Psychopharmacology (1991) 105:361-367 003331589100219W
Psychopharmacology © Springer-Verlag 1991
Pharmacology of saccadic eye movements in man 1. Effects of the benzodiazepine receptor ligands midazolam and flumazenil David M. Ball, Paul Glue, Sue Wilson, and David J. Nutt
Reckitt and Colman PsychopharmacologyUnit, School of Medical Sciences, University Walk, Bristol BS8 1TD, UK Received September 7, 1990 / Final version February 14, 1991 Abstract. A paradigm for assessing benzodiazepine re-
ceptor sensitivity was developed using intravenous midazolam in normal volunteers. After administration of incremental doses of midazolam, alterations in saccadic eye movement parameters and psychological self ratings were assessed. Significant changes included dose-dependent slowing of peak velocity, peak acceleration, peak deceleration, reduced saccade acceleration/deceleration ratio and saccade accuracy, and increased sedation selfratings. Changes in saccade variables and sedation ratings were significantly correlated, and also correlated with plasma midazolam concentrations. No significant changes were seen in saccade latency or anxiety selfratings. Pharmacological specificity of these changes was demonstrated by their reversal with the benzodiazepine antagonist flumazenil. This challenge paradigm appears to be a sensitive means of assessing benzodiazepine receptor function in man. Key words: Benzodiazepine receptors - Flumazenil Midazolam - Saccadic eye movements
Although the pharmacology and therapeutic efficacy of benzodiazepines are well established, it has been more difficult to quantify benzodiazepine receptor function in man. Techniques have included subjective and performance psychological testing, changes in endocrine and biochemical parameters, radionucleide imaging, and physiological measurements. For instance, the sedative and anxiolytic effects of benzodiazepines are well known. Sedation has been shown to alter in a dose-dependent manner following administration of benzodiazepines and the magnitude of these effects may provide a crude index of receptor sensitivity (Hommer et al. 1986). A more sensitive measure may be provided by alterations in various performance tasks, such as critical flicker fusion or digit substitution tests, although there is little work on Off'print requests to:
P. Glue
whether these changes are dose-dependent (Wittenborn 1979; Hindmarch 1980). Benzodiazepines increase plasma concentrations of growth hormone (Shur et al. 1983; Hommer et al. 1986), and reduce noradrenaline (Hossman et al. 1980; Roy-Byrne et al. 1989), although these changes are too variable to provide quantitative information on receptor function. Recent positron emission tomography (PET) studies using radiolabelled benzodiazepine agonists and antagonists in man have provided interesting preliminary data on in vivo receptor binding, but not function, in man (Persson et al. 1985). However, PET scanning may be used to give some functional information, as demonstrated recently by looking at the effects of benzodiazepines on 18F-2-deoxyglucose uptake in man (Buschbaum et al. 1987). Perhaps the most interesting recent investigation into assessment of benzodiazepine receptor function looked at the effect of benzodiazepines on parameters of saccadic eye movement. Saccadic eye movements (from the French "saccade", meaning jerk) are rapid, steplike, conjugate changes of gaze, the purpose of which is to centralise objects of interest on the fovea. Once initiated, these movements have minimal cognitive or conscious input. They are stable and reproducible within subjects, both between tests and within a testing period (Mercer et al. 1990; Wilson et al., submitted manuscript). They have been reported to provide a superior measure of performance impairment than, for instance, critical flicker fusion or digit-symbol substitution tests (Mercer et al. 1990; Glue 1991). Hommer et al. (1986) showed that diazepam produced an easily quantifiable and reliable dose-dependent slowing of peak velocity of saccadic eye movements. A strong negative correlation was noted between saccade velocity and plasma diazepam concentration. The purpose of the present study was to develop a benzodiazepine challenge paradigm, with improved patient acceptability in terms of comfort and test duration, that would be suitable for outpatient studies. To this end, we used the water-soluble benzodiazepine, midazolam, which has a much shorter half-life than diazepam, and
362 p r o d u c e s fewer a d m i n i s t r a t i o n p r o b l e m s , s u c h as p a i n o n i n j e c t i o n . A d d i t i o n a l l y , H o m m e r et al. (1986) o n l y des c r i b e d t h e effects o f d i a z e p a m o n p e a k s a c c a d e velocity, w h i l e we also e x a m i n e d the effect o f m i d a z o l a m o n o t h e r p a r a m e t e r s o f s a c c a d i c eye m o v e m e n t , s u c h as a c c e l e r a t i o n a n d d e c e l e r a t i o n . M o r e o v e r , to d e m o n s t r a t e t h e p h a r m a c o l o g i c a l specificity o f b e n z o d i a z e p i n e - i n d u c e d s l o w i n g o f s a c c a d i c eye m o v e m e n t s , we r e v e r s e d the effects o f m i d a z o l a m w i t h the b e n z o d i a z e p i n e a n t a g o n i s t f l u m a z e n i l ( H a e f e l y a n d H u n k e l e r 1988). F i n a l l y , s a c c a d e measurement was performed using an electrooculog r a p h i c s y s t e m , w h i c h is t e c h n i c a l l y easier t h a n the inf r a r e d o c u l o g r a p h i c m e t h o d u s e d b y H o m m e r ( M e r c e r et al. 1990).
Materials and methods Subjects. Fourteen healthy, drug-free volunteers (ten males and four females, age 28.9± 1.2 years) took part in the study following informed voluntary consent. The study was approved by the local Ethics Committee. All subjects were light social drinkers and abstained from alcohol for 24 h prior to testing. Physical examination and routine blood investigations prior to testing were satisfactory. Experimental design. Testing was carried out in a quiet testing room in an outpatient research unit. On the day of testing subjects attended the unit after following their normal daytime routine. Butterfly cannulae were inserted into each forearm, and their patency maintained with heparinised saline. Subjects rested semi-supine throughout the test, and were not allowed to sleep. The testing schedule is summarised in Fig. 1. At each time point saccadic eye movements and subjective ratings were recorded (see below for details). Baseline (i. e. pre-infusion) assessments were recorded at - 15, - 8 and - 3 rain. Following intravenous infusion of saline vehicle at time 0 min, recordings were taken at 3, 8 and 13 rain. At 15 rain an infusion of midazolam 6 gg/kg was administered and recordings performed at 18, 23 and 28 min. A second infusion of midazolam 6 gg/kg at 30 min was followed by recordings at 33, 38 and 43 min. The third infusion ofmidazolam 12 ~tg/kg, given at 45 min, was followed by recordings at 48, 53, 58 and 70 rain. At 75 rain flumazenil 500 ~tg was infused and the final recordings taken at 78, 90 and 100 min. All infusions were made up to 10 ml in normal saline, and given over 60 s. Blood for midazolam concentrations was sampled at 29, 44 and 60 min through the cannula opposite to that used for drug administration. Time (rain). ~ e -15 -8 | ~ d ~ , visual a n a l ~ u e -3 saline ~ 0 3 saccades, visual analogue 8 13 rnidazolam 6~g/kg ~ 15 18 saccades, visual analogue 23 midazolam level 28 midazolam 6~g/kg ~ 30 33 saccades, visual analogue 38 midazolam level 43 midazolam 12~g/kg ~ 45 48 saccades, visual analogue 53 midazolam level 58 7O flumazenil 500~g ~ 75 78 saccades, visual analogue 9O I00 Infusions
~
~ ~
I ~
Fig. 1. Testing schedule
ratings
ratings
ratings
ratings
ratings
ratings
Subjective psychological responses were recorded on visual analogue scales. Subjects were asked to rate feelings of sedation and anxiety on 100 mm scales, with 0 mm representing "not at all ..." and 100 mm "the most ... ever".
Measurement of eye movements. The CSGAAS5 system we used has been fully documented elsewhere (Marshall et at. 1985; Mercer et al. 1990; equipment available from Dr. R. Marshall, Department of Pharmacology and Therapeutics, University of Wales College of Medicine, Heath Hospital, Cardiff) but a brief outline will be presented here. An electrooculographic (EOG) method is used to measure lateral eye movements. In this method, electrodes near the eyeball detect changes in electrical activity caused by movement of the eye, and therefore of the electrical field due to the corneo-retinal potential. These changes are relatively large compared with other electrophysiological signals recorded from the same area (eye movements of 40 degrees often produce potential differences across the eyeball of about 1 mV recorded at the surface) and are therefore easily detected. The EOG potential is proportional to the angle of eyeball displacenaent up to about 45 degrees (see Mercer et al. 1990). We use silver/silver chloride disposable neonatal E K G electrodes (Medicotest, Denmark) with a small amount of electrode gel, although traditional EEG cup electrodes have similar recording characteristics. These are placed 1 cm laterally to the outer canthus of each eye and on the glabella, after the skin has been scarified with a little Skinpure cream (Nihon Kohden). Electrode impedances are measured and confirmed to be less than 5 kohm. They are connected to a Fylde 260 IA optically isolated DC amplifier with a gain of x t000 and allowed to settle for a few minutes so that the standing potentials on the electrodes are equalized. Output from the amplifier is sampled 256 times per second via an analog to digital converter with a resolution of 4096 units, and the resulting digital information sent to an IBM-compatible personal computer for analysis. Only lateral saccades are studied, because the movement of the eyelids during vertical eyeball movements significantly alters the EOG amplitude in a non-linear way. Saccades are elicited by asking the subject to watch an array of light emitting diodes (LEDs) (dominant frequency 690 nm) placed at a distance of 67 cm from the glabella, level with the subject's eyes, such that a movement of the eyes from one end of the display to the other represents 40 degrees of arc. Head movement is prevented by supporting the subject's head and neck with pillows. To minimise the risk of "startle saccades", which have a different pathway of generation from anticipated saccades (see Wurtz and Goldberg 1989), subjects are asked not to anticipate movement of the LED stimulus. Those which do occur are rejected during data analysis (see below). Initially the central LED is illuminated and the subject asked to fixate on it so that the amplifier DC offset may be set to zero. The room lighting is then reduced and the target sequence started. The subject is asked to look at the LED that is illuminated and move his eyes to the new position when that LED is extinguished and the next is lit. The program causes a LED on either side of the midline to be lit alternately for 1.5 s, starting with the LEDs at the outer extremes of the display (a saccade angle of 40 degrees), and gradually using those nearer the midline so that saccade angles of 35, 30, 25, 20 and 15 degrees are elicited. Only LEDs at the end of the saccade path are turned on, with no intervening LEDs illuminated. The LED stimuli do not appear to be in apparent motion. It has been found that a sequence of 24 saccades (taking about 36 s) can be performed before any noticeable fatigue occurs, and on each occasion we record two such sequences with a brief rest in between so that a sufficient number of saccade parameters may be obtained to allow the programme to analyse the EOG data (see below). Next, the program analyses the 48 individual digitized EOGs from the two sets of saccade sequences. These data are smoothed using a digital filter, and each data set is validated in the following way. The baseline before each saccade is checked, using coefficient of variance, allowing rejection of data sets when the subject's eyes were moving at the time the stimulus changed position, or when there were artefacts from other sources. In the remaining sets the
363 program determines whether the change in baseline caused by the saccade occurs within 200 ms of the stimulus change (this can be altered by the user). These sets are also rejected, on the basis that the subject was anticipating the stimulus change. If required additional manual rejection of data sets may be performed. The remaining data sets are now analysed further. The program measures the first saccade peak velocity by differentiation of the displacement curve of the EOG, peak acceleration and peak deceleration by further differentiation, and latency from target movement to saccade onset. The ratio between the time spent in the acceleration phase of the saccade and that spent in the deceleration phase (acceleration/deceleration ratio or a/d ratio) is calculated. Accuracy of saccades is determined by making the assumption that the eye position at the end of the 1.5-s epoch is on the target, and comparing it with the displacement at the end of the first saccade. Secondary, or correcting saccades may be included in the analysis if required; we have excluded these in the present study. The relationship between angle of eyeball displacement and measurements such as peak velocity is important, because it is this relationship which remains intact even when voluntary control of saccades is attempted, and is normally referred to as the main sequence (Baloh et al. 1975). Each of the measures derived above is therefore plotted against angle of displacement and a curve fitted to the values. It has been shown that this curve reaches asymptote at about 35 degrees so all values used for statistical analysis have used interpolation at 35 degrees. Data from other saccade angles are not reported on.
Midazolam assay. Blood samples were placed in lithium heparin tubes, immediately stored in ice, and centrifuged as soon as possible afterwards. Plasma was stored at - 2 0 ° C until analyzed. Plasma midazolam was extracted in a two-stage solvent technique, using flurazepam as an internal standard to monitor recovery. This was between 50 and 75%. Midazolam concentrations were assayed by gas liquid chromatography with specific nitrogen detection. Assay sensitivity was approximately 1-2 ng/ml in 1 ml extracted plasma. lntraassay coefficient of variability was 5 10% over a range of 0-200 ng/ml, while interassay coefficient of variability was 5 %. All samples from each subject were analyzed in a single run. Statistics. Group means of each parameter were calculated at individual time points. One-way analysis of variance (ANOVA) was performed on the post-saline infusion data points, to establish the stability of these recordings. Changes in variables after each infusion were compared with the post-saline group mean using Student's t test. Correlation of variables at each dose of midazolam was by Pearson's R. The SAS statistical package was used for all analyses.
Results
accel: F = 0.04, P = 0.99; decel: F = 0.09, P = 0.96). After each midazolam infusion significant dose-dependent slowing was noted. After the flumazenil infusion, values returned towards baseline although they were significantly lower (accel: t = 4 . 1 , P < 0 . 0 1 ; decel: t = 2 . 4 , P < 0 . 0 5 ) . A / D ratio values were not altered after saline infusion ( A N O V A : F = 0 . 1 3 , df=3, P = 0 . 9 4 ) . After midazolam, the a/d ratio became smaller, that is, the time spent in deceleration increased m o r e than time spent in acceleration. Again this was dose dependent, and was reversed by flumazenil, with values not significantly different f r o m baseline ( t = 0.5, P = ns) (Fig. 2c). M e a n saccade latency was not significantly altered c o m p a r e d with baseline over the duration of the study (Fig. 2d). Saccade accuracy was not changed after saline infusion ( A N O V A : F = 1.72, df= 3, P = 0.17). Saccade accuracy values fell after each dose of midazolam (indicating saccade undershoot), although there was a rapid return to baseline values before subsequent infusions (Fig. 2e).
Psychological self~ratinos Mean self-ratings of sedation were unaltered after saline infusion ( A N O V A : F = 0 . 0 9 , df=3, P = 0 . 9 7 ) , and increased after each infusion o f midazolam (Fig. 3). Unlike saccade parameters, sedation ratings continued to increase after each dose of midazolam, and only at the last recording after the highest dose of midazolam did they start to fall. Following the administration of flumazenil, recordings returned to baseline. Midazolam did not significantly change anxiety ratings, although there was a trend towards reduced anxiety and a brief nonsignificant increase after flumazenil (Fig. 3).
Midazolam concentrations Midazolam concentrations increased in a dose dependent fashion after each administration (dose 1 (6 ~tg/kg): 13.7 ± 2.0 ng/ml; dose 2 (6 gg/kg): 27.6 ± 3.6 ng/ml; dose 3 (12 gg/kg): 50.8 i 5.7 ng/ml) (all values mean 4-SEM). Although the assay was capable of detecting the ahydroxy metabolite, this was not detected in the assay. Correlations of the above variables are summarised in Table 1.
Saccade parameters Discussion Mean peak velocity values over the course of the study are shown in Fig. 2a. N o change was noted after saline infusion ( A N O V A F = 0.03, P = 0.99). Each midazolam infusion produced significant dose-dependent slowing o f peak velocity, the greatest effect being seen immediately after each infusion, with subsequent recordings tending to be closer to baseline levels. After flumazenil infusion, peak velocity recordings were significantly lower than at baseline (t--3.8, P < 0 . 0 1 ) . M e a n peak acceleration and deceleration values during the study are shown in Fig. 2b. The pattern of responses after the infusions resemble those of peak velocity, in that no change in either p a r a m eter was demonstrated after saline infusion ( A N O V A :
There are several interesting findings in the present study. We have confirmed the previous finding of H o m m e r et al. (1986) that peak saccade velocity is reduced in a dose-dependent manner by increasing doses o f benzodiazepines. The present study has shown that other saccade parameters, such as p e a k acceleration, deceleration, a/d ratio, and saccade errors, are also altered in a similar, dose-dependent manner. While the study of H o m m e r et al. (1986) used diazepam, this is the first study to use intravenous m i d a z o l a m to demonstrate such effects on saccade parameters. However, not all saccade parameters were similarly affected. F o r instance, saccade
550 "]
70
0 ¢4
g
60
"o
500 "t
55 50 450
45
'
¢J o
40
z>
35 400
30
Q.
2025--L 350 -'t--30
a
-15
0
15
30
60
45
75
90
105
time
f f saline mldazotam
r-,
120
-30
(rain)
,
,
.
.
-15
b
,
.
0
(6,6,12 pg/kg)
~
15
f saline
flumazenl! 500,ug
,
,
•
,
•
•
30
~
,
45
~ mldszolam (6,6,12 ]Jgikg)
,
,
60
.
.
,
75
.
,
,
90
.
.
,
.
105
.
,
120
~ lime flumazenfl 500}Jg
(rain)
200 -
0•8-
150
8
11
0•7'
10
E
g" e. E 0 ~J
x-~
