to linked mastoid electrodes. An abrasive paste (Omni Prep) and an electrode paste (Teca) were used to keep the electrode impedances below 5 kOhm. Bandpass filters were set to 1 and 30 Hz (two-pole Butterworth filters, 12 dB/octave roll-off). We recorded from 100 msec prestimulus to 300 msec poststimulus with a sampling rate of 1280 Hz. Eighty responses per intensity were averaged. Responses to the first 30 clicks were excluded to reduce habituation effects. To suppress artifacts, all trials were automatically excluded from averaging whenever, in any of the four leads, the voltage exceeded + 50 p V. A cursor program was used to determine the N 1 component as the most negative amplitude value during the 65 to 125-msec period and the subsequent P2 component as the most positive amplitude during the 120-to 220-msec period. In addition, the individual Nl latency and Nl/P2 amplitude were measured based on the average of the four AEPs that corresponded to the four intensity levels. Fig. 1 presents examples of the AEPs of two patients. A measure of the intensity dependence of the Nl/ P2 component was obtained by fitting a straight line to the amplitude values at each intensity level using the least square technique. The slope of the line (ASF slope) thereby indicates the Nl/ P2 amplitude change due to increasing stimulus intensity.

Fig. 1. Two examples for auditory evoked potentials to the 4 stimulus intensities pat.

Fr.

dl3

I

I

1

-100-60 latency

0 -3

100

160

0

260 (ms) latency

i 100

11

11 160

L 260 (ms)

-3

The responder to lithium prophylaxis (patient He) shows a pronounced amplitude increase with increasing stimulus intensity (steep ASF slope). This is not the case for the nonresponder (patient Fr).

184 The median slope was used as a further measure of the intensity dependence of the N I/ P2 component. The median slope was calculated from the slopes of all possible connections between the four amplitude values corresponding to the four intensities (n = 6, 52-62, 52-72, 52-82, 62-72, 62-82, and 72-82 dB). In another study of 33 healthy subjects, retested after 3 weeks, a test-retest reliability of 0.77 (Cz) was found for the median slope and of 0.74 for the ASF slope (Hegerl, 1992). Recording and processing of AEP data were performed outside the lithium clinic by a person (H.W.) who was not involved in the treatment of the patients and who was largely unaware of the responder/ nonresponder status of the patients.

Results Table 1 presents clinical characteristics of the responders and nonresponders. Lithium plasma concentrations were significantly higher in responders (t test: p < 0.05). No significant differences in the artifact rate were found between responders and nonresponders (responders: Ill+ SD 86; nonresponders: 169 + SD 143; 2 test: p = 0.17). As hypothesized, responders showed significantly steeper ASF slopes (i.e., augmenting) than nonresponders (Fig. 2, Table 2); this effect was confirmed by analysis of variance with the repeated measures factor “lead” (Cz, C3, C4) and the group factor “response” (F = 4.72; df = 1, 32; p < 0.05). Fig. 3 presents the individual ASF slopes (Cz) for responders and nonresponders. When the median slope was used, steeper slopes were again found for the responders (F = 5.67; df = 1, 32; p < 0.05, Table 2). When only the 19 patients who had not participated in the pilot study 2 years ago were considered, responders (n = 13) were again characterized by steeper slopes than nonresponders (n = 6, Table 3). This difference

Fig. 2. Amplitude/stimulus intensity function of responders and nonresponders to prophylactic lithium treatment 14.0

C3

13.0

2

c4

12.0

1:;;;

J

/

6.0,.

5.0-

I 52

62

72

I 62

1 52 rtimulur

More amplitude nonresponders.

increase with increasing

62

Intsnrlty

12

(da

stimulus intensity is found in

I

I

62

52

8 62

12

62

HL)’

responders

(steeper ASF slopes) than in

185

could not be confirmed by analysis of variance in this smaller subgroup (ASF slope: F= 2.36; df = I, 17;~ = 0.14; median-slope: Fr2.94; df= 1, 17;p=O.10). In the pilot study, shorter Nl latencies had been found for responders than for Table 2. ASF slopes (,v V/10 dB), median slopes (,~V/10 dB), and Nl latency (msec) of responders (n = 19) and nonresponders (n = 15) Nonreswnders

Resmnders ASF

slope

2.1 It 1.1

1.4 f

1.1

c3

1.4 +

0.7

0.8 +

0.8

c4

1.6 +

0.7

l.Of

0.7 1.0

Cz

Median slope Cz

2.1 f

1.1

1.3f

c3

1.4 +

0.7

0.8 +

0.9

c4

1.5 +

0.7

l.Of

0.8 8.1

Nl latency

Cz

91.2 +

8.6

87.0 +

c3

91.9 *

9.2

89.8 + 10.0

c4

90.1 + 10.9

92.5 * 10.7

Analysis of variance F=4.72; df= 1,32;p