Brain (1998), 121, 1155–1164

Differential effect of a dopaminergic agonist on prefrontal function in traumatic brain injury patients

Sharon McDowell,1,2 John Whyte1,2 and Mark D’Esposito3

1Moss Rehabilitation Research Institute, 2Department of Correspondence to: Mark D’Esposito, Department of Physical Medicine and Rehabilitation, Temple University Neurology, University of Pennsylvania Medical Center, School of Medicine, 3Department of Neurology, University 3400 Spruce Street, Philadelphia, PA 19104-4283, USA. of Pennsylvania Medical Center, Philadelphia, PA, USA E-mail: despo@mail.med.upenn.edu

Summary

We examined the effects of low-dose bromocriptine, a D2 dopamine receptor agonist, on processes thought to be subserved by the prefrontal cortex, including working memory and executive function, in individuals with traumatic brain injury. A group of 24 subjects was tested using a double-blind, placebo-controlled crossover trial, counterbalanced for order. Bromocriptine was found to improve performance on some tasks thought to be subserved by prefrontal function, but not others. Specifically, there was improvement in performance on

clinical measures of executive function and in dual-task performance, but not measures that tap the ability to maintain information in working memory without significant executive demands. Also, on control tasks not thought to be dependent on the prefrontal cortex, no improvement on bromocriptine was noted. These results demonstrate a selective effect of bromocriptine on cognitive processes which involve executive control, and provide a foundation for potential therapies for patients with prefrontal damage causing dysexecutive syndromes.

Keywords: dopamine; executive function; frontal lobes; traumatic brain injury; working memory

Abbreviations: MANOVA = multivariate analysis of variance; TBI = traumatic brain injury

Introduction

Executive function is a loosely defined term that is meant to capture a wide range of cognitive abilities such as planning, set-shifting, sustaining and dividing attention, organization and goal-integration. Numerous clinical neuropsychological measures, such as the Wisconsin Card Sorting Test and the Stroop Test (Lezak, 1995), have been designed to tap this function, and patients with lesions of the prefrontal cortex are often impaired on these measures relative to patients with posterior lesions (e.g. Milner, 1963; Vendrell et al., 1995). Also, some of the neuropsychological measures which have been adapted for study during PET scanning have demonstrated activation of the prefrontal cortex during performance by normal subjects (e.g. Bench et al., 1993; Berman et al., 1995). Such empirical evidence has provided a link between executive processes and prefrontal function.

More recently, investigators have sought to define the cognitive processes that may be subserved by the prefrontal cortex more specifically. One such process is working memory, which refers to the short-term storage of information that is not accessible in the environment, and the set of processes that keep this information active for later use in

© Oxford University Press 1998

behaviour. Baddeley (1992) has proposed a multicomponent model of working memory. One component of this model, the ‘central executive system’, analogous to Shallice’s

‘supervisory attentional system’ (Shallice, 1982), is thought to regulate the allocation of limited attentional resources and to co-ordinate the manipulation of the information required in complex or novel tasks. Dual-task paradigms, which are thought to tap these systems, have been demonstrated to be more impaired in patients with focal frontal lesions (Baddeley et al., 1997), as well as frontal lesions secondary to traumatic brain injury (TBI) (Hartman et al., 1992; Cicerone, 1996; McDowell et al., 1997). Also, a functional MRI study in normal subjects during dual-task performance showed activation of the prefrontal cortex (D’Esposito et al., 1995). Recently, Kimberg and Farah (1993) have demonstrated that impairments on a wide range of clinical executive measures thought to tap prefrontal function can be explained in a computational model by weakening associations in working memory without relying on a unique ‘central executive’. The neural correlates of working memory have been extensively studied in monkeys and humans using

delayed response tasks. It has been found that lesions to the lateral prefrontal cortex impair performance on these tasks (Funahashi et al., 1993; Verin et al., 1993), and the prefrontal cortex is activated specifically during the delay period of these tasks (Fuster and Alexander, 1971; Funahashi et al.,

1989; Jonides et al., 1993; Zarahn et al., 1996).

Dopamine seems to be an important neurotransmitter for prefrontal function. Non-human primate studies have demonstrated that there are increased concentrations of dopamine, dopamine receptors and dopamine-containing terminals in the lateral prefrontal cortex (Brown et al., 1979). Converging evidence suggests that neurochemical alterations in the dopaminergic neurotransmitter system can cause frontal lobe dysfunction. Non-human primate studies have demonstrated that dopamine depletion (Sawaguchi and Goldman-Rakic, 1991) and pharmacological dopamine blockade (Brozoski et al., 1979), in addition to focal damage to the striatum (Divac et al., 1967), cause difficulty with spatial working memory tasks. Ageing, which decreases D2 dopamine receptor levels in the prefrontal cortex (Wong et al., 1984; de Keyser et al., 1990; Rinne et al., 1990) has also been reported to impair spatial working memory in monkeys (Arnsten et al., 1995). In humans, patients with Parkinson’s disease, who have dopamine depletion from degeneration of the nigrostriatal pathway, exhibit cognitive deficits attributable to prefrontal dysfunction (e.g. Taylor et al., 1986).

Thus, the goal of this study was to investigate further the link between dopamine and cognitive processes thought to depend on prefrontal function. Based on the cognitive models summarized above, we investigated the effects of a dopaminergic agonist, bromocriptine, on (i) working memory measures, (ii) clinical measures of executive function and (iii) measures not thought to tap prefrontal function. We chose patients with TBI, who we have previously demonstrated to have working memory and executive impairments (McDowell et al., 1997). These deficits are likely to be due to frontal lobe contusions (Adams et al., 1980) and diffuse axonal injury that disrupts dopaminergic inputs to the prefrontal cortex and neural connections between the prefrontal cortex and other areas of the brain (Adams et al., 1982). Based on the work in non-human primates, it was expected that performance on measures thought to depend on prefrontal function would improve on bromocriptine.

Method

Subjects

We measured the cognitive effects of bromocriptine in

24 subjects (20 men, four women) who had TBI causing concussion with loss of consciousness (Glasgow Coma Scale

< 8) more than 4 weeks prior to testing. The subjects’ median age was 32.5 years (range 15–55, with the exception of one 73-year-old patient) and mean duration of education was 13 years (range 9–18). Subjects were excluded for

dementia (Mini-Mental Status Exam score < 26), deficits in oculomotor, uncorrected acuity or visual field, as noted on medical records or by brief bedside examination, a major mental illness, uncontrolled high blood pressure, pregnancy, or current use of centrally acting medications. Based on neuroimaging studies that were available to us, 18 out of 22 patients had evidence of frontal injury. We did not have any radiology reports for two subjects, although one was noted to have facial and orbital fractures. Table 1 gives more clinical and demographic details of our subjects. Twenty-one of the subjects in this study were also participants in a previously reported study not involving pharmacological treatment (McDowell et al., 1997). The protocol described here was approved by the University of Pennsylvania IRB; all subjects gave informed consent prior to participation.

Experimental procedure

After a practice session which was used to familiarize subjects with the tasks, and reduce the effects of practice during the actual test sessions, each subject was tested twice, once on

2.5 mg bromocriptine and once on a placebo, in a randomized, double-blind crossover design counterbalanced for the order of drug administration. Testing was scheduled to occur

90 min after oral pill administration, since the range of peak effectiveness for bromocriptine was estimated from previous studies (Durso et al., 1982; Luciana et al., 1992) to occur between 90 and 180 min after pill ingestion.

Cognitive tasks

All of the tasks used in this study were thought to depend on prefrontal cortex function. Specifically, tasks were chosen to probe different aspects of prefrontal cortex function, including (i) the process of maintenance of information in working memory buffers and (ii) executive control processes, which were examined with an experimental measure as well as commonly used clinical measures. We also included the bi-letter cancellation task, a control task which was used to test non-specific arousal, attention and motor processes (Luciana et al., 1992; Lezak, 1995) and should not depend on the prefrontal cortex. Except for the trailmaking test, a controlled oral word-association test (FAS Test) and the control task, all tasks were performed on a portable computer. The order of the tests was fixed in the order in which they are described below. Not all subjects completed all tests, because of time constraints and the desire of one patient to discontinue the testing for social reasons; for these patients two lengthy and often frustrating tests were omitted (the Verbal Span and Wisconsin Card Sorting Test).

Dual-task paradigm

Subjects performed a simple visual reaction time task and then performed it concurrently with each of two other tasks. In the primary task, subjects responded with a keypress to a

Table 1 Demographic and clinical information of the TBI patients

Age (years)	Education	Sex	Race	Handedness	GCS	Aetiology	Lesions	Skull
26	13	M	W	R	6	MVA	R frontoparietal contusion
							R basal ganglia contusion
							R occipital/L brainstem contusion
43	15	M	B	R	UC	GSW to L face	R frontal contusion
28	11	M	W	R	3	motorcycle	L frontoparietal/convexity SDH
						accident	Diffuse R hemisphere oedema	+
55	12	M	W	R	UC	Hit by object	L frontoparietal SDH
						falling 40'	Small L temporal SDH	+
73	12	F	W	R		Fell down steps	L frontoparietal SDH
53	13	M	W			MVA	R frontal contusion
							R temporal contusion
31	13	F	W	R	DC	MVA	Bilateral frontal contusions
							Diffuse oedema
28	12	M	W	L	UC	GSW to R	R frontoparietal SDH
						lateral orbit	Intraventricular oedema + mass effect
							R occipital infarct/temporalcontusion	+
48	16	M	W	L	UC	MVA	Bilateral frontal contusion and SAH	+
28	16	M	W	R	UC	Aneursym	L ACoA aneurysm/diffuse L oedema
54	13	M			UC	Assault?	L frontal SAH/subfrontal contusion
							Diffuse oedema
43	12	M	W	R	CC	Pedestrian/	R inferior frontal shearing/frontal contusion
						MVA	R basal ganglia/temporal SDH
							Intraventricular SAH
							L parietal contusion
39	15	M	W	R	5	Fell 20 ft	Bilateral frontal/basal ganglia decreased
							perfusion, R occipital decreased perfusion
							Bilateral temporal contusion
20	14	M	W	R	7	MVA	Corpus callosum shearing
							R internal capsule decreased perfusion
							L intraventricular haemorrhage
43	18	M	W	R	UC	MVA	Bilateral subfrontal hygromas/SAH
							Diffuse SAH/intraventricular haemorrhage
24	15	M	H	R	3	MVA	Bilateral frontoparietal contusions
							Corpus callosum haemorrhage
							R basal ganglia/thalamus haemorrhage
							Small L temporal haemorrhage	+
24	16	M	W	R	*	1000-ft fall	L parietal punctate haemorrhage
							R temporal contusion	+
47	15	F	W	R	UC	MVA	R caudate head contusion
							R parasaggital contusion
19	12	M	W	L	8	MVA	L frontal haemorrhage
							R internal capsule, pons and midbrain
							haemorrhage, L medial temporal contusion
18	12	M	B	L	UC	MVA	L temporal SDH	+
34	17	M	W	R	3	Pedestrian/	R frontal SDH/contusions
						MVA
15	9	F	W	L	UC	MVA	No X-rays available
41	13	M	W	L		Pedestrian/	No X-rays available	+
						MVA
23	14	M	W	R	8	MVA	L frontal SAH/contusion diffuse SAH

ID Time after TBI

1 73 days

2 24 months

3 51 days

4 27 days

5 101 days

6 19 months

7 136 months

8 89 days

9 148 days

10 46 days

11 67 days

12 213 days

13 59.5 months

14 42 days

15 39 months

16 7 months

17 48 days

18 149 months

19 27 days

20 49 days

21 67 days

22 48 months

23 300 months

24 37 days

B = black; CC = confused and combative; DC = decerebrate; GCS = Glasgow coma scale; GSW = gunshot wound; H = hispanic mixed race; L = left; MVA = motor vehicle accident; R = right; SAH = subarachnoid haemorrhage; SDH = subdural haemorrhage; UC = unconscious; W = white. *GCS score at 3 days.

target on the computer screen. The target appeared following one of four possible interval delays after the previous response, each used randomly 25% of the time (0.5, 1.0, 1.5 and 2.0 s). The target was a sharply demarcated black dot which appeared in one of 16 dot positions in random order, with location counterbalanced. The dot remained on the screen until the subject responded. Performance was measured as the mean reaction time across 64 trials. One secondary task required the subject to count aloud from one to 10 repeatedly, at a self-selected rate. The other secondary task

was an oral digit span task, using the subject’s own digit span to calibrate difficulty across subjects. This was determined to be the largest digit span which the subject was able to perform correctly three times consecutively, without failing three times consecutively.

Stroop Test

Subjects were required to perform two tasks: naming the colours of rectangles and naming the ink colours in which

different colour names were printed (Stroop, 1935). For the latter task the colour names and the ink colours were always in conflict. Performance was defined as the time required to complete each form of the Stroop Test.

Spatial delayed-response task

A delayed-response spatial location task similar to that developed by Funahashi et al. (1989) and used by Luciana et al. (1992) was used. Subjects were required to recall the location of a black dot on a computer monitor after a brief delay. While the subject was observing a central fixation point, a visual stimulus appeared for 0.2 s at a peripheral location on an imaginary circle on the screen. This stimulus was presented within 10° of the fixation point (to avoid the subject’s blind spot), excluding locations of 0, 90, 180 and

270° (to avoid referencing to the exact vertical and horizontal). After the presentation of the stimulus the screen was blank for 8 s, after which there was an auditory tone which prompted the subject to identify the location of the stimulus by touching a point on the screen. This location was marked by movement of the cursor. Performance was defined as the mean distance in pixels between the stimulus and the response over the 40 trials.

Wisconsin Card Sorting Test

Subjects performed a modified version of the Wisconsin Card Sorting Test (Nelson, 1976). Subjects were asked to sort cards into different preset categories, which changed throughout the test, using only a ‘right’ or ‘wrong’ feedback after each card placement. The sequence of categories was always the same: colour, then shape, then number. The category was changed after six correct consecutive trials. The cards used were only those 24 (repeated continuously) that shared no more than one attribute with any given card, following Nelson’s variation, so that each card placement could be interpreted unambiguously. Subjects continued until they had either achieved six categories or placed 48 cards. Performance was measured as the number of perseverations (cards placed in the same wrong category as an immediately preceding incorrect response).

Verbal span task

Verbal span was assessed using a reading span test (Daneman and Carpenter, 1980), in which subjects read through a series of sentences which increased from two to five sentences, remembering the last word in each sentence.

Trailmaking test

Subjects were timed while drawing a line joining consecutive circles randomly arranged on a page. Two different sequences were used, sequential numbers only (Trails A) and sequential numbers and letters, which the subject had to connect in an alternating pattern (Trails B). Errors were pointed out as they

occurred, according to Reitan’s method (Lezak, 1995), so that the patient could always complete the test. This method penalizes for errors indirectly through increased time requirement, and the scoring is based on time alone.

Controlled oral word association test (FAS Test) Subjects named all the words they could that began with a given letter of the alphabet in three 1-min trials (Lezak,

1995). The testing used two sets of letters, one for each testing session, which were balanced for frequency (F, A and S for the first set, and C, L and T for the second set).

Control tasks

Subjects performed a bi-letter cancellation test, similar to that used by Luciana et al. (1992), in which they read through six rows of letters and crossed out all of the Es and Cs. Performance was measured as the time required to complete this task and the number of errors made. In addition, several of our tasks had baseline conditions (simple visual reaction time alone for the dual-task paradigm, colour-naming control task for the Stroop Test and trailmaking A time for the trailmaking test) which should also assess basic arousal and speed of perceptual registration, cognitive processing and motor response.

For the purpose of evaluating the effect of dopamine manipulation on cognitive function, we grouped the tasks according to different component processes that may be subserved by the prefrontal cortex. Thus, our tasks were grouped into (i) those that tapped the ability to maintain information in working memory buffers (verbal span task and spatial delayed-response task) and (ii) those that tapped executive control processes (dual-task paradigm, Stroop Test, trailmaking test, Wisconsin Card Sorting Test and the FAS Test). Finally, a third group of tasks were those not considered to be dependent on prefrontal cortex function (single-task conditions from the dual-task paradigm, the Stroop control condition, the trailmaking A condition and the bi-letter cancellation test).

Data analysis

Overall comparisons of performance on and off bromocriptine were done with three repeated measures MANOVA for the groups of cognitive tasks described above, with drug condition and specific individual test measures as within-subject factors and drug order as a between-subject factor. These statistical tests were performed with ranked data since raw data violated the assumptions of univariate and multivariate homogeneity of variance (Akritas, 1991). Data were ranked separately for each task across both drug conditions. Bonferroni correction of three MANOVAs leads to a conservative alpha level of

0.0167. We used Koch’s (1972) adaptation of the Mann– Whitney U statistic (which allows for potential period effects and non-parametric data) to perform post hoc testing of the

Table 2 Effects of bromocriptine on cognitive task performance in TBI patients on drug versus placebo, and comparison with a group of control subjects not taking medication

Task Measurement n Placebo Drug Mann– P-value Control* (mean) (mean) Whitney (mean)

Central executive system

Dual task: counting Reaction time decrement (ms) 22 198 96 27 0.028 39

Dual task: digit span Reaction time decrement (ms) 21 539 400 21 0.016 200

Trailmaking test Reaction time decrement (s) 23 83 64 22.5 0.013 23.5

Stroop interference test Reaction time decrement (s) 23 38 33 33.5 0.05 26

FAS Test Words produced 23 31 35 27.5 0.02 54

Wisconsin Card Sorting Perseverations (number) 20 2.9 1.7 23 0.041 2.1

Working memory

Spatial delayed response task	Distance error (pixels)	22	30	29
Reading span	Total correct words	19	38	37
Control
Dual task: baseline	Reaction time (ms)	22	391	429
Stroop colour control	Task time (s)	23	67	65
Trailmaking A	Task time (s)	24	45	48
Letter cancellation test	Task time (s)	19	166	181

‡ ‡ 20

‡ ‡ †

‡ ‡ 303

‡ ‡ 45

‡ ‡ 25

‡ ‡ †

P-value represents drug versus placebo comparison. *Control data from previous study (McDowell et al., 1997). †Tests not used in previous study. ‡Univariate post hoc statistical tests were not warranted in cognitive domains where the MANOVA was not statistically significant.

effect of bromocriptine on specific tests, because the data were not normally distributed. Koch’s procedure consists of ranking the period differences for all of the patients in the trial and then using the Mann–Whitney U test for the differences between the two sequence groups. The use of a test which specifically examines period effects ensures that significant differences are not attributable to anything that may cause a general change between test sessions, such as practice effects or drug carryover effects.

For those tasks which had more than one condition (dual- task, Stroop and trailmaking), performance was characterized with respect to two broad dimensions: baseline task performance and decrement on the second condition compared with baseline. Performance on each of these tasks was measured by response time; the performance decrement was therefore calculated as the difference between speeds on the secondary and the corresponding baseline tasks (Table 2), and these decrement scores were used in the MANOVAs and post hoc testing. Simple differences were an appropriate calculation since the performance decrement had been found to be unrelated to baseline speed in control subjects in a previous study (McDowell et al., 1997). Speeds were characterized by mean performance times, which are sensitive to variable and outlier data that reflect inconsistent performance after TBI.

Results

Overall behavioural performance compared with controls

A previous study which compared cognitive performance

on most of the same measures between TBI patients and demographically matched control subjects showed significant impairment in cognitive performance after TBI using a

multivariate statistical analysis (McDowell et al., 1997). Post hoc comparisons of individual tests showed that individuals with TBI had impaired cognitive function as measured by all of the tests but the Wisconsin Card Sorting Test, after Bonferroni correction for multiple statistical testing. Since that study included 21 of the 24 TBI patients in this study, and the mean performance data are similar or slightly worse in this study, the performance impairment in TBI patients remains (Table 2).

Tests of executive control

The effect of bromocriptine on performance for these cognitive tests was examined with a repeated measures MANOVA (Fig. 1). There was a large drug condition effect [F(1,13) = 10.033, P = 0.007], but no significant order effect [F(1,13) = 0.107, P = 0.749] and no significant Drug X Order interactions [F(1,13) = 4.673, P = 0.050]. However, the Drug X Test interaction was significant [F(5, 65) = 3.921, P = 0.014], showing that performance on certain measures was more significantly improved by bromocriptine than on others. In fact, the effect size of the differences ranged from 0.34 for the Stroop Test to 0.71 for the trailmaking test. Post hoc univariate analyses confirmed a beneficial effect of bromocriptine on the dual-task, the trailmaking test, the Stroop Test, the Wisconsin Card-Sorting Test and the FAS Test (Table 2). The average performance improvement in these executive control tests was just >25%.

Tests of working memory maintenance

The effect of bromocriptine on the performance of tasks which seem to require only the maintenance of information without any active manipulation of stored information was

Fig. 1 Effect of bromocriptine on different tasks. For each task, this chart shows the effect size of the performance change with bromocriptine [mean of (drug – placebo)/standard deviation of (drug – placebo)], a positive value indicating improved performance on the drug. Effect size was used only for graphical purposes to allow a comparison of many different tasks on the same scale. Since most measures are time-related, where greater values indicate worse performance, the difference was calculated as the placebo measurement minus the drug measurement, although this difference was reversed (drug measurement minus placebo measurement) for the FAS Test and the spatial delayed-response task, for which greater values indicate better performance, to maintain consistency of direction with the other measures on this chart. For this graph, upward bars indicate drug benefit. For tasks with more than one level the corresponding decrement scores were subtracted. RT = reaction time; WCS = Wisconsin Card Sorting Test.

assessed similarly to the tests above (Fig. 1). Bromocriptine had no significant effect on these tasks [F(1,15) = 0.001, P = 0.978], with no significant effects of order or Drug X Order interaction. Individual test scores are shown in Table

2, with a non-significant trend towards worse performance. Post hoc testing was not performed to reduce type I error, since the multivariate test was not significant.

Control tests

The impact of bromocriptine on the control task and baseline conditions of the other executive measures was assessed. Basic attentional and sensorimotor processes were not improved on bromocriptine (Fig. 1). Statistical tests analogous to those performed above showed no significant main effect of the drug on these tests [F(1,14) = 3.203, P = 0.095], although there was a trend for performance to be worse on the medication. Neither was there any significant order effect or DrugXOrder interaction. Individual test scores are shown in the table. Again, post hoc testing was not performed.

Speed–accuracy assessment

Performance on some tasks (the bi-letter cancellation test, the Stroop Test and the dual task with concurrent digit span)

can be measured by both time and accuracy, and both could be affected by the medication. To make certain that the beneficial effect of bromocriptine on speed for these tasks was not due to speed–accuracy trade-off, the effect of medication on task accuracy was also assessed with the Wilcoxon signed rank test. Accuracy was not significantly affected by bromocriptine for any of these tasks (Z = –1.23 to –1.85, P = 0.065–0.22), and the non-significant changes in function that did occur with bromocriptine were also in the direction of improvement.

Discussion

The pattern that emerges from the data is that bromocriptine, a dopaminergic agonist, generally improved performance in individuals with TBI on tasks thought to engage executive processes (i.e. the dual task, the trailmaking test, the Stroop Test, verbal fluency and the Wisconsin Card Sorting Test). However, performance did not improve on the baseline condition of several of our executive measures (i.e. dual task, trailmaking test and Stroop Test), working memory measures with minimal executive demands (i.e. verbal span and delayed response task), or the control task. This differential effect of bromocriptine suggests that its action is not mediated through

a non-specific increase in arousal, attention or response speed, but probably through improvement in specific cognitive processes.

A lack of improvement on our control measures may be somewhat surprising, because dopaminergic agents have been used to improve general arousal and attention in individuals with brain damage (Van Woerkom et al., 1982; Lal et al.,

1988). Reports of this effect, however, are mostly case reports, not controlled studies; also they used higher medication doses and did not strictly measure arousal to distinguish it from other cognitive processes. In other words, the reported improvement in arousal and attention may actually be related to an improvement in prefrontal function, which is less easily defined and not often measured. Dopamine agonists have also been used to improve motor speed in patients with Parkinson’s disease (Godwin-Austen and Smith, 1977), but they did not have that effect here, as demonstrated by a lack of improvement in speed-related measures not dependent on executive function (i.e. the bi-letter cancellation task and control tasks of the Stroop, trailmaking and dual-task paradigms). This disparity may be explained by a differential dose effect, because Parkinson’s disease patients are usually treated with substantially higher doses of dopamine agonists than those used in the present study, and their motor slowing may be of a different type than that seen in TBI.

Since this differential behavioural effect on prefrontal function was unexpected, alternative explanations should be considered. For example, these results could be due to differences in the interval after drug administration that each of the tasks were given. However, this is unlikely for several reasons. First, the half-life of bromocriptine, which is ~3–7 h, is longer than the time it took to complete our complete battery of tasks. Secondly, tasks that showed a drug effect were intermixed in time with those that did not. Finally, our statistical analysis accounted for the possibility of any significant differences between the drug and placebo being due to general changes between test sessions, such as practice effects or drug carryover effects. Another possible explanation for our results is that the tasks that did not show improvement on bromocriptine were less sensitive in the detection of differences. Again, this is unlikely, since the range of performance on these tasks by patients was quite broad. For example, on the spatial delayed response task, difference scores between sessions ranged from an improvement in spatial error of 22.7 pixels to a decrement of 12.8 pixels (raw data ranged from 8.8 to 68.4 pixels). Likewise, the verbal span test difference scores ranged from an improvement of six correct words to a decrement of 10 words (raw data ranged from 16 to 52 words recalled). Related to this issue, it is possible that our patients were more impaired on those tasks that responded to bromocriptine than on tasks that did not respond to bromocriptine. Such a task-difficulty explanation seems unlikely since our subjects were impaired, relative to control subjects, on all of the tasks we administered except for the Wisconsin Card Sorting Test, and even this task showed improvement with bromocriptine.

Although it has been predicted from non-human primate studies that tasks that require the ability to maintain information in working memory, such as delayed-response paradigms, should be modulated by dopamine (Brozoski et al., 1979; Sawaguchi and Goldman-Rakic, 1991), the evidence from humans is less clear. Two studies from the same laboratory have shown that performance on a spatial delayed-response task in normal subjects, similar to that used in non-human primates and identical to the task used in our study, is improved on bromocriptine (Luciana et al., 1992; Luciana and Collins, 1997). However, in our own previous study with a group of normal subjects (Kimberg et al., 1997), and in the current study with frontally damaged patients, we did not find improvements on this spatial delayed-response task or a verbal span measure. Similarly, a preliminary study from another group did not find improvement on a delayed- response paradigm with bromocriptine, but they did find improvement with pergolide, an agonist of D1 and D2

receptors (Mu¨ ller et al., 1998).

Although the literature is slightly conflicting regarding the role of dopamine in maintenance processes, we believe that bromocriptine has a greater effect on tasks that require executive control for successful performance. This suggests that executive control processes and maintenance processes may be supported by distinct neuroanatomical or neurochemical substrates. Petrides and colleagues have proposed a model of functional organization of the prefrontal cortex that would be consistent with this hypothesis (Petrides,

1994). According to their ‘two-stage model’, there are two processing systems, one dorsal and the other ventral, within the lateral prefrontal cortex. It is proposed that the ventral region is the site where information is initially received from posterior association areas and where active comparisons of information held in working memory are made. In contrast, the dorsal region is recruited only when monitoring and manipulation within working memory is required. Consistent with this model is the demonstration that dorsal prefrontal regions are activated during spatial memory tasks that have maximal monitoring requirements compared with a spatial memory task with minimal monitoring requirements, which activated only a ventral frontal region (Owen et al., 1996). Furthermore, a meta-analysis of all functional neuroimaging studies of working memory reported to date also supports this type of prefrontal organization (D’Esposito et al., 1998a) as does a direct empirical functional MRI test of this hypothesis (D’Esposito et al., 1998b).

Although there are no other empirical data suggesting that a neurochemical dissociation exists along the lines mentioned above, there is evidence that Parkinson’s disease patients, who are dopamine-depleted, are impaired differentially on different components of working memory. For example, Parkinson’s disease patients in the early stages of disease are impaired on spatial and not object working memory tasks, suggesting that these different processes have different sensitivities to dopamine depletion (Owen et al., 1997; Postle et al., 1997). Consistent with these findings is the

demonstration in this study that, in normal subjects, bromocriptine administration improved performance on a spatial but not an object working memory task. Moreover, haloperidol, a dopaminergic antagonist, impaired performance on only the spatial task (Luciana and Collins, 1997). More pertinent to our findings, Parkinson’s disease patients that have been studied off and on their medications show differential impairments on prefrontal tasks comparable with those used in our study. For example, Parkinson’s disease patients on their dopaminergic medication (compared with off it) have been shown to perform better on a wide range of executive function tasks, such as the Wisconsin Card Sorting Task, a verbal fluency task and the Tower of London task (Bowen et al., 1975; Cooper et al., 1992; Lange et al., 1992, 1995). In one study, whilst on their medication Parkinson’s disease patients were better on executive measures, whereas no difference was found on a verbal and spatial short-term span test (Lange et al., 1995).

The neural basis of the improvement on prefrontal tasks in our patients is not entirely clear. Although most of the primate studies link prefrontal function to D1 receptor activity, bromocriptine is a D2 agonist. That is, monkeys improve on spatial working memory tasks when administered D1 agonists (Arnsten et al., 1994) and are impaired when given D1 antagonists (Sawaguchi and Goldman-Rakic, 1991; Arnsten et al., 1994). However, a recent study from the same laboratory is conflicting and difficult to interpret—the prefrontal cortex neurons in monkeys demonstrated increased activity during spatial working memory tasks with iontophoretic infusion of a D1 antagonist (Williams and Goldman-Rakic, 1995). Very low doses of bromocriptine have some D1 antagonist activity (Goetz and Diederich,

1992), and this D1 antagonism could be responsible for the

improvement seen on spatial delayed-response paradigms in

humans (Luciana and Collins, 1997). This effect may be separate from the D2 agonist effects of bromocriptine. The presence of D2 receptors in the human cortex is not well established, because very low levels of dopamine in the cortex preclude accurate biochemical analysis. Some studies, though, have shown stratification of D2 receptors to layer V of the prefrontal cortex (Goldman-Rakic et al., 1990). However, most D2 receptors are in the striatum (Camps et al., 1989). Finally, a recent study from Goldman-Rakic’s laboratory shows that a D2 agonist can improve spatial delayed-response task performance in monkeys (Arnsten et al., 1995). Further testing showed this effect to be mediated by interactions between D1 and D2 receptors. Therefore, the effects of bromocriptine on our subjects could be the result of direct prefrontal D2 receptor activity, or of the activity of striatal D2 receptors, which in turn activate the prefrontal cortex through mesocortical dopaminergic loops. Nevertheless, the relationship between dopamine and prefrontal function, as demonstrated in human pharmacological studies and Parkinson’s disease studies, seems strong.

In summary, our empirical findings have shown that dopamine appears to modulate executive processes which

are impaired after damage to the prefrontal cortex. This finding is consistent with a similar study performed in our laboratory studying the effects of bromocriptine in normal young adults (Kimberg et al., 1997). In that study it was shown that the effect of bromocriptine depended upon the working memory capacity of the individual. That is, bromocriptine was shown to improve performance on prefrontal measures in subjects who had lower working memory capacity, whereas it impaired performance in subjects with higher working memory capacity. Our TBI patients all had low working memory capacity, and as a group they consistently improved on bromocriptine. Taken together, these two studies highlight the spectrum of dopamine function that exists between normal control subjects and patients with brain damage, and its link with cognitive processes dependent on the prefrontal cortex.

Although individuals with frontal lobe pathology can perform well on standard neuropsychological tests, they often exhibit significant functional difficulties (Eslinger and Damasio, 1985; Shallice and Burgess, 1991). These practical difficulties, which seem to be caused by deficits in organization, planning and goal integration, are probably due to impairments in the processes that improved on bromocriptine in our study. Since incomplete recovery of prefrontal function prevents full reintegration into society, the development of more effective treatment approaches for these patients, such as pharmacological intervention, will be an extremely important component of their care. Thus, our findings provide a foundation for potential therapies that may not only improve executive function in patients with prefrontal damage, but also decrease the disability associated with these cognitive impairments.

Acknowledgements

We wish to thank Dan Kimberg for helpful discussions of this work. This project was supported in part by NIH grants HD07425, HD01097, NS01762 and AG13483, the McDonnell-Pew Program in Cognitive Neuroscience, and by funds from the Moss Rehabilitation Research Institute.

References

Adams JH, Scott G, Parker LS, Graham DI, Doyle D. The contusion index: a quantitative approach to cerebral contusions in head injury. Neuropathol Appl Neurobiol 1980; 6: 319–24.

Adams JH, Graham DI, Murray LS, Scott G. Diffuse axonal injury due to non-missile head injury in humans: an analysis of 45 cases. Ann Neurol 1982; 12: 557–63.

Akritas MG. Limitations of the rank transform procedure: a study of repeated measures designs, Part I. J Am Stat Assoc 1991; 86:

457–460.

Arnsten KT, Cai JX, Murphy BL, Goldman-Rakic PS. Dopamine

D1 receptor mechanisms in the cognitive performance of young

adult and aged monkeys. Psychopharmacology (Berl) 1994; 116:

143–51.

Arnsten AF, Cai JX, Steere JC, Goldman-Rakic PS. Dopamine D2 receptor mechanisms contribute to age-related cognitive decline: the effects of quinpirole on memory and motor performance in monkeys. J Neurosci 1995; 15: 3429–39.

Baddeley A. Working memory. [Review]. Science 1992; 255: 556–9. Baddeley A, Della Sala S, Papagno C, Spinnler H. Dual-task

performance in dysexecutive and nondysexecutive patients with a

frontal lesion. Neuropsychology 1997; 11: 187–94.

Bench CJ, Frith CD, Grasby PM, Friston KJ, Paulesu E, Frackowiak RS, et al. Investigations of the functional anatomy of attention using the Stroop Test. Neuropsychologia 1993; 31: 907–22.

Berman KF, Ostrem JL, Randolph C, Gold J, Goldberg TE, Coppola R, et al. Physiological activation of a cortical network during performance of the Wisconsin Card Sorting Test: a positron emission tomography study. Neuropsychologia 1995; 33: 1027–46.

Bowen FP, Kamienny RS, Burns MM, Yahr M. Parkinsonism:

effects of levodopa treatment on concept formation. Neurology

1975; 25: 701–4.

Brown RM, Crane AM, Goldman PS. Regional distribution of monoamines in the cerebral cortex and subcortical structures of the rhesus monkey: concentrations and in vivo synthesis rates. Brain Res 1979; 168: 133–50.

Brozoski TJ, Brown RM, Rosvold HE, Goldman PS. Cognitive deficit caused by regional depletion of dopamine in prefrontal cortex of rhesus monkey. Science 1979; 205: 929–32.

Camps M, Corte´s R, Gueye B, Probst A, Palacios JM. Dopamine receptors in human brain: autoradiographic distribution of D2 sites. Neuroscience 1989; 28: 275–90.

Cicerone KD. Attention deficits and dual task demands after mild traumatic brain injury. Brain Inj 1996; 10: 79–89.

Cooper JA, Sagar HJ, Doherty SM, Jordan N, Tidswell P, Sullivan EV. Different effects of dopaminergic and anticholinergic therapies on cognitive and motor function in Parkinson’s disease. Brain 1992;

115: 1701–25.

D’Esposito M, Detre JA, Alsop DC, Shin RK, Atlas S, Grossman M. The neural basis of the central executive system of working memory. Nature 1995; 378: 279–81.

D’Esposito M, Aguirre GK, Zarahn EK, Ballard D, Shin RK, Lease J. Functional MRI studies of spatial and non-spatial working memory. Cog Brain Res. In press 1998a.

D’Esposito M, Postle BR, Ballard D, Lease J. Maintenance versus manipulation of information held in working memory: an event- related fMRI study. Brain and Cognition. In press 1998b.

Daneman M, Carpenter PA. Individual differences in working memory and reading. J Verb Learn Verb Behav 1980; 19: 450–66.

de Keyser J, Ebinger G, Vauquelin G. Age-related changes in the human nigrostriatal dopaminergic system. Ann Neurol 1990; 27:

157–61.

Divac L, Rosvold LE, Szwarcbart MK. Behavioral effects of

selective ablation of the caudate nucleus. J Comp Physiol Psychol

1967; 63: 184–90.

Durso R, Fedio P, Brouwers P, Cox C, Martin AJ, Ruggieri SA, et al. Lysine vasopressin in Alzheimer’s disease. Neurology 1982;

32: 674–7.

Eslinger PJ, Damasio AR. Severe disturbance of higher cognition after bilateral frontal lobe ablation: patient EVR. Neurology 1985;

35: 1731–41.

Funahashi S, Bruce CJ, Goldman-Rakic PS. Mnemonic coding of visual space in the monkey’s dorsolateral prefrontal cortex. J Neurophysiol 1989; 61: 331–349.

Funahashi S, Bruce CJ, Goldman-Rakic PS. Dorsolateral prefrontal lesions and oculomotor delayed-response performance: evidence for mnemonic ‘scotomas’. J Neurosci 1993; 13: 1479–97.

Fuster JM, Alexander GE. Neuron activity related to short-term memory. Science 1971; 173: 652–4.

Godwin-Austen RB, Smith NJ. Comparison of the effects of bromocriptine and levodopa in Parkinson’s disease. J Neurol Neurosurg Psychiatry 1977; 40: 479–82.

Goetz CG, Diederich NJ. Dopaminergic agonists for the treatment of Parkinson’s disease. [Review]. Neurol Clin 1992; 10: 527–40.

Goldman-Rakic PS, Lidow MS, Gallager DW. Overlap of dopaminergic, adrenergic, and serotoninergic receptors and complementarity of their subtypes in primate prefrontal cortex. J Neurosci 1990; 10: 2125–38.

Hartman A, Pickering RM, Wilson BA. Is there a central executive deficit after severe head injury? Clin Rehabil 1992; 6: 133–40.

Jonides J, Smith EE, Koeppe RA, Awh E, Minoshima S, Mintun MA. Spatial working memory in humans as revealed by PET [see comments]. Nature 1993; 363: 623–5. Comment in: Nature 1993;

363: 583–4.

Kimberg DY, Farah MJ. A unified account of cognitive impairments following frontal damage: the role of working memory in complex organized behaviour. J Exp Psych: Learn Mem Behav 1993; 122:

411–28.

Kimberg DY, D’Esposito M, Farah MJ. Effects of bromocriptine on human subjects depend on working memory capacity. Neuroreport

1997; 8: 3581–5.

Koch GG. The use of non-parametric methods in the statistical analysis of the two-period change-over design. Biometrics 1972;

28: 577–84.

Lal S, Merbtiz CP, Grip JC. Modification of function in head- injured patients with sinemet [see comments]. Brain Injury 1988;

2: 225–33. Comment in: Brain 1989; 3: 319–22.

Lange KW, Robbins TW, Marsden CD, James M, Owen AM, Paul GM. L-dopa withdrawal in Parkinson’s disease selectively impairs cognitive performance in tests sensitive to frontal lobe dysfunction. Psychopharmacology (Berl) 1992; 107: 394–404.

Lange KW, Paul GM, Naumann M, Gsell W. Dopaminergic effects of cognitive performance in patients with Parkinson’s disease. J Neural Transm Suppl 1995; 46: 423–32.

Lezak MD. Neuropsychological assessment. 3rd ed. New York: Oxford University Press; 1995.

Luciana M, Collins PF. Dopaminergic modulation of working memory for spatial but not object cues in normal humans. J Cogn Neurosci 1997; 9: 330–47.

Luciana M, Depue RA, Arbisi P, Leon A. Facilitation of working memory in humans by a D2 dopamine receptor agonist. J Cogn Neurosci 1992; 4: 58–68.

McDowell S, Whyte J, D’Esposito M. Working memory impairments in traumatic brain injury: evidence from a dual-task paradigm. Neuropsychologia 1997; 35: 1341–53.

Milner B. Effects of different brain lesions on card sorting. Arch

Neurol 1963; 9: 90–100.

Mu¨ ller U, von Cramon DY, Pollmann S. D1 versus D2-receptor modulation of visuo-spatial working memory in humans. J Neurosci

1998; 18: 2720–8.

Nelson HE. A modified card sorting test sensitive to frontal lobe defects. Cortex 1976; 12: 313–24.

Owen AM, Evans AC, Petrides M. Evidence for a two-stage model of spatial working memory processing within the lateral frontal cortex: a positron emission tomography study. Cereb Cortex 1996;

6: 31–8.

Owen AM, Iddon JL, Hodges JR, Summers BA, Robbins TW. Spatial and non-spatial working memory at different stages of Parkinson’s disease. Neuropsychologia 1997; 35: 519–32.

Petrides M. Frontal lobes and behaviour. [Review]. Curr Opin

Neurobiol 1994; 4: 207–11.

Postle BR, Jonides J, Smith EE, Corkin S, Growdon JH. Spatial, but not object, delayed response is impaired in early Parkinson’s disease. Neuropsychology 1997; 11: 171–9.

Rinne JO, Lonnberg P, Marjamaki P. Age-dependent decline in human brain dopamine D1 and D2 receptors. Brain Res 1990; 508:

349–52.

Sawaguchi T, Goldman-Rakic PS. D1 dopamine receptors in prefrontal cortex: involvement in working memory. Science 1991;

251: 947–50.

Shallice T. Specific impairments of planning. Philos Trans R Soc

Lond B Biol Sci 1982; 298: 199–209.

Shallice T, Burgess PW. Deficits in strategy application following frontal lobe damage in man. Brain 1991; 114: 727–41.

Stroop JR. Studies of interference in serial verbal reactions. J Exp

Psychol 1935; 18: 643–62.

Taylor AE, Saint-Cyr JA, Lang AE, Kenny FT. Parkinson’s disease and depression. A critical re-evaluation. Brain 1986; 109: 279–92.

Van Woerkom TC, Minderhoud JM, Gottschal T, Nicolai G. Neurotransmitters in the treatment of patients with severe head injuries. Eur Neurol 1982; 21: 227–34.

Vendrell P, Junque C, Pujol J, Jurado MA, Molet J, Grafman J. The role of prefrontal regions in the Stroop task. Neuropsychologia

1995; 33: 341–52.

Verin M, Partiot A, Pillon B, Malapani C, Agid Y, Dubois B. Delayed response tasks and prefrontal lesions in man—evidence for self generated patterns of behaviour with poor environmental modulation. Neuropsychologia 1993; 31: 1379–96.

Williams GV, Goldman-Rakic PS. Modulation of memory fields by dopamine D1 receptors in prefrontal cortex [see comments]. Nature

1995; 376: 572–5. Comment in: Nature 1995; 376: 549–50.

Wong DF, Wagner HN Jr, Dannals RF, Links JM, Frost JJ, Ravert HT, et al. Effects of age on dopamine and serotonin receptors measured by positron tomography in the living human brain. Science

1984; 226: 1393–6.

Zarahn E, Aguirre GK, D’Esposito M. Delay-specific activity within prefrontal cortex demonstrated during a working memory task: a functional MRI study. Soc Neurosci Abstr 1996; 22: 968.

Received January 5, 1998. Accepted January 21, 1998