Specific cognitive deficits in mild frontal variant frontotemporal dementia

Transcription

1 Brain (1999), 122, Specific cognitive deficits in mild frontal variant frontotemporal dementia Shibley Rahman, 1 Barbara J. Sahakian, 1 John R. Hodges, 2,3 Robert D. Rogers 4 and Trevor W. Robbins 4,5 Departments of 1 Psychiatry and 2 Neurology, 3 MRC Cognition and Brain Sciences Unit, 4 Department of Experimental Psychology and the 5 MRC Centre for Brain Repair, University of Cambridge, Cambridge, UK Summary Eight patients with relatively mild frontal variant frontotemporal dementia (fvftd) were compared with age- and IQ-matched control volunteers on tests of executive and mnemonic function. Tests of pattern and spatial recognition memory, spatial span, spatial working memory, planning, visual discrimination learning/ attentional set-shifting and decision-making were employed. Patients with fvftd were found to have deficits in the visual discrimination learning paradigm specific to the reversal stages. Furthermore, in the decision-making paradigm, patients were found to show genuine risk-taking behaviour with increased deliberation times rather than merely impulsive behaviour. It was especially notable that Correspondence to: Dr Barbara J. Sahakian, Department of Psychiatry, University of Cambridge School of Clinical Medicine, Addenbrooke s Hospital, Cambridge CB2 2QQ, UK these patients demonstrated virtually no deficits in other tests that have also been shown to be sensitive to frontal lobe dysfunction, such as the spatial working memory and planning tasks. These results are discussed in relation to the possible underlying neuropathology, the anatomical connectivity and the hypothesized heterogeneous functions of areas of the prefrontal cortex. In particular, given the nature of the cognitive deficits demonstrated by these patients, we postulate that, relatively early in the course of the disease, the ventromedial (or orbitofrontal) cortex is a major locus of dysfunction and that this may relate to the behavioural presentation of these patients clinically described in the individual case histories. Keywords: frontotemporal dementia; decision-making; reversal learning; orbitofrontal prefrontal cortex; ventromedial prefrontal cortex Abbreviations: ANOVA 5 analysis of variance; CANTAB 5 Cambridge Neuropsychological Test Automated Battery; FTD 5 frontotemporal dementia; fvftd 5 frontal variant FTD; HMPAO 5 99m Tc-hexamethyl-propyleneamineoxide; MMSE 5 Mini-Mental State Examination; NART 5 National Adult Reading Test; SPECT 5 single photon emission computed tomography; WAIS 5 Wechsler Adult Intelligence Scale; WCST 5 Wisconsin Card Sorting Test Introduction The term frontotemporal dementia (FTD), coined in the consensus statement from the Lund and Manchester Groups (1994), represents a heterogeneous group of disorders with variable clinical and neuropathological manifestations. The term encompasses a number of conditions in which progressive dementia with bilateral degeneration of the frontal and/or temporal cortex occurs, previously referred to as Pick s disease (Pick, 1892), dementia of the frontal type (Neary et al., 1988), frontal lobe degeneration of the non- Alzheimer type (Brun, 1987, 1993; Gustafson, 1993), dementia lacking distinct histopathological features (Knopman et al., 1990), primary progressive aphasia (Mesulam, 1982; Hodges and Patterson, 1996; Snowden et al., 1996a, b), and semantic dementia (Snowden et al., 1989; Hodges et al., 1992, 1999). Neuropathological findings Oxford University Press 1999 post-mortem may vary in such cases. Some have tau-positive intraneuronal inclusions (Pick bodies), whereas others have severe neuronal depletion with spongiosis but lack distinct inclusions; Alzheimer pathology is virtually never found (e.g. Mann, 1998). The precise mapping between clinical phenotype, distribution of pathology, type of pathology, and genotype needs further clarification. In a proportion of cases, however, the disorder is familial and linked to tau gene mutations on chromosome 17 (Hutton et al., 1998; Spillantini et al., 1998); it has been suggested that these familial chromosome-17-linked cases have a greater preponderance of parkinsonian features and lack Pick bodies, but share a distinctive abnormality of tau (Spillantini et al., 1998). This study concerns aspects of cognition in patients with what we have termed frontal variant FTD (fvftd) (Hodges

2 1470 S. Rahman et al. et al., 1999), in which frontal involvement is the principal cause of the circumscribed presentation of a behavioural syndrome. This terminology parallels that used by Edwards- Lee and colleagues (Edwards-Lee et al., 1997) to describe patients with predominantly temporal lobe involvement, who typically have profound language deficits with the features of semantic dementia. There is little controversy over the fact that fvftd affects younger patients, with the peak onset in the presenium (Neary and Snowden, 1996). Cases above 75 years of age appear relatively rarely (Miller et al., 1991; Gregory and Hodges, 1996). The exact prevalence of FTD is currently uncertain, but there is emerging evidence that FTD is the second commonest form of dementia in the presenium, and may represent up to a fifth of cases (Knopman et al., 1990). A biopsy study of patients with presenile dementia revealed a prevalence of 17% with non-alzheimer pathology causing a syndrome of FTD (Neary et al., 1986). On clinical grounds alone, Neary et al. (1988) estimated that 19% of 130 cases seen in the neurological unit over a period of 5 years met criteria for frontal lobe dementia. Ultimately, patients develop global cognitive impairment and severe disabilities in the functions of daily living. It is, however, of considerable importance for effective patient management that patients are identified in the earliest stages of the disease, particularly in the light of the psychological, social and financial consequences within families (e.g. Sperlinger and Furst, 1994). Clinically, the majority of patients with fvftd are brought along to a clinic unaware of the major pervasive changes in their personality, behaviour and social conduct that have been observed by the informants. Patients may appear apathetic or withdrawn, or they may become socially disinhibited with facetiousness and inappropriate jocularity. Mental rigidity and an inability to appreciate the subtler aspects of language, such as irony, are common, as is stereotypical or ritualistic behaviour. Their ability to plan and organize complex activities (e.g. work, social engagements, etc.) is almost invariably impaired. This is traditionally considered to be due to deficits in goal-setting and attainment, as well as in mental flexibility and setshifting (Gregory and Hodges, 1993, 1996). There is often indifference to domestic and occupational responsibilities, a lack of empathy for family and friends, and gradual withdrawal from all social interactions. A form of severe self-neglect may occur as a direct consequence of frontal lobe dysfunction (Orrell et al., 1989). Many patients show a change in eating habits with an escalating desire for sweet food coupled with reduced satiety; the increasing gluttony may lead to enormous weight gain. The other components of the Klüver Bucy syndrome (hypersexuality and oral exploratory behaviour) are less commonly observed (Cummings and Duchen, 1981). Language deficits include a mild anomia or dynamic aphasia (Snowden et al., 1996a, b) and repetitive speech acts, with muteness supervening in advanced cases. Visuospatial skills are typically remarkably preserved. Memory is typically intact, but patients may have difficulty with concentration and immediate working memory. Patients with fvftd may present with prominent psychiatric symptoms (Neary et al., 1988; Miller et al., 1991; Gregory and Hodges, 1993, 1996). Depression may occur but is rarely severe (Gregory and Hodges, 1993, 1996). Compulsive disorders are more common and include, amongst others, cleaning, hygiene rituals, checking, and preoccupation with symmetry (Miller et al., 1995). Furthermore, Gustafson and Risberg (1992) report that hallucinations and delusions are prevalent in dementia (including FTD) and that psychotic traits may give the impression of a functional psychosis such as schizophrenia; however, in other series overt psychotic features in fvftd have been found to be very rare and fleeting (e.g. Gregory et al., 1998; Gregory and Hodges, 1996). In contrast to the clinical descriptions of profound frontal changes in patients with fvftd, formal studies have often failed to show a specific profile, at least until late in the course of the disease (Miller et al., 1991). The failure to find clear neuropsychological differences between dementia of the Alzheimer type and FTD is likely to relate to variation in patient selection and to the particular choice of neuropsychological tests (Miller et al., 1991; Elfgren et al., 1994; Frisoni et al., 1995; Gregory et al., 1997). For example, Pachana and colleagues (Pachana et al., 1996) compared 15 patients with FTD (including both frontal and temporal cases) and 16 cases of dementia of the Alzheimer type, but did not match for severity or duration of dementia. They found a difference in the severity of non-verbal memory functioning in the predicted direction, but no differences in a range of language, attentional and executive tasks. More recently, Gregory and colleagues (Gregory et al., 1997) assessed the success of a brief battery of bedside cognitive tasks in distinguishing between groups of patients with dementia of the Alzheimer type and those with the purely frontal variant of FTD matched on the Mini-Mental State Examination (MMSE). Traditional tests of memory, attention, language comprehension and executive function failed to separate the groups. Given the potential utility of cognitive neuropsychology in the accurate diagnosis of fvftd, the performance of patients on various frontal tests is of special interest. One longitudinal study of patients demonstrated an unchanged cognitive profile across a series of assessments apart from on frontal tests, which showed progressive worsening (Venneri et al., 1996). Groups of patients with frontal lobe dementia have previously been shown to be impaired in verbal fluency tasks (Miller et al., 1991; Elfgren et al., 1993), but this finding is non-specific. The use of letter-initial verbal fluency performance in dementia of the frontal type, dementia of the Alzheimer type and control subjects matched for age, sex and level of education, was examined in a recent study (Pasquier et al., 1995). It was found that verbal fluency tests, whilst sensitive tools for detecting dementia, did not seem useful in distinguishing between patients with dementia of the Alzheimer type and those with dementia of the frontal type, as the two groups did not differ in the number of words generated, intrusions or perseverations. One other test that

3 Cognition in frontotemporal dementia 1471 has been used most consistently in the assessment of frontal lobe dementia is the Wisconsin Card Sorting Test (WCST), which in a clinical setting has been applied to identify frontal lobe dysfunction (Milner, 1963). Although results across studies may be difficult to compare as different administration and sorting principles are used, many patients with fvftd have been found to demonstrate deficits in the WCST, particularly those with lower MMSE scores, i.e. in the dementing range (Neary et al., 1988; Miller et al., 1991; Elfgren et al., 1993). Deficits on the WCST are difficult to attribute to specific neural damage, both within and outside prefrontal cortical regions (Robinson et al., 1990; Anderson et al., 1991; Corcoran and Upton, 1993). The lack of neural specificity of the WCST probably arises because of the complexity of the test, which has many cognitive components (e.g. including response inhibition and working memory). Indeed, decompositions of the test requirements in monkeys have shown that different forms of perseveration may result from damage to different sectors of the prefrontal cortex (Dias et al., 1996). It is therefore clear that performance on the WCST cannot always be used to define with confidence the neural loci of cognitive deficits in patients with fvftd. It also emerges from the literature that some patients, despite marked neurobehavioural changes and eventual clear-cut atrophy on MRI, may achieve all categories. As an example, Neary and colleagues (Neary et al., 1988) provide a graphic description of a male electrician who was suspended from work at the age of 55 years because of increasingly clownish behaviour and a reduction in his level of personal and social behaviour. His behaviour became progressively more puerile and disinhibited, resulting in his retirement and forcing his wife to assume responsibility for all household and financial affairs. Nonetheless, his performance on the modified form of the WCST remained intact, achieving six complete categories. Kertesz (1998) also recently noted that three out of 12 patients with a diagnosis of frontal lobe dementia had normal or nearly normal scores on frontal lobe function tests despite severe behavioural disturbances. A recent observation of Gregory and colleagues (Gregory et al., 1999) may also be of some relevance. They describe two patients who presented with a marked change in personality and behaviour, and who indeed progressed over a number of years in a way conforming to a clinical picture of fvftd, but in whom there remained remarkably little abnormality on neuropsychological testing. Whilst they had normal HMPAO ( 99m Tc-hexamethylpropyleneamineoxide)-single photon emission computed tomography (SPECT) imaging when first investigated, both patients subsequently developed abnormalities on HMPAO- SPECT, frontal atrophy on MRI and a neuropsychological profile usually considered typical of fvftd. There may be a neuroanatomical basis for the discrepancy between the behaviour of patients and their neuropsychological performance. Specific architectonic differentiation and connectivity of the prefrontal cortex has been identified (Barbas and Pandya, 1991; Yeterian and Pandya, 1994). It is now widely accepted that a series of parallel subcortical circuits exist that link regions of the prefrontal cortex to the basal ganglia (Alexander et al., 1986; Alexander and Crutcher, 1990). Therefore, arguably, as the various subdivisions of the frontal lobes may subserve distinct and specific functions, traditional neurocognitive assessment may fail to test the functions of the areas of the prefrontal cortex which are predominantly affected by pathology. It has been recognized previously that successful identification of the neuroanatomical region involved may help ultimately in the direct selection of behavioural investigations and, conversely, to study which behaviours are correlated with particular brain regions (Stuss, 1993). In the present study we examined the nature of cognition in patients with fvftd with particular emphasis upon frontal lobe function. We have included only patients who are affected relatively mildly, defined (in parallel with patients with mild dementia of the Alzheimer type) for patients with a MMSE score of 20 (Folstein et al., 1975). In attempting to determine the major domains of neuropsychological deficit in patients with mild fvftd, and thereby to develop suitable probes to examine deficits and the putative underlying loci of pathology, it is useful to note that the patients are similar to those with defined neurosurgical lesions of the ventromedial prefrontal cortex. In a classic paper, Eslinger and Damasio (1985) described a patient (EVR) who by the age of 35 years was successful in a professional occupation, happily married and the father of two children. After successful resection of an orbitofrontal meningioma involving a bilateral excision of the orbital and lower mesial cortices, he could not meet personal and professional responsibilities. He was unable to hold down his job, and his marital life eventually deteriorated until his wife left home with the children and filed for divorce after 17 years of marriage. Despite these problems, results of neurological examination and standard cognitive neuropsychological assessment were entirely normal (e.g. normal performance on the WCST and Word Fluency Test) and his intelligence was superior [97th percentile on the Wechsler Adult Intelligence Scale (WAIS)]. Clinically, lesions of the dorsolateral frontal lobe have traditionally been found to tend to affect planning and other higher cognitive functions, whereas lesions of the orbitofrontal cortex with its connections with the limbic system tend to have a greater effect on social behaviour and emotion (e.g. Sarazin et al., 1998). A paradigm has been developed by Bechara and his coworkers to bring out the cognitive deficits in patients such as EVR. This paradigm requires subjects to sample repeatedly from four decks of cards. After each selection, they receive a given amount of reward consisting of facsimile money. Two of the four packs produce large payouts with larger penalties, but the other two packs produce small payouts with smaller penalties. The most productive strategy is therefore to sample from the two packs with small payouts and penalties, which indeed control subjects adopt. Patients with damage to the ventromedial, but not dorsolateral, sectors

4 1472 S. Rahman et al. of prefrontal cortex have been found to persist in drawing cards from the high payout/penalty decks (Bechara et al., 1994). The task developed by Bechara and his colleagues is attractive because it models aspects of real-life decisionmaking that patients with ventromedial lesions find difficult. These decisions typically involve choices between actions associated with differing magnitudes of reward and punishment under conditions where the underlying contingencies relating actions to relevant outcomes remain hidden. To explain the nature of deficits in patients such as EVR, the somatic marker hypothesis has been posited (Damasio, 1995, 1996). The key idea in this hypothesis is that somatic marker signals influence the processes of response to stimuli at multiple levels of operation, some of which occur overtly (consciously in mind ) and some of which occur covertly (non-consciously, in a non-minded manner). Damasio speculates that the neural network required for such markers to operate includes the (i) ventromedial prefrontal cortices, (ii) central autonomic effectors such as the amygdala, which can activate somatic states in the viscera, vascular bed, endocrine system and non-specific neurotransmitter systems, and (iii) somatosensory cortices (SI, SII and insula) which can receive signals from the soma. In two further hypotheses a relationship between the cognitive and behavioural presentation of patients is also proposed. Rolls and colleagues (Rolls et al., 1994) have observed that patients with ventral frontal lesions make more errors in the reversal (or similar extinction) tasks and complete fewer reversals than control patients with damage elsewhere in the frontal lobes or other brain regions. Subjects were required to learn to obtain points by touching one stimulus when it appeared on a video monitor but withholding a response when a different visual stimulus appeared. After the subjects had acquired the visual discrimination, the reinforcement contingencies unexpectedly reversed. Patients with ventral frontal (orbitofrontal) lesions often reported verbally that the contingencies had changed, but were unable to alter their behaviour appropriately. Some aspects of the performance of the patients were particularly intriguing. The perseverative touching of a previously rewarded stimulus observed in humans appeared to be consistent with work with non-human primates showing impaired reversal and extinction after orbitofrontal lesions. Furthermore, the performance on reversal and extinction tests in the group of patients with damage to the ventral part of the frontal lobes was found to correlate highly with responses to a questionnaire examining the degree of disinhibited and socially inappropriate behaviour. Rolls therefore argued that neuropsychological deficits in patients with ventral frontal lesions might be manifested as disorders of emotion, if various emotions are to be considered as states elicited by rewarding and punishing stimuli; for example, fear may be a state elicited by stimuli associated with punishment (after Mowrer, 1947), and joy may be a state which is associated with rewarding stimuli. Therefore, the difficulty in modifying stimulus reward associations may contribute to the emotional changes shown in daily life. This may be consistent with the hypothesis that the orbitofrontal cortex is normally involved in executing behaviour when the behaviour is performed by evaluating the associations of environmental stimuli with reinforcement (Rolls, 1996b). Interestingly, the same patients with ventral frontal lesions that Rolls described performed normally on the Tower of London task. In an alternative hypothesis, proposed by Plaisted and Sahakian (1997), the reversal learning deficits are also considered, and its starting-point is the well established finding that damage to the prefrontal cortex results in loss of inhibitory control over inappropriate responses to any current situation. Recently, in the primate literature, Dias and colleagues (Dias et al., 1996, 1997) have reported a double dissociation in the prefrontal cortex of affective and attentional shifts in the marmoset following excitotoxic lesions to the lateral and orbitofrontal cortex. Whereas damage to the lateral prefrontal cortex (Brodmann area 9) in monkeys causes a loss of inhibitory control in attentional selection (an extradimensional shift), damage to the orbitofrontal cortex in monkeys causes a loss of inhibitory control in affective processing (a reversal), thereby impairing the ability to alter behaviour in response to fluctuations in the emotional significance of stimuli. Plaisted and Sahakian, therefore, in explaining the marked deficits in social cognition observed in patients with fvftd, put emphasis upon an inability of orbitofrontal inhibitory mechanisms to suppress inappropriate behaviours elicited by the immediate environment. They argue that such disinhibition would prevent the selection of alternative and more appropriate action plans which are dictated by long-term goals. Behaviour would hence be dominated by the immediate emotional evaluation of the stimuli, regardless of any available emotional or somatic information about the consequences of more distal action plans. In the present study, we considered how the theories described above, which have tried to specify how dysfunction of the orbitofrontal cortex can lead to deficits in social cognition, may apply in this particular clinical disorder. We were therefore interested not only in identifying the cognitive deficits of patients with mild fvftd, but also in examining whether these patients were unimpaired on certain tests of executive and mnemonic function. One obvious implication of our study concerns the future development of tests sensitive to the cognitive impairments demonstrated by patients in the mild stages of fvftd. In relation to Damasio s hypothesis, we attempted to examine the precise nature of decision-making deficits using a task in which subjects were required to make choices or decisions between contingencies that were presented in a readily comprehensible visual format. This task allowed us to assess the degree to which deficits in decision-making shown by mild fvftd patients were sensitive to the quality of information available about the identity of the reinforced response, as well as differences in opportunities to earn reinforcement. Using this task, we were also able to model

5 Cognition in frontotemporal dementia 1473 a further feature of real-life decision-making. Many taxing decisions involve the commitment of resources (e.g. money, time) to particular plans at the expense of others; it is therefore important to weigh up available opportunities against each other, judge the relative probabilities of a successful outcome and then choose how much of current resources or reward to commit to the chosen strategy. The decision-making paradigm used in this study enabled us to assess how much of their resources (in this instance, an accumulated points score) subjects were willing to commit to a given contingency in the hope of earning yet more reinforcement. Secondly, in relation to the theories of Rolls (1994) and Plaisted and Sahakian (1997), we examined specifically whether patients with mild fvftd performed differently from matched controls in the reversal stages of an attentional set-shifting paradigm. This particular paradigm of visual discrimination learning, which also included intradimensional and extradimensional shifts, has been described in greater detail elsewhere (Roberts et al., 1988; Downes et al., 1989). It allowed us to examine whether patients with fvftd had specific deficits in the reversal stages, and, if so, what were the precise nature of these deficits. Methods Subjects Patients Eight male out-patients with a clinical diagnosis of fvftd and with mean age 57.9 years (SD 9.60 years) took part in the study. Written informed consent was obtained from all participants in this study prior to subsequent assessment. The study was approved by the Cambridge Local Research Ethics Committee (reference number LREC96/113). These patients attended the Memory and/or the Early Dementia Clinics at Addenbrooke s Hospital, Cambridge. These are multidisciplinary clinics where patients are seen by a neurologist, psychiatrist and neuropsychologist. All patients were examined by a senior neurologist (J.R.H.). The patients were diagnosed on the basis of criteria summarized in Table 1 (Gregory and Hodges, 1993, 1996), with some similarities to the Lund and Manchester criteria (Lund and Manchester Groups, 1994). Care was taken to include only patients with relatively mild FTD. Assessment of the cases included an informant history, physical examination, neuropsychological tests, routine screening blood tests and structural (CT or MRI) and functional (HMPAO-SPECT) brain imaging. A battery of traditional neuropsychological tests was also administered at various intervals following diagnosis. Table 2 summarizes all the available results from these tests, which were administered at a similar time to the computerized neuropsychological assessment and provided references for detailed descriptions of the tests themselves. Patients with a past history of known or suspected transient or cerebral ischaemic event or stroke, alcoholism, head injury or other Table 1 Criteria for the diagnosis of fvftd I. Presentation with an insidious and progressive disorder of personality and behaviour of at least 6 months duration characterized by at least five of the following features: loss of insight, disinhibition, restlessness, distractibility, emotional lability, reduced empathy or unconcern for others, lack of foresight and planning, impulsivity, social withdrawal, apathy or lack of spontaneity, poor self-care, reduced verbal output, verbal stereotypes or echolalia, perseveration (verbal or motor), features of the Klüver Bucy syndrome (gluttony, sexual hyperactivity). II. Historical evidence of preserved memory for day-to-day events. III. Neuropsychological evidence of disproportionate frontal dysfunction. IV. Absence of history of significant head injury, stroke, score.4 on the Hachinski scale, alcohol abuse, diagnosis of Parkinsonism or other movement disorder. V. Psychiatric phenomena may be present. major medical illnesses (e.g. cancer or thyroid dysfunction) were excluded. One patient was actively receiving medication during the study. Case H was receiving nabumetane, meptazimide and amitriptyline. One other patient had also been receiving medication, but this was only previous to the time of testing. Case D had been prescribed fluvoxamine to control extreme eating habits, but this was discontinued after the development of a tremor. Case D had also been prescribed thioridazine to improve his sleep pattern, but this was also discontinued. Control subjects To compare the performance of the patients with that of controls, we selected eight subjects from the MRC Cognition and Brain Sciences Unit s control panel, matched for age, sex and premorbid predictions of IQ, as measured by the National Adult Reading Test (NART) (Nelson, 1982). Subjects with a history of alcoholism, drug abuse, learning disability or neurological or psychiatric illness were excluded, as were subjects with clinically apparent hearing and/or visual handicaps liable to affect their performance. All patients and controls were able to complete the full battery of tests. Subject characteristics One-way analysis of variance (ANOVA) revealed that the two groups did not differ in terms of age [F(1,14) , P ] or in estimates of WAIS full-scale, verbal and performance IQs as predicted from the number of errors made on the NART [full-scale IQ: F(1,14) , P ; verbal IQ: F(1,14) , P ; performance IQ: F(1,14) , P ]. The fvftd patients and the control group were both administered the

6 1474 S. Rahman et al. Table 2 Performance of the patients in the study on a battery of background neuropsychological tests FTD patients Controls Comparison NART errors 16.3 (10.3) 13.1 (7.18) P Memory span (forwards) 6.38 (1.30) 6.63 (1.19) P Memory span (backwards) 5.13 (1.36) 4.88 (1.13) P Logical memory (immediate/24) 11.9 (3.66) 13.0 (4.57) P Logical memory (delayed/24) 7.90 (2.72) 9.55 (2.43) P TROG (number correct/80) 76.9 (2.61) 78.7 (2.00) P TROG (blocks correct/20) 18.0 (1.91) 19.4 (0.79) P VOSP: object (incomplete letters/20) 18.8 (1.33) (all patients above cut-off*) VOSP: object (silhouettes/30) 19.7 (5.13) (all patients above cut-off*) VOSP: object (object decision/20) 18.5 (1.64) (all patients above cut-off*) VOSP: space (position discrimination /20) 18.7 (1.03) (all patients above cut-off, except one*) VOSP: space (number location/10) 8.17 (1.94) (all patients above cut-off, except one*) VOSP: space (cube analysis/10) 9.33 (0.82) (all patients above cut-off*) Data shown are mean (standard deviation) values. Controls were taken from the MRC Cognition and Brain Sciences Unit control panel, and were matched on the basis of age and verbal IQ estimated by the NART. Controls 5 control subjects. Forward and reverse memory span from the Wechsler Memory Scale (WMS-R) (Wechsler, 1987); logical memory subtest from the WMS-R (Wechsler, 1987); TROG 5 test for the reception of grammar (Bishop, 1983); VOSP 5 Visual Object and Space Perception Battery (Warrington and James, 1991).*5% cut-off scores for subjects on the VOSP incomplete letters, silhouettes, object decision, position discrimination, number location and cube analysis tests are 17, 16, 15, 18, 7 and 6 (age,50 years) and 16, 15, 14, 18, 7 and 6 (age 50 years), respectively. Table 3 Subject characteristics FvFTD patients Control subjects Age (years) Mean 57.9 (9.60) 58.1 (9.60) Range Sex (M/F) 8/0 8/0 FSIQ 114 (8.38) 116 (6.28) VIQ 114 (9.33) 116 (6.97) PIQ 113 (6.64) 115 (5.01) MMSE score Mean 27.8 (2.49) 29.9 (0.35) Range Data shown are mean (standard deviation) values. FSIQ 5 predicted full-scale IQ; VIQ 5 predicted verbal IQ; PIQ 5 predicted performance IQ. MMSE (Folstein et al., 1975). One-way ANOVA revealed that the mean MMSE scores for two groups were significantly different [F(1,14) , P ]. Subject characteristics are detailed in Table 3. Patient case histories The clinical histories of each of the eight patients in the present study are described below. A summary of the clinical characteristics of these patients is provided in Table 4. Case A This patient is a married financial director (born in 1942) without past personal history of psychiatric illness but with a family history of suicide (maternal grandfather) and depression (maternal aunt). He was referred to a psychiatrist by his general practitioner in October 1991, after his wife had received a letter from his employers stating that his job was in jeopardy because of his extremely poor performance. His wife provided a 3-year history of concern regarding his deterioration in intellectual function and an insidious personality change characterized by increasing irritability, antisocial and inappropriate behaviour (e.g. retiring to bed during visits from friends). His wife also reported that he showed loss of judgement with his personal finances, running up considerable debts on his credit cards and buying unnecessary, expensive items. He lacked any interest in his wife and children, and his driving had also become erratic, such that his family felt unsafe with him at the wheel. The patient showed no insight into his problems and was mystified by the letter from his employers. An assessment by a psychiatrist found no evidence of depression or formal psychiatric disorder. He was referred to the Memory Clinic in Examination and all investigations including routine haematology, biochemistry, EEG, CT and MRI were normal. The SPECT scan was minimally abnormal, with a suggestion of diffuse patchy hypoperfusion. Neuropsychological evaluation, with the exception of verbal fluency, was normal. A tentative diagnosis of fvftd was made on clinical grounds. Since 1991 there has been a gradual decline in his personality and behaviour. By March 1995 he had become extremely apathetic and had begun to overeat, gaining several stones in weight. He currently spends his time at home watching TV. He does not engage in any spontaneous conversation. He attends to a few specified household chores, which he performs in a rigid and highly stereotyped manner, e.g. hoovering at exactly 4 p.m. daily. He has also become sexually disinhibited, increasingly antisocial and shows no interest in his family. He has occasional urinary incontinence. Subsequent MRI imaging in 1995 and 1998 showed

7 Cognition in frontotemporal dementia 1475 Table 4 Summary of clinical characteristics of the patients in the study Initial Age, sex Length of time Relevant aspects of MMSE Neuroimaging since clinical family history diagnosis (years) A 54, M 7 Maternal grandfather died 29 SPECT (1991): minimally abnormal with a suggestion of by suicide; depression diffuse patchy hypoperfusion. (maternal aunt) SPECT (1995, 1998): bifrontal hypoperfusion. MRI (1995, 1998): progressive frontal lobe atrophy. B 70, M 1 28 SPECT (1998): reduced perfusion in left frontal and temporal regions. MRI (1998): marked frontal lobe atrophy. C 58, M 3 22 SPECT (1995): dramatic extensive reductions in perfusion in the frontal and anterior temporal regions. SPECT (1997): marked progression in the changes in the frontal lobes. MRI (1997): gross degree of frontal atrophy. D 55, M 5 29 SPECT, CT, MRI (1993): normal. SPECT (1997): reduced perfusion in both frontal cortices (in particular the inferior part and especially on the right). E 53, M 1 28 CT (1996): marked frontal lobe atrophy. SPECT (1997): marked reduction in perfusion bilaterally to the frontal lobes. F 43, M 7 Paternal grandparents and 30 CT (1991): selective frontal atrophy. uncle all died by suicide SPECT (1991): patchy reduced uptake of tracer. G 73, M 2 Mother died at 27 CT (1996): mild degree of cerebral atrophy. age 81 years, having MRI (1996): a degree of atrophy more prominent in the possibly been suffering region of the anterior portion of the left temporal lobe and from dementia for the the frontal lobe bilaterally. preceding 5 years H 57, M 1 Father died of motor 29 MRI (1998): degree of global atrophy, with accentuation neuron disease (mother in the frontal and anterior temporal lobes, particularly on still alive and well) the left side. MMSE 5 Mini-Mental State Examination, M 5 male. progressive frontal lobe atrophy. Repeat HMPAO-SPECT scans in 1995 and 1998 have also revealed bifrontal hypoperfusion. Case B This patient (born in 1927) is now aged 70 years. He left school at 15 years and became an apprentice, but only a few years later, after joining the Burma Oil Company in India, he underwent rapid promotion so that within 10 years he was a manager and in charge of several hundred employees. At that time, he was under very considerable strain and began to have peculiar attacks of loss of consciousness which would last for a number of hours. These were subsequently diagnosed as strokes, although this seems dubious. He was unwell enough to be sent back from India and underwent extensive investigation in the early 1970s. In about 1975 he had a brain scan and was told that this showed brain atrophy due to a presenile dementia. In the 10 years after his retirement he underwent a progressive change in personality, behaviour and social interaction. The features of this included poor motivation and apathy, with a complete loss of interest in his wife and inability to organize any activity. He took up a couple of jobs but they lasted for only a few weeks. His current state is of complete inertia. Many days can pass in the house without him initiating conversation, although he still reads novels and apparently can be motivated if taken to museums, particularly if they relate to the Second World War. There have been some minor changes in eating and his manners. Approximately 15 years ago there was a period of sexual disinhibition but he subsequently appears to have lost his libido. His wife describes him as being non-existent. On examination, there were a number of physical signs. Most striking was his restriction in eye movements. The horizontal range appeared limited, but he had virtually nonexistent upgaze. There was also poor balance and a hint of some rigidity in the limbs. There was a positive pout, no clear grasping and no limb apraxia. On basic cognitive testing, he was fully orientated with good registration of information and rapid serial calculation. He had good

8 1476 S. Rahman et al. visuospatial function and naming ability; he was also very good on current events. By contrast, his verbal fluency was markedly reduced. In March 1998 he underwent a number of investigations. An MRI scan confirmed marked frontal lobe atrophy, compatible with the diagnosis of FTD. A SPECT scan revealed reduced perfusion in the left frontal and temporal regions. All other routine investigations were normal. Case C This patient (born in 1939) presented in He admitted to no symptoms, although he did realize that his family had been finding it difficult to cope with him. His family reported that he was letting things go on the farm, whereas previously he was methodical and well organized. In addition, he had become excessively socially withdrawn and was initiating very little conversation at home. He needed reminding to change his clothes and to wash, and his dietary habits had changed in that he now liked to consume lots of sweet food. He came from a very large family with 10 siblings, an elderly mother and a father who died of a heart attack at the age of 75 years. There was no family history of dementia. He had a surly, unsociable manner, but this impression was occasionally dispelled by a sudden smile. Examination revealed a bilateral palmomental reflex, pout reflex and glabellar tap, and difficulty with Luria three-step commands. He was well orientated with a normal digit span, but had reduced verbal fluency. His acquisition and recall of a name and address was surprisingly good. On formal neuropsychological assessment, language skills were well preserved, although his conversational ability was reduced. Traditional assessment of frontal lobe function revealed disorganized and somewhat erratic performance. In 1997, formal neuropsychological assessment revealed that he was unable to perform normally on formal tests of memory, but performance on tests of perceptual and visuospatial analysis was within normal limits. Unlike previously, he failed to sort the Weigl tokens according to shape and colour. In 1995 a cerebral perfusion SPECT scan revealed dramatic extensive reductions in perfusion in the frontal and anterior temporal regions. Therefore, taken with the clinical findings, a tentative diagnosis of fvftd was made. A cerebral perfusion SPECT scan was repeated in February 1997, which showed quite a marked progression in the changes in the frontal lobes. An MRI scan performed at that time demonstrated a gross degree of frontal atrophy. pulling out into the path of oncoming traffic. He was also noted to have lost his sense of humour. At work it was stated that he had difficulty in coping with the children during class, but he had no problems in completing his regular newspaper crossword. In February 1990 he had suffered a myocardial infarction from which he made a full recovery, and was found to have elevated cholesterol levels. By March 1993 he was noted to have substantial difficulty in reading music and playing the piano simultaneously, and had problems playing and conducting, tasks at which he had previously excelled. He showed poor judgement and forethought. His affect was flat and his behaviour stereotyped, and he started to have difficulty controlling his food intake. He was referred to the Memory Clinic at Addenbrooke s Hospital and underwent full assessment in A psychiatric assessment found no evidence of functional illness, in particular no evidence of an affective disorder. There was no family history of psychiatric disorder, but the patient had suffered an episode of depression in his 30s. On physical examination he had positive pout and palmomental reflexes. Neuropsychological testing showed that his verbal and performance IQs were well maintained. His memory, naming and perceptual abilities were normal. On two simple frontal lobe tests (verbal fluency and a modified version of the WCST), he showed no deficit. Routine haematological and CSF examination were normal, as were SPECT, CT scan and MRI. A tentative diagnosis of fvftd was made on the basis of the history. He has been assessed neuropsychologically on a number of occasions. By July 1993 he was gaining weight excessively and would often eat two full meals at one sitting. At school, he began to behave in a disinhibited manner, making inappropriate sexual remarks. Because of this he was asked to leave his job. By December 1994 he had become markedly apathetic and, uncharacteristically, would sit watching TV for hours at a time. His conversational abilities were reduced and he became increasingly rigid and repetitive in his behaviour. He also began to eat inedible substances. In September 1997 a cerebral perfusion SPECT scan was undertaken, and this revealed a reduction in perfusion in both frontal cortices (in particular the inferior part, and especially on the right). The rest of the cerebral perfusion appeared within normal limits. At the present time (1998) he does little except keep to the routine of taking his dog for a daily walk, the route of which is never varied. He does not help spontaneously in the house, and his self-care and social manners have deteriorated greatly. Case D This patient, a married music teacher born in 1942, was referred in November 1992 by his general practitioner following concern regarding his behaviour from both relatives and colleagues at work. It was reported that he had become increasingly vague and subdued, appeared slowed and lacked initiative. His driving had become rather hazardous, e.g. Case E Patient E (born in 1944) was referred to the Memory Clinic from the Burton Hospitals NHS Trust in 1997 with the provisional clinical diagnosis of fvftd. He himself denied any symptoms, but his wife and close friend reported a gradual change in personality over the last 2 3 years. He began to act in rather strange ways, for instance, grabbing

9 Cognition in frontotemporal dementia 1477 someone else s steak at dinner because it looked larger than his own, and getting up and leaving half-way through a dinner party to watch TV. They noticed that he was becoming progressively withdrawn and at other times he would act in a rather disinhibited and blunt fashion. He also had difficulty with planning and organizing and dealing with more than one thing at once. There was no change in his self-care or eating habits, and his wife did not observe any obsessive or compulsive behaviour. He did express a slightly paranoid idea of conspiracy, but this may have been related to the fact that he felt that there was nothing wrong with him and therefore could not understand why he was being referred for investigations. Despite the changes in personality and behaviour, his memory remains very good and he is extremely knowledgeable about Manchester United Football Club and also the stock market. There is no past history of significant head injury. He has never abused alcohol. There is no family history of dementia. His mother is aged 78 years and still alive. His father died of a brain tumour. He has one sister, who is well, and a daughter aged 23 years. On examination, he was found to have a positive pout response but no other release signs. He was well orientated in time and place, and his memory was good. There were no language or visuospatial deficits, but he did show a marked reduction in verbal fluency for both letters and categories. More formal neuropsychological testing showed borderline performance on tests of memory, with normal language and perceptual ability but poor performance on tests sensitive to frontal lobe function. A cerebral perfusion SPECT scan performed at Addenbrooke s Hospital in March 1997 showed a marked reduction in perfusion bilaterally to the frontal lobes, which concurred with the results of a CT scan performed at Burton NHS Trust in August 1996 that had revealed marked frontal lobe atrophy. Case F This patient was initially referred to a psychiatrist in May 1989, at the age of 35 years, with a short history of odd ideas. He believed that his workmates were smiling at him in suggestive ways and that he was a special and wonderful person. He also believed that his thoughts were being interfered with by an electronic gadget and that the TV and radio were making specific reference to him. Furthermore, he believed that his actions were being controlled by an external force. There was no past psychiatric history, and he was physically fit. There was a strong family psychiatric history; his paternal grandparents and uncle all died by suicide. His developmental, sexual and occupational histories were unremarkable, except that they were characterized by stability. On further questioning, his wife reported a gradual onset of mental inflexibility and social withdrawal, a decline in his sense of humour and a loss of interest in hobbies for ~1 year before this psychotic illness. He made a good response to neuroleptic medication but never developed any insight into his illness. Over the next 2 years he was maintained on neuroleptic medication, and any attempt at withdrawal resulted in reappearance of his paranoid beliefs. Over this period, however, his behaviour changed substantially. He began to collect and hoard women s underwear and pornographic magazines, which he hid in the cupboard of his bedroom, but which were obvious to his wife whenever she opened the cupboard doors. He became sexually disinhibited, masturbating frequently. Over the next few months all his psychotic symptoms disappeared and he was left apathetic and disinhibited with poor foresight and planning. He appeared very flat all of the time, and even during a crisis at home he remained unperturbed. During all of this, he denied any problems at all and said that he was fine. He also denied any problems with his memory, thinking or organizational abilities. On physical examination he was found to have pout and grasp reflexes. Bedside cognitive testing revealed him to be fully orientated, with good registration and recall. He showed reduced verbal fluency. Detailed neuropsychological tests (e.g. the WCST) confirmed the presence of frontal dysfunction. A CT scan revealed selective frontal atrophy, but a SPECT scan was unhelpful and showed only patchily reduced uptake of tracer. In August 1991 a diagnosis of fvftd was made. During the period after diagnosis his clinical condition has deteriorated further. He has become gradually less able to organize himself or activities, is extremely impulsive and has occasional urinary incontinence. He requires constant supervision and is about to embark on respite admissions to his local psychiatric hospital. Case G This patient, a retired chartered accountant born in 1924, was referred to Addenbrooke s Hospital in March 1996 with a 3-year history of a gradual decline in motivation and interest. He had demonstrated a marked reduction in spontaneous conversation, often appeared apathetic and showed no empathy with family members. He had previously been a rather flamboyant Australian. His wife reported that his dayto-day episodic memory remained good. He denied any symptoms. Interestingly, his mother died in a home at the age of 81 years, possibly (in retrospect) having been suffering from dementia for the preceding 5 years. His father had a stroke in his 60s. He has seven siblings. He is an ex-smoker and drinks alcohol only occasionally. On examination he was found to be fully orientated. His digit span was 7 forwards and 7 backwards. His verbal fluency was a little low for his background. Formal neuropsychological assessment showed quite marked anomia, which seemed to be in word-finding rather than semantic disorder. He performed quite well on visuospatial tests. His recognition memory for faces was at a good level but his recognition of memory for words was impaired. On tasks traditionally considered to be sensitive to frontal lobe function (verbal fluency, bimanual hand co-ordination, drawing to rule, Weigl, Stroop), the impression was of frontal inefficiency,

10 1478 S. Rahman et al. although he did not show marked failure. A CT scan showed a mild degree of cerebral atrophy. An MRI scan demonstrated a degree of atrophy more prominent in the region of the anterior portion of the left temporal lobe, and the frontal lobe bilaterally. A clinical diagnosis was made of fvftd. Case H This patient (born in 1940) was referred via the Orthopaedic and Psychiatric Departments to the Memory Clinic at Addenbrooke s Hospital, Cambridge. During a recent stay for investigation of sciatica and back pain, it was noted that his behaviour and affect on the ward were odd. For instance, he reported that his normal amount of exercise consisted of running at least 100 miles per week and he made inappropriately personal comments to the staff. He had been employed in a local security company, but had left his job 3 years previously, at the time of his marital break-up. He was unable to give a clear explanation of these events. He made no complaints related to his mental or intellectual state. It was difficult to elicit a clear history because of his circumlocutory style, and his tendency to tell rather grandiose tales about his time in the Army. For example, he described an incident in Lebanon when he awoke suddenly in the middle of the night, woke his 27 fellow soldiers and killed 900 enemy soldiers who were creeping up on them. His son reported that he had always exaggerated about his Army career, but that he had started telling extremely unlikely tales only about a year previously. His ex-wife had also noted a change in personality in the last few years, with an exaggeration of his premorbid state. He has become disinhibited, with a tendency to tell inappropriate and rude jokes, with episodes of verbal aggression and emotional lability. His father died of motor neuron disease, further details of which are pending. His mother is aged 79 years and is still alive and well, and he has five siblings who are all fit. He has been quite a heavy smoker for all of his life, and had consumed alcohol whilst in the Army but not more recently. On examination there were no physical signs. He was well orientated. His acquisition of a name and address was poor. He was unable to name the Prime Minister, the last Prime Minister or the American President. His verbal fluency was poor; he was able to suggest the names of only nine animals in a minute. He scored within the normal range on the Cognitive Estimates Test. He had a striking impairment in abstract thinking, as he was quite unable to interpret proverbs and unable to answer even the simplest questions on similarities. Formal neuropsychological assessment showed him to have a rather low premorbid IQ but, even allowing for this, his performance on tests of naming and drawing was poor, and he was unable to complete the WCST. In February 1998, the clinical diagnosis was made of early fvftd. Blood tests and CSF investigation revealed no abnormality. An MRI scan showed a degree of global atrophy, with accentuation in the frontal and anterior temporal lobes particularly on the left side. EMG was normal. Computerized neuropsychological assessment Procedure The majority of tests were taken from the Cambridge Neuropsychological Test Automated Battery (CANTAB). These tests were administered using a portable Datalux 486 microcomputer fitted with a touch-sensitive screen. Subjects were seated comfortably ~0.5 m from the touchscreen, and it was explained to them that they would have to respond to stimuli presented on the monitor screen by touching them. In order to introduce subjects to the apparatus, they were initially given a motor screening task in which they were asked to respond to a series of 10 flashing crosses presented at varying locations on the monitor screen by placing the index finger of their preferred hand on the centre of the cross. This followed a short demonstration by the tester in which three consecutive crosses were touched. After completion of this task, subjects were given the following cognitive tasks in the order described below. Pattern and spatial recognition (Sahakian et al., 1988) Two tasks designed to assess recognition memory for both patterns and spatial locations were administered. In the pattern recognition task, subjects are presented with a series of 12 abstract patterns and their task is to remember them. Following a 5-s delay, each pattern is re-presented in reverse order, paired with a novel pattern, and subjects are asked to touch the pattern they have seen previously. This procedure is then repeated with a further 12 patterns. In the spatial recognition task, five squares are presented sequentially in different locations around the screen. In the recognition phase, subjects are presented with a choice of two squares in different locations, one of which is novel; they must touch the location in which the square has previously appeared. The procedure is repeated a further three times. Spatial span (Owen et al., 1990) This is a computerized version of the Corsi Block Tapping task (Milner, 1971). Briefly, each trial begins with nine white boxes presented in fixed locations on the monitor screen. Initially, two of the boxes change colour, one after the other, in a predetermined sequence. The end of the sequence is indicated by a tone. Subjects are then asked to point to the boxes in the order in which they have changed colour. After successful completion of a sequence, the number of boxes changing colour increases by one, up to a maximum of nine. The test is terminated when three consecutive failures occur at any one sequence length. During each trial, the number in

11 Cognition in frontotemporal dementia 1479 the bottom left corner of the screen indicated the length of the current sequence. The spatial span was calculated as the highest level at which the subject has successfully recalled at least one sequence of boxes. This measure was used to assess the ability of subjects to hold information on-line in order to plan a series of moves in tasks such as the spatial working memory task and the one-touch Tower of London task. Spatial working memory (Owen et al., 1990) This is a test of spatial working memory for humans which is formally analogous to the Olton radial arm maze (Olton and Samuelson, 1976), a spatial working memory task for the rat. Both tasks require memory for locations already visited. In this task, subjects are required to search through a number of coloured boxes presented on the monitor screen (by touching each one) in order to find blue tokens, which are hidden inside. On any one trial, only a single token is hidden in one of the boxes. Once found, the next token is hidden. The key instruction is that once a token has been found within a particular box, then that box will not be used again to hide a token. Two types of error are possible. First, a subject may return to open a box in which a token has already been found (a between-search error). Secondly, a subject may return to a box already opened and shown to be empty earlier in the same trial (a within-search error). There are four trials with each of four, six and eight boxes. The task is scored according to the number of between- and within-search errors at each level of difficulty and also for the use of an efficient search strategy (Owen et al., 1990). Both scores are a measure of spatial working memory, but the between-search error is a more stringent one as the subject is required to remember across searches which boxes had contained the blue tokens while conducting a new search (Joyce and Robbins, 1991). A particularly efficient strategy for completing this task is to follow a predetermined search sequence, starting with a particular box and then returning to start each new sequence with that same box as soon as a token is found (editing the sequence when a token is found in that box). The extent to which such a strategy is used is estimated from the number of search sequences starting with a novel box for just the more difficult six- and eight-box problems. The total of these scores provides a measure of strategy for each subject, a high score (many sequences starting with a novel box) representing poor use of a strategy and vice versa. Visual discrimination learning/attentional set shifting (Downes et al., 1989) The task requires subjects to learn a series of two-alternative forced-choice visual discriminations using feedback provided automatically by the computer. The task is composed of nine stages presented in the same fixed order. For each stage, continuation to the next stage is dependent on a criterion of six successive correct discriminations being reached. If criterion is not reached at the 50th trial of a stage, the test is discontinued and the subject is not able to proceed to the following stage. The task starts with a simple discrimination and its reversal for stimuli varying in just one dimension (e.g. two different white-line configurations). A second, alternative dimension is then introduced (purple-filled shapes) and compound discrimination and its reversal are tested. To succeed, subjects must continue to respond to the previously relevant dimension whilst ignoring the presence of the new, irrelevant dimension. At the intradimensional shift stage, novel exemplars of each of the two dimensions are introduced and subjects must continue to respond to one of the two exemplars from the previously relevant dimension. Following another reversal, the extradimensional shift and its reversal are presented, again using novel exemplars of each stimulus dimension. In order to succeed at the extradimensional shift stage, the subject must shift attentional or response set to the previously irrelevant stimulus dimension, whilst ignoring the previously relevant dimension. The extradimensional shift stage is akin to a change in category in the Wisconsin Card Sorting Test. One-touch Tower of London (Owen et al., 1995) This task is a modified version of the CANTAB Tower of London task, itself ultimately derived from the Tower of London task developed by Shallice and McCarthy (Shallice, 1982). Subjects are first trained with a number of problems from the original Tower of London task and are then presented with the following modified task. Two sets of coloured balls appear on the screen, one in the upper half of the screen and one in the lower half. They are described to the subjects as snooker or pool balls as they appear to hang in vertical pockets. There are three pockets in each half of the screen, capable of holding three balls, two balls and one ball, respectively. The numbers 1, 2, 3, 4 and 5 are printed in large boxes across the bottom of the screen. At the start of each trial, the upper and lower pockets appear empty on the screen. After a delay of 1 s, a tone alerts the subject to the screen and a red ball, a blue ball and a green ball are placed in a predetermined arrangement in the pockets of the upper and lower displays. The subjects are instructed to examine the positions of the balls on the screen and then to imagine how they might rearrange the balls in the lower display to match those in the upper display without actually moving any of the balls. For any given arrangement, the subjects are asked to find the solution that requires the minimum number of moves, and then to press the corresponding number on the bottom of the screen. If a subject s first response is incorrect, he is required to try again until the correct number is selected. The importance of accuracy rather than speed of response is emphasized. After the practice problems, subjects are given test trials of varying problem difficulty arranged in a constant, pseudorandom order.

12 1480 S. Rahman et al. Fig. 1 Decision-making task. This figure shows a typical display from the decision-making task (see text for details). Note that the ratio of red to blue boxes (i.e. the relative probabilities of the two outcomes) changed from trial to trial. Decision-making (Rogers et al., 1999a) Figure 1 shows a typical display from a decision-making task. The subject is told that the computer has hidden at random a yellow token inside one of the red or blue boxes arrayed at the top of the screen, and that he has to decide whether this token is hidden inside a red box or a blue box. The subject indicates his decision by touching either the response panel marked red or the response panel marked blue. After making this initial choice, the subject attempts to increase a total points score, shown in green towards the left-hand side of the display, by placing a bet on this choice being correct. The available bets appear in a sequence, one after another, centred in the box positioned towards the righthand side of the display. Each bet is displayed for a period of 5 s before being replaced by its successor, and the subject is free to select any bet by touching the box in which the sequence appeared. Immediately following such a selection, one of the boxes opens to reveal the location of the token, accompanied by either a You win! message and a short rising musical scale, or a You lose! message and a low tone. If a subject chooses the correct colour the bet placed is added to the total points score; if the subject chooses the wrong colour the bet is subtracted. The subject is instructed to treat the points as being valuable and to accumulate as many as possible during the test. However, no real monetary significance is attached to the total points accumulated by the end of the task. The subject performs the task in two separate conditions. In one condition ( ascending ), the first bet offered is small but is replaced by larger and larger bets until the subject makes a selection. In the other condition ( descending ), the first bet offered is large but is replaced by smaller and smaller bets. Each bet represents a fixed percentage of the current total points score, although this is never made clear to the subject. Five bets are offered on each trial, so that in the ascending condition the order of available bets is as follows: 5%, 25%, 50%, 75% and 95%. In the descending condition the order is reversed. In both conditions, each bet is presented with a short tone whose pitch corresponds to the size of the bet: higher tones accompany larger bets and lower tones accompany lower bets. If the subject fails to select a bet by the end of a sequence, the last bet is chosen automatically.

13 Cognition in frontotemporal dementia 1481 The ascending and descending conditions both consist of four sequences of nine displays. At the start of the sequence, the subject is given 100 points and is asked to increase this total by as much as possible. If a subject s score falls to just one point the current sequence ends and the next begins. The order of presentation of the two conditions is counterbalanced across the two subject groups. The same set of fixed, pseudorandom sequences is presented to both the subject groups. Three features of this task are important. First, by manipulating the ratio of red and blue boxes from trial to trial it is possible to examine a subject s decision-making behaviour over a variety of differentially weighted contingencies. For example, some ratios (e.g. 9 red : 1 blue) represent two contingencies that are quite unequal in terms of the probabilities associated with the two possible outcomes. In contrast, other ratios (e.g. 4 red : 6 blue) represent contingencies that are more balanced. Therefore, a subject s choice of contingency, speed of choice and size of bet are expected to differ as a function of the ratio of red to blue boxes. Secondly, by allowing subjects to determine for themselves how much of their points score they wish to bet after each red/blue decision, we were able to assess individual efficiency in placing accumulated reinforcement at risk with the intention of acquiring more reward. For example, one might suppose that a ratio of 9 red : 1 blue represents an opportunity to bet more points on a red decision in order to gain more reward, while a ratio of 6 blue : 4 red represents a situation in which more conservative behaviour might be appropriate. In this way, the aspect of decision-making in real-life situations can be modelled, i.e. individual choices about the investment of resources (such as time and money, etc.) in available opportunities. Finally, offering the bets in both an ascending and a descending order enables the isolation of merely impulsive behaviour from genuine risk-taking behaviour (after Miller, 1992). One would expect a subject, if impulsive, to be unable to withhold manual responses to the sequence of bets as they are presented, and therefore to choose early bets in both the ascending and the descending sequence. In contrast, one would expect a subject, if actively risk-taking, to choose late bets in the ascending condition and early bets in the descending condition. A large difference between the mean percentage bet in these two conditions therefore indicates impulsivity whereas a low difference indicates risk-seeking. The three principal measures in this task are: (i) the speed of decision-making, i.e. how long it takes the subject to decide which colour of box is hiding the token, as measured by the mean deliberation time; (ii) the quality of decisions as an ineffective approach is to bet continuously on the least likely of the two possible outcomes, one measure is how much of the time the subject chooses the most likely outcome; and (iii) risk adjustment, i.e. the rate at which a subject increases the percentage of the available points bet in response to more favourable ratios of red to blue boxes (e.g. 9 red : 1 blue versus 4 red : 6 blue). Table 5 Group comparisons of visuospatial memory tasks Group Pattern Spatial Spatial span recognition recognition fvftd patients 77.6 (15.7) 73.1 (16.2) 4.88 (1.46) Control subjects 81.8 (7.70) 81.9 (8.84) 5.63 (0.92) Data shown are mean (standard deviation) values. Data analysis The majority of the experimental data were analysed using the Statistical Package for the Social Sciences [SPSS V6.1.1] (SPSS Inc., Chicago, Ill., USA) running on an Apple Power Macintosh 6100/66 computer. Data were transformed where appropriate (Tukey, 1977). Some data were not suitable for parametric analysis even after transformation, and were therefore analysed non-parametrically. The data shown in the figures always represent untransformed values. For most of the dependent variables, ANOVA was used to compare the performance of patients with fvftd with that of the controls. The ANOVA model was a two-factor design that included a between-subjects factor (group) and a withinsubject factor (e.g. difficulty level). In the analysis of the decision-making task, initially the principal measures were subjected to repeated measures ANOVA with the following between- and within-subject factors: group (fvftd versus controls); order of condition (ascending/descending versus descending/ascending); condition (ascending versus descending); decision (red versus blue); ratio (6 : 4 versus 7:3versus 8:2versus 9 : 1). The mean deliberation times (i.e. the speed of decision-making) were log 10 transformed to ensure homogeneity of variance, while the proportions of trials on which subjects chose the most likely outcome (i.e. the quality of decision-making) were arcsin-transformed, as is particularly appropriate whenever the variance is proportional to the mean (Howell, 1997). In those instances in which the additional assumption of homogeneity of covariance in repeated measures ANOVA was violated, as assessed using the Mauchly sphericity test, the number of degrees of freedom against which the F term was tested was reduced by the value of Greenhouse Geisser epsilon (Howell, 1997). Results Computerized neuropsychological assessment Pattern and spatial recognition memory The mean percentage scores for both the pattern recognition and the spatial recognition memory test are presented in Table 5. The mean percentage correct scores were compared, following arcsin transformation [f(x) 5 2 arcsin x. One-way ANOVAs revealed that there was no significant difference in the percentage of correct scores between the patients and controls on both the pattern recognition task [F(1,14) , P ] and the spatial recognition task [F(1,14) 5

14 1482 S. Rahman et al. Table 6 Spatial working memory data Group BSE BSE BSE Strategy (four (six (eight score boxes) boxes) boxes) fvftd 2.75 (3.41) 12.9 (9.89) 26.6 (10.6) 36.6 (4.78) Control subjects 1.13 (1.89) 9.50 (9.17) 22.0 (10.0) 31.8 (6.23) Data shown are mean (standard deviation) values. BSE 5 number of between-search errors. 1.64, P ]. The mean response latencies for both tasks were also compared, following logarithmic transformation [f(x) 5 log 10 (x)] to reduce positive skew in the distribution. Further one-way ANOVAs revealed that the fvftd patients were significantly slower on the pattern recognition task [F(1,14) , P ] and highly significantly slower on the spatial recognition task [F(1,14) , P ]. Spatial span Mean spatial spans for the two groups are also presented in Table 5. There was no significant difference between the two groups in their spatial spans [F(1,14) , P ]. The total numbers of errors for the fvftd and control groups were 12.6 (SD ) and 12.6 (SD ), respectively. The total usage errors for the fvftd and control groups were 3.50 (SD ) and 3.13 (SD ), respectively. There was no significant difference between the two groups in the total number of errors [F(1,14) 5 0, P 5 1] or in the total number of usage errors [F(1,14) , P ]. Spatial working memory Data for the spatial working memory task are presented in Table 6. The mean number of between-search errors made for each of the four-, six- and eight-box stages of the task was subjected to repeated measures ANOVA with group and level (four, six or eight boxes) as factors. There was no main effect of group [F(1,14) , P ] but there was a significant main effect of difficulty [F(2,28) , P, ]. There was no significant interaction between group and difficulty factors [F(2,28) , P ]. Within-search errors were at a minimal level in both groups and hence no meaningful analysis was possible. One-way ANOVA revealed no difference between the two groups in the use of an efficient search strategy [F(1,14) , P ]. The relationship between the strategy score and the sum of between-search errors obtained in the four-, six- and eight-box problems was further examined using Pearson s product moment correlation coefficient, r. As expected, there was a significant correlation between the strategy score and the number of between-search errors for patients [r , P ]. The correlation between the strategy score and the number of between-search errors tended to significance for controls [r , P ]. Visual discrimination learning/attentional set shifting All the subjects and the controls attempted each of the nine stages of the attentional set-shifting paradigm. Only two subjects (patients E and H) failed to reach criterion at the final stage, the extradimensional shift reversal; their error scores for this particular stage, therefore, represent a conservative estimate of their deficit at this stage. Figure 2A shows the errors for each stage of the task for both the group of patients and for the controls. Two-way ANOVA of the errors to criterion at the intradimensional and extradimensional shift stages, following square-root transformation [f(x) 5 (x 1 0.5)], showed no significant effect of group [F(1,14) , P ], a significant effect of stage [F(1,14) , P ] and no significant group 3 stage interaction [F(1,14) , P ]. The total number of errors at the compound discrimination reversal, intradimensional shift reversal and extradimensional shift reversal stages, as well as the total number of errors for the corresponding non-reversal stages, was calculated for each subject. Figure 2B shows graphically the average total number of errors made by the two groups in the reversal and non-reversal stages. As the assumptions of ANOVA were violated even after transformation, these data were analysed non-parametrically. To examine specifically whether the effect of reversal differed between the two groups, the total number of errors made on the non-reversal stages was subtracted from the total number of errors made on the reversal stages for each subject. A two-tailed Mann Whitney U test on this difference was then used, with a Bonferroni correction applied to the analysis such that the critical region was defined by P, The Mann Whitney U test revealed that the two groups differed significantly in the difference in the total number of errors made on the reversal and non-reversal stages [U 5 8, exact two-tailed P ]. Therefore, the patients with fvftd did appear to diverge in performance from the controls specifically in the reversal stages of the task, the patients with fvftd demonstrating a specific deficit on the reversal stages compared with controls. In addition, the number of trials before each subject changed his previously correct response (a perseverative error score) was noted for each of the three stages, compound discrimination reversal, intradimensional shift reversal and extradimensional shift reversal. The total number of perseverative errors made in these three stages was then calculated for each subject. Interestingly, one-way ANOVA revealed that there was no significant difference between the two groups in the total number of perseverative errors [F(1,14) , P ]. One-touch Tower of London The main measure of planning accuracy was the percentage of problems solved correctly at each level of difficulty. A graph of this measure for both patients and controls is shown in Fig. 3A.

15 Cognition in frontotemporal dementia 1483 Fig. 2 Attentional set-shifting task: (A) errors; (B) total number of errors at reversal stages of compound discrimination and intradimensional and extradimensional shift (CD_R 1 ID_R 1 ED_R) and non-reversal stages (CD 1 ID 1 ED). Error bars represent 1 SEM. Solid circles 5 fvftd; open circles 5 controls. Fig. 3 One-touch Tower of London task: (A) mean percentage correct; (B) mean latency. Error bars represent 1 SEM. Solid circles 5 fvftd; open circles 5 controls. These data were subjected to repeated measures ANOVA with group and difficulty level as factors. There was a highly significant effect of difficulty [F(3,42) , P, ] but no effect of group [F(1,14), 0.001, P ] and no group3difficulty interaction [F(3,42) , P ]. Another measure of planning accuracy was the mean number of attempts required before correct solution at each level of difficulty. These data were subjected to repeated measures ANOVA with group and difficulty level as factors, following square-root transformation [f(x) 5 (x 1 0.5)]. There was a highly significant effect of difficulty [F(3,42) , P, ] but no effect of group [F(1,14) , P ] and no group 3 difficulty interaction [F(3,42) , P ]. One further measure from this task was analysed following log 10 transformation. This measure, the mean latency at each level of difficulty, is shown in Fig. 3B. For the mean latencies overall, there was a strong tendency towards a significant effect of group [F(1,14) , P ], a highly significant effect of difficulty [F(3,42) , P, ] and a significant group 3 difficulty interaction [F(3,42) , P ].

16 1484 S. Rahman et al. Fig. 4 Decision-making (speed) task. Deliberation times associated with deciding which colour of box is hiding the yellow token, as a function of the red and blue boxes (see text for details). (A) Individual characteristics in the form of a bar chart showing data for each of the patients and the average performance of the controls. (B) Group characteristics. Error bars represent 1 SEM. Solid circles 5 fvftd; open circles 5 controls. Decision making Speed of decision-making. Figure 4A shows the mean deliberation times for the group of controls and for each of the eight patients associated with deciding which colour of box is hiding the yellow token, as a function of the ratio of red to blue boxes. Figure 4B shows the mean deliberation times for the groups of patients and controls. Notably, the fvftd group showed large and significant increases in the time needed to make their decisions, so that the mean deliberation time was 3325 ms for the controls but 6302 ms for the fvftd patients [F(1,12) , P ]. All subjects took more time to make their decisions on the first occasion that they completed the task compared with the second [average for the first occasion ms versus 4363 ms for the second occasion]. In other words, the latency was slower in the ascending condition for those subjects whose order of conditions was ascending descending and in the descending condition for those subjects whose order was descending ascending, yielding a significant two-way interaction between order and condition [F(1,12) , P ]. However, the three-way interaction between these two factors and that of group was not significant [F(1,12) , P ]. The mean time needed to decide the colour of the box was not significantly increased at any particular ratio of red to blue boxes, suggesting that the patients did not have particular difficulties in the face of relatively poor information about the identity of the rewarded response. Analysis of the enhanced performance of the group of fvftd patients and the controls revealed that neither age nor premorbid verbal IQ was a significant covariate of deliberation times (age: t , P ; verbal IQ: t , P ), and the results above were unaffected. Quality of decisions. Figure 5A shows for the group of controls and for each of the eight patients the percentage of trials on which subjects chose the most likely of the two possible outcomes (i.e. the colour with the greatest number of boxes), as a function of the number of red and blue boxes. It is noteworthy that case C, with marked hypoperfusion of the frontal, frontoparietal and anterior temporal regions revealed by SPECT and with a MMSE score of 22, demonstrated marked impairment in the quality of decisionmaking. Figure 5B shows the same measure (the quality of decision-making) for the groups of patients and controls. Averaging over group, the percentage of choices of the most likely outcome increased significantly at the more favourable compared with the less favourable ratios (ratio 6 : 4, 88.5%; ratio 7 : 3, 90.5%; ratio 8 : 2, 94.4%; ratio 9 : 1, 96.8%). Overall, there was no difference in the tendency to make optimal choices between the patients with fvftd and controls [F(1,12) , P ]. Analysis of the performance of the group of fvftd patients and the controls revealed that neither age nor premorbid verbal IQ was a significant covariate of the probability that the most likely outcome was chosen (age: t , P ; verbal IQ: t , P ), and the results above were unaffected. Risk adjustment. Figure 6A shows the percentage of their total points score that each of the eight subjects and the group of controls were prepared to risk in order to earn more points, as a function of the ratio of red to blue boxes. Most patients showed markedly increased bets compared with controls. Figure 6B shows the percentage of the points score for each of the groups of patients and controls. It is important to note that this analysis is restricted to those trials on which the subjects did in fact choose the most likely outcome, since it is only by comparing performance on such trials that we can assess subjects sensitivity to the available opportunities to earn reward represented by different ratios.

17 Cognition in frontotemporal dementia 1485 Fig. 5 Decision-making (quality) task. Percentage choice of the most likely outcome as a function of red and blue boxes (see text for details). (A) Individual characteristics in the form of a bar chart showing data for each of the patients and the average performance of the controls. (B) Group characteristics. Error bars represent 1 SEM. Solid circles 5 fvftd; open circles 5 controls. Fig. 6 Decision-making (risk adjustment) task. Percentage points risked in order to earn more points as a function of red and blue boxes (see text for details). (A) Individual characteristics in the form of a bar chart showing data for each of the patients and the average performance of the controls. (B) Group characteristics. Error bars represent 1 SEM. Solid circles 5 fvftd; open circles 5 controls. Irrespective of group, subjects increased the percentage bet as a function of the ratio of red to blue boxes at a relatively constant rate (ratio 6 : 4, 48.3%; ratio 7 : 3, 56.9%; ratio 8 : 2, 65.0%; ratio 9 : 1, 71.4%). The patients with fvftd bet more points at each ratio than the controls (patients: ratio 6 : 4, 60.3%; ratio 7 : 3, 69.0%; ratio 8 : 2, 73.0%; ratio 9 : 1, 77.6%; controls: ratio 6: %; ratio 7 : 3, 44.9%; ratio 8 : 2, 56.9%; ratio 9 : 1, 65.2%). There was therefore a highly significant group difference [F(1,12) , P ]. All subjects placed larger bets in the descending compared with the ascending condition [F(1,12) , P ]. In order to determine if the order of ascending and descending conditions affected the outcome, for example as a consequence of impulsive choices in the fvftd group, we examined the group 3 condition interaction. This was not, in fact, significant [F(1,12), 0.001, P ]. There was also no significant group 3 probability interaction [F(2,36) , P ]. Analysis of the performance of the group of fvftd patients and the controls revealed that neither age nor premorbid verbal IQ was a significant covariate of the bets (age: t , P ; verbal IQ: t , P ), and the results above were unaffected. Discussion In this study we examined the cognitive performance of patients in the relatively mild stages of fvftd on a broad

18 1486 S. Rahman et al. range of computerized tests of executive and visuospatial function that have not previously been used in the assessment of fvftd patients. As predicted, the patients with fvftd were selectively impaired in the decision-making paradigm and there was some evidence that there was an impairment in the reversal stages of the attentional set-shifting paradigm. It is also of considerable interest that, in the present study, these patients were unimpaired in their performance in the spatial span and spatial working memory tasks. In addition, although an executive impairment was suggested by the difference in the groups on the mean latency measure of the one-touch Tower of London task, the accuracy of planning performance assessed using the same task was found to be relatively unimpaired. Deficits of working memory have previously been implicated in disorders affecting the prefrontal cortex (Goldman-Rakic, 1987); furthermore, deficits of planning, demonstrated using the Tower of London task, have been found to be consequent upon disorders affecting the frontal lobes (Shallice, 1982). The performance of the fvftd patients reported here differs specifically from that of unilateral or bilateral frontal excision patients (Owen et al., 1990, 1995) in a number of key areas. The frontal-lesioned patients exhibited significant impairments in error scores on the spatial working memory task and in their use of strategy (Owen et al., 1990). These patients were also impaired on the onetouch Tower of London task in the accuracy of performance (Owen et al., 1995). The working memory and planning deficits were not considered to be due to an impairment in retaining a sequence of spatial moves in short-term memory for a sufficient length of time, as the patients had an unimpaired spatial span. In considering the precise nature of cognitive deficits demonstrated by patients with mild fvftd, it is certainly useful to note that dissociations of function within areas of the prefrontal cortex have been reported previously, and may shed some light on the specific deficits that were observed in the patients with fvftd in this study. Recently, Bechara and colleagues (Bechara et al., 1998) have reported a dissociation in performance on neuropsychological assessment in patients with frontal lobe lesions. In their study, all subjects with ventromedial lesions were impaired on their gambling task, whereas only a subset of these subjects those with the most anteriorly placed lesions were normal on the working memory tasks; however, subjects with right dorsolateral/high mesial lesions were impaired on the working memory tasks but not on the gambling task. An explanation for why the patients with fvftd in this study, with presumed frontal lobe pathology, are relatively unimpaired in some traditional frontal tasks (spatial working memory and planning assessed using the one-touch Tower of London) may be derived from what is known about the precise neural substrates underlying successful performance in these tasks, together with consideration of the possible site of pathology in fvftd. First, in recent PET studies Owen et al. (1996) have investigated components of working memory using a close analogue of the spatial working memory task described in the present study. These studies indicated two major areas of activation in the dorsolateral and ventrolateral prefrontal cortex. Further data suggest that the ventrolateral area (Brodmann area 47) may play a role in the passive receipt of items for memory (as in spatial span performance), whereas the dorsolateral prefrontal cortex (Brodmann areas 46 and 9) is important for the monitoring role by which candidate sequences are compared with the goal sequence. Secondly, in an investigation by Baker and colleagues (Baker et al., 1996) using the same Tower of London task as used in this study, regional increases in cerebral blood flow were produced in a distributed cortical network that included the superior occipitoparietal cortex and three main zones in the frontal cortex: the premotor area, a band of activation including the dorsolateral prefrontal cortex bilaterally and the frontopolar cortex on the right. In the present study, one possible reason, therefore, for the relative lack of impairment in patients with fvftd in the spatial span, spatial working memory and the one-touch Tower of London tasks is that the site of pathology in fvftd involves predominantly the orbitofrontal or ventromedial part of the frontal lobes, which is not associated with altered regional cerebral blood flow in these tasks. The future use of volumetric structural MRI or other forms of neuroimaging may be able to establish, with superior spatial resolution compared with SPECT, whether the primary site of damage in mild fvftd is the orbitofrontal or the ventromedial part of the frontal lobe. However, in early disease it is far from clear whether structural changes will be evident, thus necessitating a neuropsychological approach, as adopted here. Visual discrimination learning/attentional setshifting In the present study, we report evidence for an impairment in the attentional set-shifting (intradimensional/ extradimensional) paradigm in patients with fvftd that is relatively specific to the reversal stages demonstrated using only eight patients. Notably, the fvftd patients were unimpaired compared with controls with respect to errors at the extradimensional shift stage, a stage which had previously been shown to pose particular difficulties for patients with frontal lobe excisions (Owen et al., 1991), medicated patients with severe Parkinson s disease (Owen et al., 1992) and patients in the preclinical and early clinical stages of Huntington s disease (Lawrence et al., 1998, 1996). The extradimensional shift stage is akin to the category shift in the WCST, and the lack of impairment concurs with reports of normal performance on the WCST by patients with nonprogressive frontal lobe damage associated with sparing of the dorsolateral cortex (Drewe, 1974; Stuss and Benson, 1984; Eslinger and Damasio, 1985). A recent PET imaging study using an adapted version of the attentional set-shifting paradigm showed that, compared with the intradimensional

19 Cognition in frontotemporal dementia 1487 shift scans, the extradimensional shift activated the frontal pole on the left side and areas 9/46 on the right side, whereas reversal learning engaged neural circuitry associated with the ventromedial prefrontal cortex (Rogers et al., 1999b). In relation to this and the non-human primate work by Dias and colleagues (Dias et al., 1996, 1997), the specific reversal learning deficit of patients with fvftd early in the course of the disease is consistent with the hypothesis that the pathology predominantly affects the orbitofrontal cortex or its connections rather than the dorsolateral, prefrontal cortex. The orbitofrontal cortex is considered to be especially involved with stimulus reward learning, and may participate in a distributed network comprising the temporal cortex, amygdala, ventral striatum and mediodorsal thalamus. Reasons for the impairment in the reversal stages may include a loss of inhibitory control in response to the previously rewarded response, a failure to learn new stimulus reinforcement associations, or an interaction between the two. A classic study (Jones and Mishkin, 1972) showed that, for both object reversal and spatial reversal, lesions of the orbitofrontal cortex caused the animals to perseverate in the formerly relevant response. Once they had overcome this tendency they were able to learn the new stimulus reward association. The study by Dias and colleagues (Dias et al., 1996) also emphasized the loss of inhibitory control following orbitofrontal lesions in affective shifting. The low number of perseverative errors made by the fvftd patients in this study in the reversal stages suggests that the increased number of errors in the reversal stages made by the patients may arise from deficits in the formation of new stimulus reinforcement associations. In relation to this, it is interesting to note that damage to the orbitofrontal cortex has been found to produce disturbances in learning stimulus reinforcement associations; in a study by Iversen and Mishkin (Iversen and Mishkin, 1970), medial orbital-lesioned monkeys took many more trials to learn to respond to the previously unrewarded stimulus than controls, but were no slower than controls to inhibit their responding to the previously rewarded stimulus. This certainly concurs with more recent studies that demonstrate that lesions of the medial orbital region do disrupt the ability of monkeys to learn stimulus reward associations (Baylis and Gaffan, 1991), and therefore impairments in learning about the affective properties of stimuli may be dissociable from deficits in inhibitory control, and may themselves contribute to some of the emotional impairments associated with damage to the orbitofrontal cortex. This concept is supported by examination of the function of the orbitofrontal cortex at the neuronal level. There is major visual input to many neurons in the orbitofrontal cortex, and many neurons represent the reinforcement association of visual stimuli (Rolls et al., 1996a). In one study in which the primary reinforcer used was taste, many neurons were found to show visual taste reversal in one trial or a very few trials (Thorpe et al., 1983). In addition to these neurons, which encode the reward association of visual stimuli, other neurons in the orbitofrontal cortex were found that detected non-reward. These responded, for example, when an expected reward was not obtained when a visual discrimination task was reversed. Therefore, it would seem that the orbitofrontal cortex does enable the rapid reversal of behaviour by stimulus reinforcement association relearning when the association of stimuli with reinforcers is altered. The results presented here appear consistent with the inhibition hypothesis proposed by Plaisted and Sahakian (Plaisted and Sahakian, 1997), as the inability to form stimulus reinforcement associations when contingencies change would be sufficient to prevent these patients from responding to future social situations and to be able to form coherent action sequences appropriate to long-term goals. It might appear that the results are also broadly consistent with the position of Rolls and colleagues (Rolls et al., 1994) in that social emotional changes may be due to a failure to alter behaviour as a consequence of difficulties in the remapping of stimulus reinforcement associations. It is, however, very clear that patients with severe reversal deficits do not necessarily exhibit clinically disinhibited behaviour. At the single-case level, for example, case G scored only three errors in total in the reversal stages, but did demonstrate clinically marked behavioural disinhibition during cognitive testing. There is converging evidence that the deficits in reversal learning may be related to abnormalities in the functioning of serotonergic neurotransmitter systems in the prefrontal cortex. Normal subjects with tryptophan depletion induced by a low-tryptophan drink, when required to perform a simple rule-reversal in the presence of an irrelevant dimension (the reversal of compound discrimination stage), needed more trials to learn this rule compared with subjects receiving placebo (Park et al., 1994). This is similar to the performance of the fvftd patients in this study using the same paradigm. Low tryptophan intake is known to reduce the supply of tryptophan to the brain by lowering the plasma tryptophan level and by restricting the entry of plasma tryptophan into the brain through competition from other amino acids for transport across the blood brain barrier (Young et al., 1985). Nomura (Nomura, 1992) has reported an analogous effect in rodents: rats given a low-tryptophan diet showed impaired learning in an operant-type discrimination learning paradigm. There is currently some neurochemical evidence, derived from post-mortem studies, to suggest that there exist profound postsynaptic serotonergic defects in patients with FTD in comparison with controls (Sparks and Markesbery, 1991). It is noteworthy that Sparks and Markesbery (Sparks and Markesbery, 1991) reported in their study reduced total serotonin receptor binding to samples of tissue from autopsy-proven cases of FTD from the frontal pole (Brodmann areas 10 and 11), temporal pole (Brodmann area 38) and the hypothalamus, but not in the nucleus basalis of Meynert, in comparison with controls. It is nonetheless important to consider that the deficits in

20 1488 S. Rahman et al. reversal learning may also reflect pathology affecting the caudate nucleus. Divac and colleagues (Divac et al., 1967) originally showed in primates that a lesion involving the ventrocaudal neostriatum impaired visual discrimination reversal. The hypothesis has therefore emerged that the ventrocaudal striatum is a critical link in the stimulus response or habit-learning circuit (Mishkin et al., 1984). Neurones which reflect the responses of orbitofrontal neurons are found in the ventral head of the caudate nucleus and the ventral striatum, which receive input from the orbitofrontal cortex (Williams et al., 1993). The existence of severe degeneration of the caudate in frontotemporal degenerative dementia has long been recognized (e.g. Filley et al., 1994); the possibility of a minor degree of atrophy in the caudate nucleus in some of the patients who took part in this study cannot be completely excluded. Decision-making Previously, Damasio (Damasio, 1996) has argued that patients with ventromedial lesions in his paradigm perform badly because they may have difficulties in their ability to sense, overtly or not, which decks are risky and which are profitable. That paradigm has not, however, been able to distinguish between the principal factors determining their risk-taking behaviour. Using a different decision-making paradigm, we have shown that the patients with fvftd were different from age- and IQ-matched controls in that they were willing to bet a much higher proportion of their accumulated reward at all ratios. Furthermore, they showed no significant difference from controls in their tendency to choose the most likely outcome. Therefore, to summarize, although patients with fvftd may make accurate probability judgements, they are unable to adjust the levels of their bets accordingly. As they are not merely impulsive in making their decisions, they appear to be true risk-takers. There are broad similarities in the performance of the patients with fvftd and the patients with ventromedial lesions studied by Damasio and colleagues (Bechara et al., 1994, 1998). Both groups of patients make abnormal decisions that are no longer personally advantageous, and in particular have difficulty planning their future over immediate, medium and long ranges. Furthermore, Bechara (Bechara et al., 1998) has argued that failure of inhibitory control is not a primary reason for the poor performance of subjects with ventromedial lesions on their gambling task. This argument was based partly on the observation that, in their task, subjects switch decks whenever they receive punishment, just as normal controls do, but they return more often to the decks that yield high immediate reward (the bad decks). The switch away from the bad deck immediately after punishment did not, arguably, indicate a lack of inhibition of the natural tendency to switch decks after a negative outcome (Bechara et al., 1994). There is little evidence that defective inhibitory control is the primary reason for the poor performance of the fvftd patients in the decision-making task presented here, as the patients exhibited much longer deliberation times, did not consistently choose early bets in both the ascending and the descending sequence, and generally appeared able to adjust their bet successfully as a function of the ratio of coloured boxes (and hence the information encoded by them about the likelihood of a rewarded response), albeit at a rather inflated level. In general, neither the group described by Damasio nor our patient group were able to adjust their behaviour successfully appropriately to the opportunities offered by the task. One explanation that has been proposed for the poor performance in the gambling task of Damasio s patients with ventromedial lesions is that there is a failure to anticipate future outcomes (Bechara et al., 1997). Indeed, there is some evidence from a recent electrophysiological study to support the hypothesis that the orbitofrontal cortex is involved in encoding information that is used to guide goal-directed behaviour (Schoenbaum et al., 1998). In that study, the activity of neurons in the rat orbitofrontal cortex and basolateral amygdala was recorded during instrumental learning in an olfactory discrimination task. Neurones in both these regions fired selectively during the anticipation of rewarding or aversive outcomes; this emerged early in training before the rats had learned reliably to avoid the aversive outcome. Schoenbaum and colleagues speculated that expectations about the occurrence of future events might be useful in forming associations that represent accurate predictive information, and that experience was required to determine whether anticipated outcomes were matched by actual outcomes. Within this conceptual framework, one possible reason for the poor performance of the patients with fvftd in the decision-making task is that there exists a failure to anticipate expected outcomes, as may be the case for patients with ventromedial lesions, as a consequence of the pathology affecting the ventromedial (or orbitofrontal) cortex. In their performance in the decision-making task, it is particularly noteworthy that the patients with fvftd are strikingly similar to patients with orbitofrontal prefrontal cortex lesions, in that both groups demonstrate much higher deliberation times associated with their decisions (Rogers et al., 1999a). This behavioural difference is manifest in situations requiring choice between competing courses of action or response. It has been noted previously that similar increases in deliberation times in the decision-making task may be produced in those normal volunteers with acutely reduced tryptophan and reduced 5-hydroxytryptamine function who demonstrate the most severe deficits in decisionmaking. Therefore, in addition to the reversal learning deficits following a low-tryptophan drink reported by Park and colleagues (Park et al., 1994) and described above, as well as neuropathological post-mortem observations (e.g. Sparks and Markesbery, 1991), the results from the decision-making task provide some support for the hypothesis that abnormalities in the serotonergic system contribute to the neuropsychological profile of patients with fvftd.

21 Cognition in frontotemporal dementia 1489 Putative neural substrates underlying the cognitive profile in fvftd One important consideration in concluding that patients with mild fvftd demonstrate a relative lack of deficits on tests sensitive to lateral prefrontal function is that these tests may exhibit different sensitivity to those used to examine ventromedial or orbitofrontal function. Whilst this may be a concern, we feel that it is unlikely to be the case for a number of reasons. First, the one-touch Tower of London was shown to be very sensitive to the effects of frontal lobe lesions (Owen et al., 1995). Power analysis of those data suggests that it would have been possible to obtain significant deficits in that study using the same number of patients as used here. Secondly, the spatial working memory task and the Tower of London tasks have previously been found to be sensitive to cognitive deficits in patients with Huntington s disease (Lange et al., 1995; Lawrence et al., 1998, 1996), medicated patients with Parkinson s disease (Owen et al., 1992) and patients with multiple system atrophy and Steele Richardson Olsewski syndrome (Robbins et al., 1994b). Thirdly, recent results from our laboratory (L. H. A. Watkins, R. D. Rogers, B. J. Sahakian, A. E. Rosser and T. W. Robbins, unpublished observations) have demonstrated that patients with Huntington s disease, whilst impaired on the one-touch Tower of London task, are unimpaired on the decisionmaking task. When taken together with results from the present study, such a double dissociation arguably provides strong evidence against a difference in task sensitivity (Kinsbourne, 1971). Indeed, the deficits seen in different patient groups are consistent with what is known about the neural substrates underlying successful performance in these tasks in normal healthy volunteers discussed earlier (Baker et al., 1996; Owen et al., 1996). Finally, in relation to the performance of patients with fvftd on the attentional setshifting task, as the extradimensional shift is known to be more difficult than reversal shifting for normal subjects (Roberts et al., 1988), it is unlikely that the deficit in reversal shifting described here (or indeed the aforementioned deficit in extradimensional shifting in other patient groups) is simply due to differences in task sensitivity. Overall, the finding that patients with mild fvftd demonstrate deficits in tasks sensitive to orbitofrontal or ventromedial function, rather than function of more lateral areas, may be consistent with a number of other different converging lines of evidence. In their formulation of the neuropathological criteria for FTD, the Lund and Manchester groups (Lund and Manchester Groups, 1994) described the distribution of gross changes as typically being predominant in the frontal and anterior temporal lobes, adding that, in two of the three subtypes (frontal lobe degeneration and motor neuron disease type), there is usually no gross atrophy of the striatum, amygdala or hippocampus, in contrast to Pick disease type, where asymmetry and striatal atrophy appear to be common. At the microscopic level, they noted that changes in FTD were most notable in the frontal convexity cortex, sometimes within the orbitofrontal cortex. Various studies have recognized that the anterior and medial temporal areas and the orbitofrontal cortex show the most severe atrophic changes (Sanders et al., 1939; Cummings and Duchen, 1981; Frisoni et al., 1996); post-mortem studies of single cases have often revealed focal atrophy within the orbitofrontal region of the frontal lobe and medial temporal structures (see, for example, a case reported by Hodges and Gurd, 1994). The neuropsychological findings described in this study are further supported by neuroimaging findings regarding the specificity of changes in cerebral blood flow in patients with frontal lobe dementia. In one HMPAO-SPECT study (Starkstein et al., 1994), blood flow was found to be significantly lower in the frontal lobes, the anterior temporal cortex and the basal ganglia in patients with frontal lobe dementia compared with patients with dementia of the Alzheimer type. Within the frontal lobe dementia group, blood flow was significantly lower in the orbital than in the dorsal frontal cortex, and in the anterior temporal than in the dorsal temporal cortex. These findings concur with those from other neuroimaging studies using PET. For example, Kumar and colleagues (Kumar et al., 1990) demonstrated metabolic deficits bilaterally in the orbitofrontal, the anterior cingulate and the temporal lobes with relative sparing of the parietal lobes in three patients with marked behavioural changes including sexual disinhibition, irritability, laughing bursts and loud speech. It is possible that a number of the various clinical symptoms of patients with fvftd may be the consequence of a specific disruption of the frontostriatal circuitry involving the ventromedial (or orbitofrontal) prefrontal cortex (for a review, see Masterman and Cummings, 1997). An important aspect of future research will be to specify the relative contributions of areas other than the prefrontal cortex with which the prefrontal cortex may indeed share connectivity to the development of cognitive and behavioural symptomatologies as demonstrated in this study. It is known that in a few cases focal lesions to nonfrontal areas may produce a frontal-like syndrome (e.g. Strub, 1989; Petty et al., 1996). As expected for patients with pathology localized predominantly to the anterior cortical regions, the patients with fvftd in this study were unimpaired in accuracy measures on tests of mnemonic function (pattern and spatial recognition memory). Performance on the pattern-recognition task has been shown to be impaired in patients with temporal lobe excisions and amygdalohippocampectomy, whilst the spatial recognition task is sensitive to frontal lobe excisions (Owen et al., 1995). The recognition tasks are also sensitive to the cognitive deficits seen in probable dementia of the Alzheimer type, mild and moderate patients being equivalently impaired on pattern and spatial recognition (Sahakian et al., 1988). The increased latencies on these tasks in our patient group may indicate some impaired ability to identify and retrieve the appropriate memory trace. These results are consistent with the general lack of mnemonic

22 1490 S. Rahman et al. deficits in such background neuropsychological tests as the Logical Memory subtest used to detect posterior cortical involvement (Table 2). These patients with mild fvftd were also found to have relatively preserved language and visuospatial function in background neuropsychological assessment, consistent with previous findings in the literature (e.g. Gustafson et al., 1992). Finally, it is interesting to note that, out of the eight patients included in this study, only one had an MMSE score (22) below the cut-off score for dementia (24). This supports the view that this tool is generally inadequate for the screening of fvftd (e.g. Gregory and Hodges, 1996). Conclusion We have found that patients with fvftd in the relatively mild stages of disease demonstrate cognitive deficits not usually identified using traditional neuropsychological testing. Specifically, these patients exhibited more marked impairments in the reversal stages of the attentional setshifting paradigm and also in risk-taking behaviour in the decision-making paradigm. These neuropsychological deficits may be interpreted as a consequence of pathology affecting a distinct part of the prefrontal cortex (the orbitofrontal or ventromedial cortex), and appear to be consistent with current evidence from functional neuroimaging and post-mortem neuropathological studies. The understanding of this condition is of particular relevance to the development of better methods of early diagnosis and assessment of efficacious therapeutic interventions. It may also, in the longer term, be helpful for the development of a strategy for the rehabilitation of patients with fvftd. Crucial to this, there is a clear clinical need for the development of further neuropsychological tasks to provide sensitive measures of orbitofrontal or ventromedial prefrontal lobe dysfunction. Acknowledgements The authors wish to thank two anonymous referees for their helpful comments on a previous version of the manuscript, M. A. Mehta for statistical advice, S. Davies and N. Graham for assistance in compiling Table 2, and all those who took part in this study. This work was supported by a Programme Grant from the Wellcome Trust to T.W.R., B.J.S., Professor B. J. Everitt and Dr A. C. Roberts, a MRC Programme Grant to J.R.H., a MRC LINK grant to B.J.S., J.R.H., Dr J. Semple and T.W.R. and an MRC studentship to S.R. References Alexander GE, Crutcher MD. Functional architecture of basal ganglia circuits: neural substrates of parallel processing [see comments]. [Review]. Trends Neurosci 1990; 13: Comment in: Trends Neurosci 1991; 14: Alexander GE, Delong MR, Strick PL. Parallel organization of functionally segregated circuits linking basal ganglia and cortex. [Review]. Annu Rev Neurosci 1986; 9: Anderson SW, Damasio H, Jones RD, Tranel D. Wisconsin Card Sorting Test performance as a measure of frontal lobe damage. J Clin Exp Neuropsychol 1991; 13: Baker SC, Rogers RD, Owen AM, Frith CD, Dolan RJ, Frackowiak RS, et al. Neural systems engaged by planning: a PET study of the Tower of London task. Neuropsychologia 1996; 34; Barbas H, Pandya DN. Pattern of connections in the prefrontal cortex in the rhesus monkey associated with cortical architecture. In: Levin HS, Eisenberg HM, Benton AL, editors. Frontal lobe function and dysfunction. New York: Oxford University Press; p Baylis LL, Gaffan D. Amygdalectomy and ventromedial prefrontal ablation produce similar deficits on food choice and in simple object discrimination learning for an unseen reward. Exp Brain Res 1991; 86: Bechara A, Damasio AR, Damasio H, Anderson SW. Insensitivity to future consequences following damage to human prefrontal cortex. Cognition 1994; 50: Bechara A, Damasio H, Tranel D, Damasio AR. Deciding advantageously before knowing the advantageous strategy [see comments]. Science 1997; 275: Comment in: Science 1997; 275: Bechara A, Damasio H, Tranel D, Anderson SW. Dissociation of working memory from decision making within the human prefrontal cortex. J Neurosci 1998; 18: Bishop DVM. Test for the reception of grammar. Manchester, UK: University of Manchester; Brun A. Frontal lobe degeneration of non-alzheimer type. I. Neuropathology. Arch Gerontol Geriatr 1987; 6: Brun A. Frontal lobe degeneration of non-alzheimer type revisited. Dementia 1993; 4: Corcoran, R, Upton, D. A role for the hippocampus in card sorting? [see comments]. Cortex 1993; 29: Comment in: Cortex 1996; 32: Cummings JL, Duchen LW. The Klüver-Bucy syndrome in Pick s disease: clinical and pathologic correlations. Neurology 1981; 31: Damasio AR. Descartes error: emotion, reason and the human brain. London: Picador; Damasio, AR. The somatic marker hypothesis and the possible functions of the prefrontal cortex. [Review]. Philos Trans R Soc Lond B Biol Sci 1996; 351: Dias R, Robbins TW, Roberts AC. Dissociation in prefrontal cortex of affective and emotional shifts. Nature 1996; 380: Dias R, Robbins TW, Roberts AC. Dissociable forms of inhibitory control within prefrontal cortex with an analog of the Wisconsin Card Sort Test: restriction to novel situations and independence from on-line processing. J Neurosci 1997; 17:

23 Cognition in frontotemporal dementia 1491 Divac I, Rosvold HE, Szwarcbart MK. Behavioral effects of selective ablation of the caudate nucleus. J Comp Physiol Psychol 1967; 63: Downes JJ, Roberts AC, Sahakian BJ, Evenden JL, Robbins RG, Robbins TW. Impaired extra-dimensional shift performance in medicated and unmedicated Parkinson s disease: evidence for a specific attentional dysfunction. Neuropsychologia 1989; 27: Drewe EA. The effect of type and area of brain lesion on the Wisconsin Card Sorting Test performance. Cortex 1974; 10: Edwards-Lee T, Miller BL, Benson DF, Cummings JL, Russell GL, Boone K, et al. The temporal variant of frontotemporal dementia. Brain 1997; 120: Elfgren C, Passant U, Risberg J. Neuropsychological findings in frontal lobe dementia. Dementia 1993; 4: Elfgren C, Brun A, Gustafson L, Johansen A, Minthon L, Passant U, et al. Neuropsychological tests as discriminators between dementia of Alzheimer type and frontotemporal dementia. Int J Geriatr Psychiatry 1994; 9: Eslinger PJ, Damasio AR. Severe disturbance of higher cognition after bilateral frontal lobe ablation: patient EVR. Neurology 1985; 35: Filley CM, Kleinschmidt-De Masters BK, Gross KF. Non-Alzheimer fronto-temporal degenerative dementia. A neurobehavioral and pathologic study. Clin Neuropathol 1994; 13: Folstein MF, Folstein SE, McHugh PR. Mini-mental state : a practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975; 12: Frisoni GB, Pizzolato G, Geroldi C, Rossato A, Bianchetti A, Trabucchi M. Dementia of the frontal type: neuropsychological and [99Tc]-HM-PAO SPECT features. J Geriatr Psychiatry Neurol 1995; 8: Frisoni GB, Beltramello A, Geroldi C, Weiss C, Bianchetti A, Trabucchi M. Brain atrophy in frontotemporal dementia. J Neurol Neurosurg Psychiatry 1996; 61: Goldman-Rakic PS. Circuitry of primate prefrontal cortex and regulation of behaviour by representational memory. In: Plum F, editor. Handbook of physiology, Sect 1, Vol. 5, Pt. 1. Bethesda (MD): American Physiological Society; p Gregory CA, Hodges JR. Dementia of frontal type and the focal lobar atrophies. Int J Geriatr Psychiatry 1993; 5: Gregory CA, Hodges JR. Clinical features of frontal lobe dementia in comparison to Alzheimer s disease. [Review]. J Neural Transm Suppl 1996; 47: Gregory CA, Orrell M, Sahakian BJ, Hodges JR. Can frontotemporal dementia and Alzheimer s disease be differentiated using a brief battery of tests? Int J Geriatr Psychiatry 1997; 12: Gregory CA, McKenna PJ, Hodges JR. Dementia of frontal type and simple schizophrenia: two sides of the same coin? Neurocase 1998; 4: 1 6. Gregory CA, Serra-Mestres J, Hodges JR. The early diagnosis of the frontal variant of frontotemporal dementia: how sensitive are standard neuroimaging and neuropsychological tests? J Neuropsychiatry Neuropsychol Behav Neurol 1999; 12: Gustafson L. Clinical picture of frontal lobe degeneration of non- Alzheimer type. Dementia 1993; 4: Gustafson L, Risberg J. Deceptions and delusions in Alzheimer s disease and frontal lobe dementia. In: Katona C, Levy R, editors. Delusions and hallucinations of old age. London: Gaskell; Gustafson L, Brun A, Passant U. Frontal lobe degeneration of non- Alzheimer type. [Review]. Bailliere s Clin Neurol 1992; 1: Hodges JR, Gurd JM. Remote memory and lexical retrieval in a case of frontal Pick s disease [see comments]. Arch Neurol 1994; 51: Comment in: Arch Neurol 1995; 52: Hodges JR, Patterson K. Nonfluent progressive aphasia and semantic dementia: a comparative neuropsychological study. J Int Neuropsychol Soc 1996; 2: Hodges JR, Patterson K, Oxbury S, Funnell E. Semantic dementia: progressive fluent aphasia with temporal lobe atrophy. Brain 1992; 115: Hodges JR, Patterson K, Ward R, Garrard P, Bak T, Perry R, et al. The differentiation of semantic dementia and frontal lobe dementia (temporal and frontal variants of frontotemporal dementia) from early Alzheimer s disease: a comparative neuropsychological study. Neuropsychology 1999; 13: Howell DC. Statistical methods for psychology. 4th ed. Belmont (CA): Duxbury Press; Hutton M, Lendon LC, Rizzu P, Baker M, Froelich S, Houlden H, et al. Association of missense and 59-splice-site mutations in tau with the inherited dementia FTDP-17. Nature 1998; 393: Iversen SD, Mishkin M. Perseverative interference in monkeys following selective lesions of the inferior prefrontal convexity. Exp Brain Res 1970, 11: Jones B, Mishkin M. Limbic lesions and the problem of stimulus reinforcement associations. Exp Neurol 1972; 36: Joyce EM, Robbins TW. Frontal lobe function in Korsakoff and non-korsakoff alcoholics: planning and spatial working memory. Neuropsychologia 1991; 29: Kertesz A. The quantification of behavior in frontotemporal dementia. In: Kertesz A, Munoz DG, editors. Pick s disease and Pick complex. New York: Wiley-Liss; p Kinsbourne M. Cognitive deficit: experimental analysis. In: McGaugh JL, editor. Psychobiology. New York: Academic Press; p Knopman DS, Mastri AR, Frey WH 2d, Sung JH, Rustan T. Dementia lacking in distinctive histologic features: a common non- Alzheimer degenerative dementia. Neurology 1990; 40: Kumar A, Schapiro MB, Haxby JV, Grady CL, Friedland RP. Cerebral metabolic and cognitive studies in dementia with frontal lobe behavioral features. J Psychiatr Res 1990; 24:

24 1492 S. Rahman et al. Lange KW, Sahakian BJ, Quinn NP, Marsden CD, Robbins TW. Comparison of executive and visuospatial memory function in Huntington s disease and dementia of Alzheimer type matched for degree of dementia. J Neurol Neurosurg Psychiatry 1995; 58: Lawrence AD, Sahakian BJ, Hodges JR, Rosser AE, Lange KW, Robbins TW. Executive and mnemonic functions in early Huntington s disease. Brain 1996, 119: Lawrence AD, Hodges JR, Rosser AE, Kershaw A, ffrench-constant C, Rubinsztein DC, et al. Evidence for specific cognitive deficits in preclinical Huntington s disease. Brain 1998; 121: Mann DM. Dementia of frontal type and dementias with subcortical gliosis. [Review]. Brain Pathol 1998; 8: Lund and Manchester Groups. Clinical and neuropathological criteria for frontotemporal dementia. [Review]. J Neurol Neurosurg Psychiatry 1994; 57: Masterman DL, Cummings JL. Frontal subcortical circuits: the anatomic basis of executive, social and motivated behaviors. [Review]. J Psychopharmacol (Oxf) 1997; 11: Mesulam MM. Slowly progressive aphasia without generalised dementia. Ann Neurol 1982; 11: Miller LA. Impulsivity, risk-taking, and the ability to synthesize fragmented information after frontal lobectomy. Neuropsychologia 1992; 30: Miller BL, Cummings JL, Villanueva-Meyer J, Boone K, Mehringer CM, Lesser IM, et al. Frontal lobe degeneration: clinical, neuropsychological, and SPECT characteristics [see comments]. Neurology 1991; 41: Comment in: Neurology 1992; 42: Miller BL, Darby AL, Swartz JR, Yener GG, Mena I. Dietary changes, compulsions and sexual behaviour in frontotemporal degeneration. Dementia 1995; 6: Milner B. Effects of different brain lesions on card sorting. Arch Neurol 1963; 9: Milner, B. Interhemispheric differences in the localization of psychological processes in man. [Review]. Br Med Bull 1971; 27: Mishkin M, Malamut B, Bachevalier J. Memories and habits: two neural systems. In: Lynch G, McGaugh JL, Weinberger NM, editors. Neurobiology of learning and memory. New York: Guilford Press; p Mowrer OH. On the dual nature of learning a re-interpretation of conditioning and problem solving. Harvard Educ Rev 1947; 17: Neary D, Snowden J. Fronto-temporal dementia: nosology, neuropsychology, and neuropathology. [Review]. Brain Cogn 1996; 31: Neary D, Snowden JS, Bowen DM, Sims NR, Mann DM, Benton JS, et al. Neuropsychological syndromes in presenile dementia due to cerebral atrophy. J Neurol Neurosurg Psychiatry 1986; 49: Neary D, Snowden JS, Northen B, Goulding P. Dementia of frontal lobe type. J Neurol Neurosurg Psychiatry 1988; 51: Nelson HE. National Adult Reading Test Manual. Windsor (UK): NFER-Nelson; Nomura M. Effects of bifemelane on discrimination learning of serotonergic-dysfunction rats. Pharmacol Biochem Behav 1992; 42: Olton DS, Samuelson RJ. Remembrance of places passed: spatial memory in rats. J Exp Psychol Anim Behav Proc 1976; 2: Orrell MW, Sahakian BJ, Bergmann K. Self-neglect and frontal lobe dysfunction. Br J Psychiatry 1989; 155: Owen AM, Downes JJ, Sahakian BJ, Polkey CE, Robbins TW. Planning and spatial working memory following frontal lobe lesions in man. Neuropsychologia 1990; 28: Owen AM, Roberts AC, Polkey CE, Sahakian BJ, Robbins TW. Extra-dimensional versus intra-dimensional set shifting performance following frontal lobe excisions, temporal lobe excisions or amygdalo-hippocampectomy in man. Neuropsychologia 1991; 29: Owen AM, James M, Leigh PN, Summers BA, Marsden CD, Quinn NP, et al. Fronto-striatal cognitive deficits at different stages of Parkinson s disease. Brain 1992; 115: Owen AM, Sahakian BJ, Hodges JR, Summers BA, Polkey CE, Robbins TW. Dopamine-dependent frontostriatal planning deficits in early Parkinson s disease. Neuropsychology 1995a; 9: Owen AM, Sahakian BJ, Semple J, Polkey CE, Robbins TW. Visuospatial short-term recognition memory and learning after temporal lobe excisions, frontal lobe excisions or amygdalohippocampectomy in man. Neuropsychologia 1995b; 33; 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: Pachana NA, Boone KB, Miller BL, Cummings JL, Berman N. Comparison of neuropsychological functioning in Alzheimer s disease and frontotemporal dementia. J Int Neuropsychol Soc 1996; 2: Park SB, Coull JT, McShane RH, Young AH, Sahakian BJ, Robbins TW, et al. Tryptophan depletion in normal volunteers produces selective impairments in learning and memory. Neuropharmacology 1994; 33; Pasquier F, Lebert F, Grymonprez L, Petit H. Verbal fluency in dementia of frontal lobe type and dementia of Alzheimer type. J Neurol Neurosurg Psychiatry 1995; 58: Petty RG, Bonner D, Mouratoglou V, Silverman M. Acute frontal lobe syndrome and dyscontrol associated with bilateral caudate nucleus infarctions. Br J Psychiatry 1996; 168: Pick A. Uber die Beziehungen der senilen Hirnatrophie zur Aphasie. Prag Med Wschr 1892; 17: Plaisted KC, Sahakian BJ. Dementia of frontal type living in the here and now. Aging Ment Health 1997; 1:

25 Cognition in frontotemporal dementia 1493 Robbins TW, James M, Owen AM, Lange KW, Lees AJ, Leigh PN, et al. Cognitive deficits in progressive supranuclear palsy, Parkinson s disease, and multiple system atrophy in tests sensitive to frontal lobe dysfunction. J Neurol Neurosurg Psychiatry 1994b; 57: Roberts AC, Robbins TW, Everitt BJ. The effects of intradimensional and extradimensional shifts on visual discrimination learning in humans and non-human primates. Q J Exp Psychol 1988; 40B: Robinson AL, Heaton RK, Lehman RA, Stilson DW. The utility of the Wisconsin Card Sorting Test in detecting and localizing frontal lobe lesions. J Consult Clin Psychol 1980; 48: Rogers RD, Everitt BJ, Baldacchino A, Johnson AJ, Swainson R, London M, et al. Dissociable deficits in the decision-making cognition of chronic amphetamine abusers, opiate abusers, patients with focal damage to prefrontal cortex, and tryptophan-depleted normal volunteers: evidence for monoaminergic mechanisms. Neuropsychopharmacology 1999a; 20: Rogers RD, Andrews TC, Grasby PM, Brooks D, Robbins TW. Contrasting cortical and sub-cortical PET activations produced by reversal learning and attentional set-shifting in humans. J Cogn Neurosci. In press 1999b. Rolls ET. A neurophysiological and computational approach to the functions of the temporal lobe cortical visual areas in invariant object recognition. In: Harris L, Jenkin M, editors. Computational and biological mechanisms of visual coding. Cambridge: Cambridge University Press; 1996a. p Rolls ET. The orbitofrontal cortex. [Review]. Philos Trans R Soc Lond B Biol Sci 1996b; 351, Rolls ET, Hornak J, Wade D, McGrath J. Emotion-related learning in patients with social and emotional changes associated with frontal lobe damage. J Neurol Neurosurg Psychiatry 1994; 57: Sahakian BJ, Morris RG, Evenden JL, Heald A, Levy R, Philpot M, et al. A comparative study of visuospatial memory and learning in Alzheimer-type dementia and Parkinson s disease. Brain 1988; 111: Sanders J, Schenk VWD, Van Venn P. A family with Pick s disease. In: Uitgave NV van de, editor. Amsterdam: Noord-Hollandsche; Sarazin M, Pillon B, Giannakopoulos P, Rancurel G, Samson Y, Dubois B. Clinicometabolic dissociation of cognitive functions and social behavior in frontal lobe lesions. Neurology 1998; 51: Schoenbaum G, Chiba AA, Gallagher M. Orbitofrontal cortex and basolateral amygdala encode expected outcomes during learning. Nat Neurosci 1998; 1: Shallice T. Specific impairments of planning. Philos Trans R Soc Lond B Biol Sci 1982; 298: Snowden JS, Goulding PJ, Neary D. Semantic dementia: a form of circumscribed cerebral atrophy. Behav Neurol 1989; 2: Snowden JS, Griffiths HL, Neary D. Progressive language disorder associated with frontal lobe degeneration. Neurocase 1996a; 2: Snowden JS, Neary D, Mann DMA. Frontotemporal lobar degeneration: frontotemporal dementia, progressive aphasia, semantic dementia. New York: Churchill Livingstone; 1996b. Sparks DL, Markesbery WR. Altered serotonergic and cholinergic synaptic markers in Pick s disease. Arch Neurol 1991; 48: Sperlinger D, Furst M. Service experience of people with pre-senile dementia: a study of carers in one London borough. Int J Geriatr Psychiatry 1994; 9: Spillantini MG, Bird TD, Ghetti B. Frontotemporal dementia and parkinsonism linked to chromosome 17: a new group of tauopathies. [Review]. Brain Pathol 1998; 8: Starkstein SE, Migliorelli R, Tesón A, Sabe L, Vázquez S, Turjanski M, et al. Specificity of changes in cerebral blood flow in patients with frontal lobe dementia. J Neurol Neurosurg Psychiatry 1994; 57: Strub RL. Frontal lobe syndrome in a patient with bilateral globus pallidus lesions. Arch Neurol 1989; 46: Stuss DT. Assessment of neuropsychological dysfunction in frontal lobe degeneration. [Review]. Dementia 1993; 4: Stuss DT, Benson DF. Neuropsychological studies of the frontal lobes. Psychol Bull 1984; 95: Thorpe SJ, Rolls ET, Maddison S. The orbitofrontal cortex: neuronal activity in the behaving monkey. Exp Brain Res 1983; 49; Tukey JW. Exploratory data analysis. Reading (MA): Addison- Wesley; Venneri A, Grassi F, Caffarra P. Dementia of the frontal lobe type: report of the neuroimaging and neuropsychological results of a case study. Dementia 1996; 7: Warrington EK, James MJ. The Visual Object and Space Perception Battery (VOSP). Bury St Edmunds, UK: The Thames Valley Test Company; Wechsler DA. Wechsler Memory Scale Revised. San Antonio (TX): Psychological Corporation; Williams GV, Rolls ET, Leonard CM, Stern C. Neuronal responses in the ventral striatum of the behaving macaque. Behav Brain Res 1993; 55; Yeterian EH, Pandya DN. Laminar origin of striatal and thalamic projections of the prefrontal cortex in rhesus monkeys. Exp Brain Res 1994; 99: Young SN, Smith SE, Pihl RO, Ervin FR. Tryptophan depletion causes a rapid lowering of mood in normal males. Psychopharmacology (Berl) 1985; 87: Received September 24, Revised February 13, Accepted March 18, 1999

26 Neuropsychopharmacology (2006) 31, & 2006 Nature Publishing Group All rights reserved X/06 $ Methylphenidate ( Ritalin ) can Ameliorate Abnormal Risk-Taking Behavior in the Frontal Variant of Frontotemporal Dementia Shibley Rahman 1, Trevor W Robbins 2, John R Hodges 3,4, Mitul A Mehta 5, Peter J Nestor 3, Luke Clark 2 and Barbara J Sahakian*,1 1 Department of Psychiatry, University of Cambridge School of Clinical Medicine, Cambridge, UK; 2 Department of Experimental Psychology, University of Cambridge, Cambridge, UK; 3 Department of Neurology, University of Cambridge, Addenbrooke s Hospital, Cambridge, UK; 4 MRC Cognition and Brain Sciences Unit, Cambridge, UK; 5 Institute of Psychiatry, University of London, London, UK The frontal variant of frontotemporal dementia is a significant neurological condition worldwide. There exist few treatments available for the cognitive and behavioural sequelae of fvftd. Previous research has shown that these patients display risky decision-making, and numerous studies have now demonstrated pathology affecting the orbitofrontal cortex. The present study uses a within-subjects, doubleblind, placebo-controlled procedure to investigate the effects of a single dose of methylphenidate (40 mg) upon a range of different cognitive processes including those assessing prefrontal cortex integrity. Methylphenidate was effective in normalizing the decisionmaking behavior of patients, such that they became less risk taking on medication, although there were no significant effects on other aspects of cognitive function, including working memory, attentional set shifting, and reversal learning. Moreover, there was an absence of the normal subjective and autonomic responses to methylphenidate seen in elderly subjects. The results are discussed in terms of the somatic marker hypothesis of impaired decision-making following orbitofrontal dysfunction. Neuropsychopharmacology (2006) 31, doi: /sj.npp ; published online 7 September 2005 Keywords: orbitofrontal cortex; methylphenidate; risk-taking; decision-making; frontal variant frontotemporal dementia INTRODUCTION The frontal variant of frontotemporal dementia (fvftd) is a clinically significant neurological problem that constitutes one of the most prevalent forms of early-onset dementia (Ratnavalli et al, 2002). There currently exist few treatments available for the amelioration of the cognitive and behavioral deficits in fvftd, in stark contrast to the management options for dementia of the Alzheimer type (Rahman et al, 2000). An initial study by Coull et al (1996) showed that the a 2 noradrenergic antagonist idazoxan produced cognitive improvements in patients with fvftd on tests of planning, sustained attention, verbal fluency, and episodic memory. Idazoxan was thought to increase coerulo-cortical noradrenaline (NA) activity via presynaptic effects. There has also been some interest in using selective serotonin reuptake inhibitors (SSRIs) in the treatment of frontotemporal dementia, but we have previously found *Correspondence: Professor BJ Sahakian, Department of Psychiatry, University of Cambridge School of Clinical Medicine, Addenbrooke s Hospital, Cambridge CB2 2QQ, UK. Tel: ; Fax: Received 13 January 2005; revised 17 June 2005; accepted 20 July 2005 Online publication: 3 August 2005 at Npp /default.pdf little objective benefit of a chronic dosage regimen of paroxetine (Deakin et al, 2004a). Previous work has demonstrated that the psychostimulant methylphenidate (Ritalin) improves higher level cognitive function in healthy volunteers, on tests of working memory and planning (Elliott et al, 1997; Mehta et al, 2000). Methylphenidate increases synaptic and extracellular concentrations of dopamine (DA) and NA (Scheel-Krüger, 1971) via blockade of the reuptake transporters in the striatum (Volkow et al, 1998). There has been historical interest in methylphenidate, as it has been used for many years in the treatment of attention-deficit hyperactivity disorder (ADHD) and disorders of sleep and arousal such as narcolepsy (Aron et al, 2003; Conners and Eisenberg, 1963). The cognition-enhancing effects of methylphenidate have been investigated in patients with brain injury, with some studies demonstrating improvements (Gualtieri and Evans, 1998; Plenger et al, 1996; Whyte et al, 1997), but others failing to find beneficial effects (Speech et al, 1993; Williams et al, 1998). Methylphenidate has been studied as an adjuvant medication for the treatment of depression and apathy in various disorders (Marin et al, 1995; Chatterjee and Fahn, 2002). Analyses of the neurochemical correlates of FTD based on autopsy tissue have generally implicated deficits in DA and NA neurotransmission (Bettendorff et al,

27 ; Francis et al, 1993; Gilbert et al, 1988; Nagaoka et al, 1995; Sjogren et al, 1998; Sparks and Markesbery, 1991). For example, Sjogren et al (1998) reported a significant reduction in the CSF concentration of homovanillic acid, a major metabolite of DA, in a large sample of frontotemporal dementia cases. This indication of DA pathology in fvftd highlights the therapeutic potential of methylphenidate in this condition. We originally proposed that the orbitofrontal cortex was a major locus of aberrant function in fvftd (Rahman et al, 1999; Mendez and Cummings, 2003). Neurocognitive assessment indicated risk-taking behavior in the domain of decision-making and deficits in reversal learning, which resemble the cognitive sequelae of orbitofrontal cortex lesions (Damasio, 1996; Rolls et al, 1994). These rewardbased deficits are considered to be one of the main problems that fvftd patients face (La Coco and Nacci, 2004), in contrast to the episodic memory dysfunction found in the dementia of Alzheimer type (eg Lee et al, 2003). Neuroanatomical evidence for orbitofrontal dysfunction has been borne out in subsequent post mortem and in vivo neuroimaging studies (Ibach et al, 2004; Liu et al, 2004; Diehl et al, 2004; Kobayashi et al, 1999). For example, a European multicentre PET study showed that the ventromedial frontopolar cortex was critically affected in every one of 29 fvftd cases (Salmon et al, 2003). The ascending mesocorticolimbic DA projection into the prefrontal cortex has a critical role in signaling rewardrelated information (Hollerman and Schultz, 1998), and a fronto-striatal circuit comprising the ventral striatum and orbitofrontal cortex has been implicated in a range of emotional processes and affective disorders (Alexander et al, 1986; Mega and Cummings, 1994). We hypothesized that methylphenidate would ameliorate reward-based deficits in fvftd by stimulating dopaminergic transmission in orbitofrontal fronto-striatal circuitry. We have previously reported no significant effects on decision-making cognition in healthy elderly volunteers receiving a single 40 mg dose of methylphenidate (Turner et al, 2003). However, healthy elderly volunteers tend to be more conservative and show reduced risk-taking behavior compared to young adults (Deakin et al, 2004b), whereas fvftd patients show increased risk-taking behavior compared to age-matched controls (Rahman et al, 1999). The present study sought to explore the cognition-enhancing properties of methylphenidate in fvftd on a neuropsychological assessment focusing on measures of prefrontal cortical function, including reward-based learning and working memory, and temporal lobe memory function (visual recognition memory). METHODS Participants Eight patients were recruited with a clinical diagnosis of fvftd according to the Lund and Manchester Groups (1994) criteria. The eight patients are labelled A-H in Figure 1. A full medical history was taken prior to testing by specialist registrar in neurology (PJN). Patient B was female; the rest were male. Patients were all Caucasian, with a mean age of 62.0 years (SD 10.1). Four patients were current Neuropsychopharmacology Methylphenidate and fvftd S Rahman et al smokers (A, B, F, H). The Mini Mental State Examination (MMSE; Folstein et al, 1975) was administered with an exclusion threshold of 20, which would preclude computerized assessment (mean score 27.0, SD 1.7). Premorbid verbal intelligence was assessed with the National Adult Reading Test (mean 110.7, SD 6.2). Subjects were asked to abstain from alcohol intake on the evenings before test sessions or excessive caffeine intake on the mornings of test sessions. Possible risks and benefits were explained to all patients and caregivers before seeking informed consent for the study. The study was approved by the Cambridge Local Research Ethics Committee and the Department of Health Medicines Controls Agency. Patients were excluded if the severity of dementia precluded computerized neuropsychological assessment (MMSE o20), or if they suffered from concomitant illness likely to confound the interpretation of findings. Five patients were currently receiving medication: atenolol 25 mg/day (patient H), vitamin E tablets (patient F), a nonsteroidal anti-inflammatory drug (patient A), chlorpromazine 150 mg/day, fluoxetine 20 mg/day and diazepam 15 mg/day (patient B), and hormone replacement therapy and diclofenac (patient C). A total of 11people were originally screened for the study, of which three were excluded from participating in the study, due to concomitant illnesses, which contraindicated the use of methylphenidate (glaucoma, unstable hypertension, and stable/ unstable angina). Procedure This study followed a double-blind, placebo-controlled cross-over design. In this design, the performance of each subject is measured twice, once following administration of an active drug and once following administration of a placebo. The two test sessions for each subject were separated by 1 2 weeks and treatment order was fully counter-balanced across subjects. The groups that had taken drug first (A, D, F, H) or placebo first (B, C, E, G) were matched for age, MMSE, and NART IQ estimate (age: F(1,6) ¼ 0.215, P ¼ 0.659; MMSE: F(1,6) ¼ 0.667, P ¼ 0.445; NART: F(1,6) ¼ 2.48, P ¼ 0.166). All test sessions were in the afternoon, with ingestion of the drug or placebo at approximately 1430 hours. A 40 mg dose of methylphenidate was used, with cognitive testing commencing at 90 min after administration (for about 2 h). Peak plasma concentrations of methylphenidate are reached approximately 2 h after ingestion (half-life in plasma 1 2 h, mean systemic clearance is 10 l/h/kg) (Gilman et al, 1980; see also Turner et al (2003) and Muller et al (2005) for older healthy adults). Adverse effects during the session were monitored and recorded, with particular care to identify any nervousness, headache, and gastrointestinal disturbances. Blood pressure was measured using a traditional sphygmomanometer immediately prior to ingestion of the tablet, and just prior to cognitive testing (at + 90 min). Visual analog scales (Bond and Lader, 1974) were administered prior to tablet ingestion and just prior to cognitive testing, for the scales alert drowsy, calm excited, strong feeble, muzzy clear headed, well coordinated clumsy, lethargic energetic, contented discontented, troubled tranquil, mentally slow quick witted, tense relaxed, attentive dreamy, incompetent proficient, happy sad, antagonistic

28 amicable, interested bored and withdrawn-gregarious. A restricted threshold of P ¼ 0.01 was set to represent a significant difference between scores in order to reduce the chance of a type I error. Cognitive Assessment In both test sessions, the patients were given the same assessment in a fixed order, consisting of a computerized assessment including established tasks from the CANTAB battery ( pattern recognition memory, spatial recognition memory, spatial span, spatial working memory, and ID/ED attentional set-shifting (see Rahman et al, 1999 for descriptions). The one-touch version of the CANTAB Tower of London test of spatial planning was also administered (Owen et al, 1995). Finally, the Cambridge Gamble Task (Rogers et al, 1999) was used to assess decision-making cognition. In this task, subjects are presented with an array of 10 red or blue boxes on each trial. The ratio of red : blue boxes varies across trials in a pseudo-random sequence (6 : 4, 7 : 3, 8 : 2, 9 : 1). The subject is informed that the computer has hidden a token, at random, under one of the boxes, and they must decide whether the token is hidden under a red or a blue box. After making this probabilistic judgment (they should always select the color in the majority), subjects must place a bet on their confidence in the decision. Bets are generated by the computer in either an ascending or descending sequence, with the available bets in percentages of the current points total (5, 25, 50, 75, and 95%). After the bet is placed, the location of the token is revealed and the points bet are either added or deducted from the running total. Subjects perform four blocks of 10 trials in each condition (ascending or descending bets), where at the start of each block the subject returns to 100 points. Statistical Analysis Statistical analyses were run in SPSS version 9.0 with two-tailed statistics thresholded at Po0.05. In the crossover design, it is the difference between the two test sessions that denotes the effect of the treatment. Given the small group sizes, data were analyzed using nonparametric statistics. Drug effects were tested by comparing the methylphenidate and placebo scores within-subjects using a Wilcoxon signed-ranks test (as described in Howell, 1997). This test is a distribution-free analog for related samples. RESULTS Cognitive Assessment Neuropsychological performance is shown in Table 1. There were no effects of methylphenidate on pattern recognition memory, spatial recognition memory, spatial working memory, ID/ED attentional set-shifting, and onetouch Tower of London tasks. A marginally significant (detrimental) effect on spatial span on the span score (P ¼ 0.096) was not apparent on usage errors on this task (P ¼ 0.612). Methylphenidate and fvftd S Rahman et al Table 1 Neuropsychological Performance of fvftd Patients after Methylphenidate (MPH) and Placebo (Mean (SD)) Pattern recognition memory MPH Placebo Wilcoxon z-value Percentage correct 17.9 (2.2) 19.4 (3.3) Z ¼ 1.28, P ¼ Latency (ms) 3270 (674) 3490 (1480) Z ¼ 0.560, P ¼ Spatial recognition memory Percentage correct 14.0 (3.0) 13.6 (3.5) Z ¼ 0.736, P ¼ Latency (ms) 3230 (908) 3570 (346) Z ¼ 0.280, P ¼ Spatial span Span 4.25 (0.71) 4.88 (0.64) Z ¼ 1.67, P ¼ Errors 2.52 (0.70) 2.72 (1.25) Z ¼ 0.507, P ¼ Spatial working memory Total between-search errors 52.6 (21.6) 48.2 (26.9) Z ¼ 0.840, P ¼ Strategy score 37.9 (3.6) 36.5 (3.3) Z ¼ 0.916, P ¼ ID/ED attentional set shifting ID stage errors 1.63 (3.81) 1.50 (2.27) ED stage errors 10.8 (12.0) 9.88 (10.0) Reversal errors 14.8 (16.8) 14.2 (14.9) Nonreversal errors 7.17 (5.27) 10.3 (7.92) One-touch tower of London Mean attempts (2 moves) 1.36 (0.748) 1.29 (0.756) Mean attempts (3 moves) 1.29 (0.393) 1.21 (0.567) Mean attempts (4 moves) 1.54 (0.366) 1.85 (0.627) a Z ¼ 0.314, P ¼ b Z ¼ 0.105, P ¼ Mean attempts (5 moves) 2.21 (0.940) 2.14 (0.643) c Z ¼ 0.406, P ¼ Mean latency (2 moves) (s) 13.7 (8.7) 15.9 (20.0) Mean latency (3 moves) (s) 20.9 (17.2) 30.6 (38.4) Mean latency (4 moves) (s) 43.5 (27.0) 42.7 (26.5) Mean latency (5 moves) (s) 59.4 (34.4) 70.1 (47.0) d Z ¼ 0.338, P ¼ ID ¼ intradimensional shift, ED ¼ extradimensional shift. a The difference between ED errors and ID errors was tested. b The difference between reversal and nonreversal errors was tested. c The average mean attempts across all levels of difficulty was tested. d The average latency across all levels of difficulty was tested. Data for the Cambridge Gamble Task are shown in Figures 1 and 2. There were no effects of methylphenidate on the average deliberation time (Z ¼ 0.280, P ¼ 0.779) or choice of the likely outcome (Z ¼ 0.170, P ¼ 0.865) (see Figure 2). The difference in betting behavior (Figure 1) in the drug and placebo conditions was statistically significant (Z ¼ 2.38, P ¼ 0.017), with patients demonstrating reduced betting on methylphenidate. This was a consistent effect across all eight patients. To further evaluate the changes in betting behavior on methylphenidate, a two (drug: methylphenidate, placebo) 2 (condition: ascend, descend) 4 (ratio: 9 : 1, 8 : 2, 7 : 3, 6 : 4) 2 (order of drug administration) mixed-model ANOVA was performed. There was a significant main effect of drug (F 1,6 ¼ 7.73, P ¼ 0.032), Neuropsychopharmacology 653

29 654 a % bets b % bets c 100 % bets :1 8:2 7:3 6:4 Ratio of boxes Placebo Drug Healthy* Placebo Drug Drug Placebo A B C D E F G H Patient Figure 1 (a) Methylphenidate significantly reduced betting behavior on the Cambridge Gamble Task. Error bars indicate the standard errors of the mean. (b) The reduction in betting behavior in fvftd patients on methylphenidate normalizes task performance, that is, performance approaches of healthy older adults. *Healthy data from the placebo condition of Turner et al (2003), in 20 male volunteers aged years (mean age 61 years). (c) Individual data in the fvftd group show that the reduction in betting is apparent in every subject. supporting the effect in the nonparametric test. The main effects of order (F 1,6 ¼ 0.570, P ¼ 0.479) and drug order interaction term (F 1,6 ¼ 1.37, P ¼ 0.287) were both nonsignificant. The main effect of condition (F 1,6 ¼ 1.60, P ¼ 0.253) was not significant, although, on average, subjects did place higher bets in the descend condition than in the ascend condition (methylphenidate: ascend Neuropsychopharmacology Methylphenidate and fvftd S Rahman et al a10000 Deliberation time (msec) b Proportion choices of likely outcome :1 8:2 7:3 6:4 Ratio of boxes 9:1 8:2 7:3 6:4 Ratio of boxes mean ¼ 58.8 (SD 22.0), descend mean ¼ 65.5 (SD 19.8); placebo: ascend mean ¼ 66.5 (SD 22.8), descend mean ¼ 75.9 (SD 13.0)), consistent with previous studies using the Cambridge Gamble Task (eg Clark et al, 2003; Mavaddat et al, 2000). Similarly, the main effect of ratio was not significant (Greenhouse Geisser correction F 1.42,8.52 ¼ 2.17, P ¼ 0.177), although, on average, subjects placed higher bets at the 9 : 1 ratio compared to the 6 : 4 ratio (see Figure 1). All interaction terms in the ANOVA model were nonsignificant (P40.10). Cardiovascular and Subjective Measures Drug Placebo Drug Placebo Figure 2 Methylphenidate treatment did not significantly affect deliberation times (a) or choice of most likely outcome (b) on the Cambridge Gamble Task. Cardiovascular measures at baseline and prior to cognitive testing are shown in Table 2. Baseline measurements of pulse (Z ¼ 0.632, P ¼ 0.528) and systolic blood pressure (Z ¼ 1.20, P ¼ 0.231) did not differ significantly between drug and placebo sessions, although the baseline measurements of diastolic blood pressure approached significance (Z ¼ 1.87, P ¼ 0.061). There was no overall difference in pulse (ie change from baseline) for the drug condition compared to the placebo condition (Z ¼ 0.943, P ¼ 0.345). While there was a general increase in blood pressure after methylphenidate, the differences in systolic and diastolic

30 Table 2 Cardiovascular Measures of Pulse, Systolic Blood Pressure (bp), and Diastolic Blood Pressure (bp) on Drug and Placebo Sessions (Mean (SD)) blood pressure for the drug condition compared to the placebo condition approached significance (systolic blood pressure: Z ¼ 1.78, P ¼ 0.075; diastolic blood pressure: Z ¼ 1.78, P ¼ 0.075). However, these comparisons were nonsignificant when the subject receiving atenolol was excluded from the analysis (systolic blood pressure Z ¼ 1.48, P ¼ 0.138; diastolic blood pressure Z ¼ 1.58, P ¼ 0.115). For the visual analog scales, the change from baseline score was calculated for each drug session. There was no effect of drug on any of these changes compared to placebo. DISCUSSION Methylphenidate Placebo Baseline Pretesting Baseline Pretesting Pulse 77.2 (11.9) 84.5 (14.9) 72.9 (15.3) 73.1 (10.7) Systolic bp 132 (12.1) 140 (14.6) 140 (18.2) 138 (19.3) Diastolic bp 76.8 (9.79) 86.1 (9.88) 85.4 (6.84) 84.3 (11.4) This study reported the effects of methylphenidate upon subjective mood, cardiovascular activity, and a range of cognitive functions in patients with fvftd. The key finding was an attenuation of risk-taking following methylphenidate, on a laboratory measure of decision-making (the Cambridge Gamble Task). This was a relatively selective effect: there were no significant effects of drug treatment on tasks of memory function associated with temporal lobe integrity (recognition memory) or executive tasks associated with dorsolateral prefrontal function (planning, extra-dimensional shifting and working memory). In a previous study of methylphenidate effects in healthy older adults (aged years, mean age ¼ 61 years), we did not detect any significant effects of this medication on risktaking on the Cambridge Gamble Task (Turner et al, 2003). We infer that the significant effect of methylphenidate on risk-taking in patients with fvftd is associated with the behavioral disturbance induced by the dementia. This effect also shows some neurochemical specificity: we have shown previously that chronic treatment with the selective SSRI paroxetine did not affect performance on the Cambridge Gamble Task in an independent group of fvftd patients (Deakin et al, 2004a). The amelioration of risk-taking behavior in fvftd by methylphenidate carries important implications for rehabilitative approaches, given that neurobehavioral deficits including risky behavior and disinhibition represent a significant barrier for social interaction and everyday functioning within society. Under baseline conditions, we have shown previously that patients with fvftd displayed increased betting behavior on the Cambridge Gamble Task compared to matched healthy controls (Rahman et al, 1999). This increased betting was apparent across all ratios of boxes, in both the ascending Methylphenidate and fvftd S Rahman et al and descending conditions of the task (Rahman et al, 1999). The action of methylphenidate in the present study was to normalize risk-taking behavior, bringing the fvftd patients toward the typical performance of healthy older adults (see Figure 1b, with healthy performance taken from Turner et al (2003)). The order of drug administration was fully counter balanced and order did not have a significant effect in the analysis of betting performance. The discrepancy between the ascending and descending conditions on the Cambridge Gamble Task can provide an index of impulsivity or delay aversion. Impulsive or impatient subjects are expected to place high bets in the descend condition but low bets in the ascend condition, whereas a genuine riskpreferent subject would wait in order to place high bets in the ascend condition (see Cools et al, 2003). In the present study, the effect of methylphenidate to reduce betting did not interact with the ratio of boxes, and did not interact with the ascend vs descend betting condition. As such, it seems unlikely that the normalizing effect of methylphenidate results from an antiimpulsive action similar to that seen in ADHD (Aron et al, 2003; Tannock et al, 1989). The beneficial effects of methylphenidate on decisionmaking in fvftd may be mediated at the level of the orbitofrontal cortex, the striatum, or the connectivity between these two regions. Neuropathology in the orbitofrontal cortex is a consistent feature of fvftd (eg Salmon et al, 2003) and is thought to underlie the risk-taking behavior seen under baseline conditions (Rahman et al, 1999). However, the DA transporters that are targeted by methylphenidate are located predominantly within the striatum. The DA projection from the midbrain to the ventral striatum is known to signal reward-related information (Schultz, 2002): specifically, the temporal discrepancy between the occurrence and the prediction of reward (the reward prediction error) (Hollerman and Schultz, 1998). It is possible that neuropathology affecting the DA system in fvftd (eg Sjogren et al, 1998) may increase reward-driven behavior under baseline conditions. Stimulation of DA neurotransmission by methylphenidate may conceivably normalize these changes. In addition, DA stimulation by methylphenidate appears to modulate orbitofrontal cortex activity: methylphenidate administration significantly increased glucose metabolism in the orbitofrontal cortex of cocaine addicts (Volkow et al, 2005), who also display increased risk taking (Bartzokis et al, 2000). These increases in orbitofrontal metabolism were correlated with methylphenidate-induced changes in thalamic DA binding, suggesting a modulatory role of the mesothalamic DA projection (Volkow et al, 2005). An alternative explanation is that the effect on decisionmaking cognition may reflect the action of methylphenidate to produce or modulate central somatic markers (Damasio, 1996) via stimulation of the ascending catecholamine systems. Through connectivity with the amygdala and somatosensory cortex, the orbitofrontal cortex may contribute to decision-making by retrieving somatic information associated with the outcomes of similar decisions in the past (Bechara, 2003; Bechara et al, 2003; Damasio, 1996). These somatic states facilitate an exhaustive cost benefit analysis of decision-making. The amygdala may predominantly mediate the elicitation of emotional responses by conditioned or unconditioned stimuli ( primary inducers ), Neuropsychopharmacology 655

31 656 whereas the orbitofrontal cortex elicits emotional arousal by thoughts, memories, and hypothetical decisions ( secondary inducers ) (Bechara et al, 2003). Via dopaminergic innervation of the orbitofrontal cortex (Oades and Halliday, 1987) and amygdala (Fallon et al, 1978), methylphenidate may modulate or mimic these somatic representations in a decision-making context. Dysfunction in somatic- and emotion-related circuitry may also impair the monitoring of internal state. In the present study, we were unable to demonstrate any significant effects of methylphenidate on subjective mood. This is perhaps remarkable in itself: in previous research, methylphenidate robustly increased subjective ratings of alertness and energy in healthy volunteers (Elliott et al, 1997; Heishman and Henningfield, 1991), including elderly volunteers (Turner et al, 2003). The lack of subjective effects in fvftd implies that these patients are unable to sense differences in their internal state following drug administration. The cardiovascular effects of methylphenidate were unclear in the present study. There was a subtle effect of methylphenidate to increase systolic and diastolic blood pressure. However, this effect only reached trend levels of significance. In addition, baseline diastolic blood pressure was somewhat lower on the methylphenidate session, and therefore, we cannot rule out a regression to the mean effect. Future research is needed in a larger group of patients to confirm whether there is a genuine dissociation between the subjective and cardiovascular effects of methylphenidate in fvftd patients. Economic models distinguish decision-making under risk (when outcome probabilities are explicit) from decisionmaking under ambiguity (when outcome probabilities are undefined) (Ellsberg, 1961; Smith et al, 2002). Bechara et al, (2001) and Bechara (2003) have recently investigated the effects of a DA manipulation on the Iowa Gambling Task. On this task, subjects make a series of choices from four decks of cards that vary in the magnitude and frequency of winning and losing. Subjects are provided with no information about the differing contingencies of the four decks; consequently, the early stages of the task emphasize decision-making under ambiguity. Bechara et al (2001, 2003) reported bidirectional modulation of choice behavior by the dopaminergic agents dextroamphetamine and haloperidol, but these effects were restricted to the early part of the task when outcome probabilities were not well defined. In contrast, the modulatory effects of methylphenidate on the Cambridge Gamble Task suggest that DA predominantly manipulates decision-making under risk, as this task explicitly presents trial-by-trial probabilities. However, it is possible that the cognitive impairment present in these fvftd patients may have affected their explicit knowledge of risk on the Cambridge Gamble Task, and as such, these decisions may have also been made under ambiguity. In the present study, there was no effect of methylphenidate on reversal learning on the ID/ED task. Reversal learning is also robustly linked to orbitofrontal integrity (Clark et al, 2004) and was impaired in fvftd patients in our previous investigation (Rahman et al, 1999). This suggests some parcellation at a neurochemical level within the orbitofrontal cortex. These data are consistent with accumulating evidence using a marmoset analog of the ID/ED task, showing that reversal learning is sensitive to Neuropsychopharmacology Methylphenidate and fvftd S Rahman et al prefrontal 5-HT depletion but not prefrontal DA depletion (Clarke et al, 2004; Crofts et al, 2001). The lack of effect of methylphenidate on traditional measures of executive function, including working memory and planning, suggests a practical limitation on the use of this agent to treat cognitive dysfunction in fvftd. It is notable that the level of performance of the patients in the present study was somewhat more impaired than that of the mild patients reported previously (Rahman et al, 1999). Mild cases may stand to benefit more from cognition-enhancing medication. A previous study by Coull et al (1996) showed that idazoxan, an a 2 NA antagonist that acts presynaptically to elevate NA activity, produced several instances of cognitive improvement in fvftd, but impaired spatial working memory. Recent neuropathological findings support the notion that NA neurotransmission is abnormal in fvftd, with levels reduced by at least 30% in the nucleus basalis, thalamus, locus coeruleus, and amygdala (Nagaoka et al, 1995). It is possible that some cognitive effects associated with DA or NA challenges may be cancelled out by the combined actions of methylphenidate on catecholamine neurotransmission. Future research may utilize more selective DA agents such as the D2 agonist bromocriptine. Some further limitations of the present study are the small number of participants, the use of an acute (ie single dose) study design, and the presence of additional medications in five of the eight patients, which may interact with the effects of methylphenidate. These encouraging findings must be treated as preliminary and require extension and replication in a larger sample size with a chronic treatment design. ACKNOWLEDGEMENTS We thank the patients and carers for participating in this study, which was funded by a Wellcome Trust Programme grant awarded to TWR, BJS, BJ Everitt, and AC Roberts, and completed within the MRC Centre for Behavioural and Clinical Neuroscience in Cambridge. BJS and TWR are consultants for Cambridge Cognition Ltd. REFERENCES Alexander GE, DeLong MR, Strick PL (1986). Parallel organization of functionally segregated circuits linking basal ganglia and cortex. Ann Rev Neurosci 9: Aron AR, Dowson JH, Sahakian BJ, Robbins TW (2003). Methylphenidate improves response inhibition in adults with attention-deficit/hyperactivity disorder. Biol Psychiatry 54: Bartzokis G, Lu PH, Beckson M, Rapoport R, Grant S, Wiseman EJ et al (2000). Abstinence from cocaine reduces high-risk responses on a gambling task. Neuropsychopharmacology 22: Bechara A (2003). Emotion, decision-making and addiction. J Gamb Stud 19: Bechara A, Damasio H, Damasio A (2001). Manipulation of dopamine and serotonin causes different effects on covert and overt decision-making. Soc Neurosci Abstracts 27: Bechara A, Damasio H, Damasio AR (2003). Role of the amygdala in decision-making. Ann NY Acad Sci 985: Bettendorff L, Mastrogiacomo F, Wins P, Kish SJ, Grisar T, Ball MJ (1997). Low thiamine phosphate levels in brains of patients

32 with frontal lobe degeneration of the non-alzheimer s type. J Neurochem 69: Bond A, Lader M (1974). The use of analogue scales in rating subjective feelings. Br J Med Psychol 47: Chatterjee A, Fahn S (2002). Methylphenidate treats apathy in Parkinson s disease. J Neuropsychiatry Clin Neurosci 14: Clark L, Cools R, Robbins TW (2004). The neuropsychology of ventral prefrontal cortex: decision-making and reversal learning. Brain Cogn 55: Clark L, Manes F, Antoun N, Sahakian BJ, Robbins TW (2003). The contributions of lesion laterality and lesion volume to decisionmaking deficits following frontal lobe damage. Neuropsychologia 41: Clarke HF, Dalley JW, Crofts HS, Robbins TW, Roberts AC (2004). Cognitive inflexibility after prefrontal serotonin depletion. Science 304: Conners C, Eisenberg L (1963). The effects of methylphenidate on symptomatology and learning in disturbed children. Am J Psychiatry 120: Cools R, Barker RA, Sahakian BJ, Robbins TW (2003). L-Dopa medication remediates cognitive inflexibility, but increases impulsivity in patients with Parkinson s disease. Neuropsychologia 41: Coull JT, Sahakian BJ, Hodges JR (1996). The alpha(2) antagonist idazoxan remediates certain attentional and executive dysfunction in patients with dementia of frontal type. Psychopharmacology 123: Crofts HS, Dalley JW, Collins P, Van Denderen JC, Everitt BJ, Robbins TW et al (2001). Differential effects of 6-OHDA lesions of the frontal cortex and the caudate nucleus on the ability to acquire an attentional set. Cereb Cortex 11: Damasio AR (1996). The somatic marker hypothesis and the possible functions of the prefrontal cortex. Philos Trans R Soc London B 351: Deakin JB, Aitken MRF, Robbins TW, Sahakian BJ (2004b). Risktaking during decision-making in normal volunteers changes with age. J Int Neuropsychol Soc 10: Deakin JB, Rahman S, Nestor PJ, Hodges JR, Sahakian BJ (2004a). Paroxetine does not improve symptoms and impairs cognition in frontotemporal dementia: a double-blind randomized controlled trial. Psychopharmacology 172: Diehl J, Grimmer T, Drzezga A, Riemenschneider M, Forstl H, Kurz A (2004). Cerebral metabolic patterns at early stages of frontotemporal dementia and semantic dementia. A PET study. Neurobiol Aging 25: Elliott R, Sahakian BJ, Matthews K, Bannerjea A, Rimmer J, Robbins TW (1997). Effects of methylphenidate on spatial working memory and planning in healthy volunteers. Psychopharmacology 131: Ellsberg D (1961). Risk, ambiguity and the savage axioms. Q J Econ 75: Fallon JH, Koziell DA, Moore RY (1978). Catecholamine innervation of the basal forebrain II. Amygdala, suprarhinal cortex and entorhinal cortex. J Comp Neurol 180: Folstein MF, Folstein SE, McHugh PR (1975). Mini mental state: a practical method of grading the cognitive state of patients for the clinician. J Psychiatr Res 12: Francis PT, Holmes C, Webster MT, Stratmann GC, Procter AW, Bowen DM (1993). Preliminary neurochemical findings in non-alzheimer dementia due to lobar atrophy. Dementia 4: Gilbert JJ, Kish SJ, Chang LJ, Morito C, Shannak K, Hornykiewicz O (1988). Dementia, Parkinsonism, and motor neurone disease: neurochemical and neuropathological correlates. Ann Neurol 24: Gilman AG, Goodman LS, Gilman A (eds) (1980). Pharmacological Basis of Therapeutics. MacMillan: New York. Methylphenidate and fvftd S Rahman et al Gualtieri CT, Evans RW (1998). Stimulant treatment for the neurobehavioural sequelae of traumatic brain injury. Brain Injury 2: Heishman SJ, Henningfield JE (1991). Discriminative stimulus effects of d-amphetamine, methylphenidate and diazepam. Psychopharmacology 103: Hollerman JR, Schultz W (1998). Dopamine neurons report an error in the temporal prediction of reward during learning. Nat Neurosci 1: Howell DC (1997). Statistical Methods for Psychology, 4th edn. Duxbury Press: Boston. Ibach B, Poljansky S, Marienhagen J, Sommer M, Manner P, Hajak G (2004). Contrasting metabolic impairment in frontotemporal degeneration and early onset Alzheimer s disease. Neuroimage 23: Kobayashi K, Hayashi M, Fukutani Y, Miyazu K, Shiozawa M, Aoki T. et al (1999). KP1 expression in ghost Pick bodies, amyloid P-positive astrocytes and selective nigral degeneration in early Pick s disease. Clin Neuropathol 18: La Coco D, Nacci P (2004). Frontotemporal dementia presenting with pathological gambling. J Neuropsychiatry Clin Neurosci 16: Lee AC, Rahman S, Hodges JR, Sahakian BJ, Graham KS (2003). Associative and recognition memory for novel objects in dementia: implications for diagnosis. Eur J Neurosci 18: Liu W, Miller BL, Kramer JH, Rankin K, Wyss-Coray S, Gearhart P et al (2004). Behavioural disorders in the frontal and temporal variants of frontotemporal dementia. Neurology 62: Lund and Manchester Groups (1994). Clinical and neuropathological criteria for frontotemporal dementia. J Neurol Neurosurg Psychiatry 57: Marin RS, Fogel BS, Hawkins J, Duffy J, Krupp B (1995). Apathy: a treatable syndrome. J Neuropsychiatry Clin Neurosci 7: Mavaddat N, Kirkpatrick PJ, Rogers RD, Sahakian BJ (2000). Deficits in decision-making in patients with aneurysms of the anterior communicating artery. Brain 123: Mega MS, Cummings JL (1994). Frontal subcortical circuits and neuropsychiatric disorders. J Neuropsychiatry Clin Neurosci 6: Mehta MA, Calloway P, Sahakian BJ (2000). Amelioration of specific working memory deficits by methylphenidate in a case of adult attention-deficit/hyperactivity disorder. J Psychopharmacol 14: Mendez MF, Cummings JL (2003). Frontotemporal dementia and the asymmetric cortical atrophies. In: Dementia: A Clinical Approach. Butterworth-Heinemann: Philadelphia, PA. pp Muller U, Suckling J, Zelaya F, Honey G, Faessel H, Williams SCR et al (2005). Plasma level-dependent effects of methylphenidate on task-related functional magnetic resonance imaging signal changes. Psychopharmacology, April 14 [Epub ahead of print]. Nagaoka S, Arai H, Iwamoto N, Ohwada J, Ichimiya Y, Nakamura M et al (1995). A juvenile case of frontotemporal dementia: neurochemical and neuropathological investigation. Prog Neuropsychopharmacol Biol Psychiatry 19: Oades RD, Halliday GM (1987). Ventral tegmental (A10) system: neurobiology. 1. Anatomy and connectivity. Brain Res 434: Owen AM, Sahakian BJ, Hodges JR, Summers BA, Polkey CE, Robbins TW (1995). Dopamine-dependent frontostriatal planning deficits in early Parkinson s disease. Neuropsychology 9: Plenger PM, Dixon CE, Castillo RM, Frankowski RF, Yablon SA, Levin HS (1996). Subacute methylphenidate treatment for moderate to moderately-severe traumatic brain injury: a preliminary double-blind placebo-controlled study. Arch Phys Med Rehabil 77: Neuropsychopharmacology 657

33 658 Methylphenidate and fvftd S Rahman et al Rahman S, Sahakian BJ, Hodges JR, Rogers RD, Robbins TW (1999). Specific cognitive deficits in early frontal variant frontotemporal dementia. Brain 122: Rahman S, Sahakian BJ, Gregory CA (2000). Therapeutic strategies in early onset dementia. In: Hodges JR (ed). Early Onset DementiaFA Multidisciplinary Approach. Oxford University Press: Oxford. Ratnavalli E, Brayne C, Dawson K, Hodges JR (2002). Prevalence of frontotemporal dementia. Neurology 58: Rogers RD, Everitt BJ, Baldacchino A, Blackshaw AJ, Swainson R, Wynne K et al. (1999). Dissociable deficits in the decisionmaking cognition of chronic amphetamine abusers, opiate abusers, patients with focal damage to prefrontal cortex, and tryptophan-depleted normal volunteers: evidence for monoaminergic mechanisms. Neuropsychopharmacology 20: Rolls ET, Hornak J, Wade D, McGrath J (1994). Emotion-related learning in patients with social and emotional changes associated with frontal lobe damage. J Neurol Neurosurg Psychiatry 57: Salmon E, Garraux G, Delbeuck X, Collette F, Kalbe E, Zuendorf G et al (2003). Predominant ventromedial frontotemporal metabolic impairment in frontotemporal dementia. Neuroimage 20: Scheel-Krüger J (1971). Comparative studies of various amphetamine analogues demonstrating different interactions with the metabolism of the catecholamines in the brain. Eur J Psychopharmacology 14: Schultz W (2002). Getting formal with dopamine and reward. Neuron 36: Sjogren M, Minthon L, Passant K, Wallin A (1998). Decreased monoamine metabolites in frontotemporal dementia and Alzheimer s disease. Neurobiol Aging 19: Smith K, Dickhaut J, McCabe K, Pardo JV (2002). Neuronal substrates for choice under ambiguity, risk, gains and losses. Management Sci 48: Sparks DL, Markesbery WR (1991). Altered serotonergic and cholinergic synaptic markers in Pick s disease. Arch Neurol 48: Speech TJ, Rao SM, Osmon DC, Sperry DT (1993). A double-blind controlled study of methylphenidate treatment in closed head injury. Brain Injury 7: Tannock R, Schachar R, Carr RP, Chajczyk D, Logan GD (1989). Effects of methylphenidate on inhibitory control in hyperactive children. J Abnorm Child Psychol 17: Turner DC, Robbins TW, Clark L, Aron AR, Dowson J, Sahakian BJ (2003). Relative lack of cognitive effects of methylphenidate in elderly volunteers. Psychopharmacology 168: Volkow ND, Wang GJ, Fowler JS, Gatley SJ, Logan J, Ding YS et al (1998). Dopamine transporter occupancies in the human brain induced by therapeutic doses of methylphenidate. Am J Psychiatry 155: Volkow ND, Wang GJ, Ma Y, Fowler JS, Wong C, Ding YS et al (2005). Activation of orbital and medial prefrontal cortex by methylphenidate in cocaine-addicted subjects but not in controls: relevance to addiction. J Neurosci 25: Whyte J, Hart T, Schuster K, Fleming M, Polansky M, Corlett HB (1997). Effects of methylphenidate on attentional function after traumatic brain injury: a randomized, placebo-controlled trial. Am J Phys Med Rehabil 76: Williams SE, Ris MD, Ayyanger R, Schefft BK, Berch D (1998). Recovery in pediatric brain injury: is psychostimulant medication beneficial? J Head Trauma Rehabil 13: Neuropsychopharmacology

34 Psychopharmacology (2004) 172: DOI /s ORIGINAL INVESTIGATION J. B. Deakin S. Rahman P. J. Nestor J. R. Hodges B. J. Sahakian Paroxetine does not improve symptoms and impairs cognition in frontotemporal dementia: a double-blind randomized controlled trial Received: 9 June 2003 / Accepted: 23 October 2003 / Published online: 10 December 2003 Springer-Verlag 2003 J. B. Deakin S. Rahman B. J. Sahakian () ) Department of Psychiatry, University of Cambridge, Addenbrooke s Hospital, Box 189, CB2 2QQ Cambridge, England, UK jenny.hall@addenbrookes.nhs.uk P. J. Nestor J. R. Hodges Department of Neurology, University of Cambridge, Addenbrooke s Hospital, CB2 2QQ Cambridge, England, UK Abstract Rationale: Patients with frontal variant frontotemporal dementia (fvftd) present with disinhibition, impulsiveness, apathy, altered appetite and stereotypic behaviors. A non-randomized clinical trial found improvement in these symptoms after treatment with a selective serotonin reuptake inhibitor (SSRI). Objectives: We aimed to subject a SSRI, paroxetine, to a more rigorous test of its efficacy using a double-blind, placebocontrolled experimental design. Methods: Ten subjects meeting the consensus criteria for FTD were entered into a double-blind, placebo-controlled crossover trial. Doses of paroxetine were progressively increased to 40 mg daily. The same regimen was used for placebo capsules. Subjects were assessed with a battery of cognitive tests in the sixth week of paroxetine and placebo treatment. At each assessment, caregivers were interviewed using the Neuropsychiatric Inventory and asked to complete the Cambridge Behavioral Inventory. Results: There were no significant differences on the Neuropsychiatric Inventory or the Cambridge Behavioral Inventory. Paroxetine caused a decrease in accuracy on the paired associates learning task, reversal learning and a delayed pattern recognition task. There were no changes on the decisionmaking task, in spatial span, spatial recognition, spatial working memory, digit span and verbal fluency. Conclusions: This study finds no evidence for the efficacy of paroxetine in the treatment of fvftd. The results suggest that a chronic course of paroxetine may selectively impair paired associates learning, reversal learning and delayed pattern recognition. This pattern of deficits closely resembles that seen after tryptophan depletion. Results are discussed with respect to current theories on serotonergic modulation of orbitofrontal/ventromedial prefrontal cortex. Keywords Decision-making Frontotemporal dementia Paroxetine Serotonin Introduction Frontotemporal dementia (FTD) is the term now preferred to describe non-alzheimer-type dementia affecting the frontal and/or anterior temporal lobes (Hodges and Miller 2001). FTD typically occurs at a younger age than most dementias, when patients still have parental and financial responsibilities for children. The prevalence in the under- 65 s is almost equal to Alzheimer s disease (Ratnavalli et al. 2002). FTD represents a continuum: at one end is the temporal form, known more commonly as semantic dementia, and, at the other end, the frontal variant of FTD (fvftd) (Orrell and Sahakian 1991). fvftd is characterized by profound changes in personality and social behavior with stereotypic ritualistic behavior (Bozeat et al. 2000; Neary et al. 2000). Patients may ignore visitors, or address them rudely, and may fail to offer food and drink to others, they may also make personal comments within earshot of strangers (Plaisted and Sahakian 1997). Changes in mood and apathy are common and many patients develop an increased desire for sweet food coupled with reduced satiety. This can result in considerable weight gain and caregivers are often led to ration food. The loss of concern and empathy for family and friends places a great emotional burden on caregivers. Impaired serotonin (5-HT) function has been implicated in the pathogenesis of depression, anxiety, obsessional compulsive and eating disorders (Mann 1999). Drugs such as the selective serotonin reuptake inhibitors (SSRIs), which decrease 5-HT re-uptake, are particularly effective in the treatment of these symptoms (Blier and de Montigny 1999). There is also increasing evidence that reduced 5-HT function is involved in impulsive behavior

35 in antisocial individuals (Dolan et al. 2001). One study in post-mortem brains reported reductions in post-synaptic 5-HT receptor binding in the frontal lobes of patients who died in the early stages of FTD (Sparks and Markesbery 1991). This raises the possibility that impaired 5-HT function contributes to the pathogenesis of some symptoms in fvftd and that SSRIs might ameliorate such symptoms. A further rationale for investigating the effect of a SSRI in fvftd comes from recent evidence that changes in 5-HT function may specifically alter performance on neuropsychological tasks mediated by orbitofrontal and not dorsolateral circuitry (Robbins 2000; Rogers and Robbins 2001), the same circuitry that has been specifically implicated in fvftd. Normal volunteers who have undergone acute tryptophan depletion, which acutely reduces central 5-HT, demonstrate the same deficits on reversal learning as seen in fvftd (Rahman et al. 1999; Rogers et al. 1999a). This same deficit is also seen in patients with damage to the orbitofrontal cortex (OFC) but not the dorsolateral cortex (DLPFC) (Rogers et al. 1999b). The pattern of results with Gamble (Rogers et al. 1999b), a decision making task, is more complex. Performance on the decision-making task is sensitive to damage of the OFC (but not the DLPFC), and tryptophan depletion (Rogers et al. 1999a, 1999b). However, the decision-making task elicits a slightly different deficit in fvftd patients (Rahman et al. 1999), which is the same deficit that is seen in patients with large lesions of the frontal lobe (Manes et al. 2002) and with aneurysms of the anterior communicating artery (Mavaddat et al. 2000). The precise effect of ventral frontal lesions on decision making appears to depend on the extent and location of the lesion. As yet, no discrete lesions of frontal cortex have been found to reproduce the fvftd deficit in decision making, nor have any pharmacological manipulations exactly mimicked it. That tryptophan depletion impairs both reversal and decision making in normal volunteers suggests that increasing 5-HT function could improve symptoms and cognition in fvftd. This would occur by reversing the possible deficit in 5-HT function and enhancing OFC function (Robbins 2000; Rogers and Robbins 2001). Since one of the effects of fvftd is the same as that seen with tryptophan depletion, and decreased 5-HT binding has been found in the brains of fvftd patients, we hypothesized that enhanced 5-HT release by chronic treatment with paroxetine would reduce the cognitive deficits. Specifically, it was hypothesized that paroxetine would improve performance of those tests that are impaired by tryptophan depletion, namely decision making (Rogers et al. 1999b), reversal learning (Rogers et al. 1999a), delayed pattern recognition (Rubinsztein et al. 2001) and paired associates learning (Park et al. 1994). Whilst Alzheimer s disease research has yielded a variety of therapeutic strategies (Grundman et al. 1998), comparatively little work has been done on the treatment of FTD. In two recent preliminary studies, Swartz et al. (1997) assessed the effect of three different SSRIs on a group of frontal and temporal variant patients, and Moretti et al. (2003) compared the effect of paroxetine with that of a g-amino butyric acid (GABA) analogue, piracetam. Both studies reported improvements in clinical symptoms after treatment with SSRIs. These findings are far from conclusive. Swartz et al. reported no statistically significant improvement; Moretti et al. observed significant improvement relative to the other therapy. Neither study used a placebo control. Therefore, the present placebo-controlled study assessed the effect of paroxetine therapy in patients with fvftd to test the prediction that increasing 5-HT function would be associated with improved performance on tests of ventral frontal lobe function. Methods Participants Patients with a clinical diagnosis of fvftd were recruited through the Memory and Early Onset Dementia Clinic at Addenbrooke s Hospital, Cambridge. Patients were screened for psychiatric or neurological illness and any serious medical conditions other than FTD. All patients met internationally agreed criteria for FTD (Neary et al. 1998) and conformed to local guidelines previously applied showing at least 5 of 12 clinical features (Gregory 1999; Gregory et al. 1999; Rahman et al. 1999; Bozeat et al. 2000). Written informed consent for all procedures was obtained from both patients and caregivers. The study was approved by the Cambridge Local Research Ethics Committee (reference number LREC 99/017). Experimental methods Patients were assessed on neuropsychological tasks and their caregivers were asked to fill out questionnaires. Two treatment assessments were made: one when the subject was on a daily 40-mg dose of paroxetine and another when taking identical capsules containing placebo. The order in which different patients received the two treatments was randomized and counterbalanced. The two treatment periods were separated by a minimum of 5 weeks to ensure complete recovery from the effects of the drug. Drug regimen Caregivers were asked to be responsible for the administration of the capsules. Patients were started on a daily 20-mg dose. This dose was increased to 30 mg in week 2 and 40 mg in week 3. The patient then remained on 40 mg for 4 weeks. The dose was cautiously reduced by 10 mg each week to avoid a SSRI discontinuation syndrome (see Table 1.). The whole treatment lasted for 9 weeks. The treatment assessment was carried out in the sixth or seventh week of the regime. The experimenter, caregiver and patient were all blind to the contents of the capsule. On every session, subjects were assessed for verbal fluency (Benton 1968), and forward and backward digit span was assessed. Computerized neuropsychology 401 The tests were taken from the Cambridge Neuropsychological Test Automated Battery (CANTAB) and the CANTAB extensions. These tests were administered using a portable Datalux 486 microcomputer fitted with a touch sensitive screen or an Advantech PPI 120-T touchscreen. Subjects were seated comfortably ~0.5 m

36 402 Table 1 Drug regimen Week away from the touch screen. In order to introduce subjects to the apparatus, they were initially given a motor screening task in which they were asked to respond to a series of ten flashing crosses presented at varying locations on the monitor screen by touching the center of the cross with the index finger of their preferred hand. The experimenter demonstrated by touching three crosses. After completion of this task, subjects were given the following tasks in the order described below. Immediate and delayed pattern recognition In the pattern recognition task, subjects are presented with 12 abstract colored patterns and asked to try to remember them. In the recognition phase, the same patterns (each paired with a novel pattern) are presented in the reverse order and subjects were asked to respond by touching the pattern they had already seen (Sahakian et al. 1988). This procedure is then repeated with 12 new patterns. The recognition phases are then repeated 20 min later to assess delayed pattern recognition. Spatial recognition In the spatial recognition task (Sahakian et al. 1988), five squares are presented sequentially in the different locations around the screen. In the recognition phase, subjects are presented with a choice of two squares in different locations around the screen, one of which is novel. Subjects are asked to touch the square which is in the location that a square has previously appeared in. This procedure is repeated a further three times. Spatial span Each trial begins with nine white boxes presented in fixed locations on the monitor screen. Initially, two of the boxes change color, one after the other in a predetermined sequence. The end of the sequence is indicated by a tone. Subjects are then asked to point to the boxes in the order in which they have changed color. After successful completion of a sequence, the number of boxes changing color increases by one, up to a maximum of nine. The test is terminated when three consecutive failures occur at any one sequence length. During each trial, the number in the bottom left corner of the screen indicated the length of the current sequence. The spatial span is calculated as the highest level at which a subject successfully recalled at least one sequence of boxes (Owen et al. 1990). This measure is used to assess the ability of subjects to hold information on-line in order to plan a series of moves such as the spatial working memory task. Spatial working memory Daily dose of paroxetine (mg) In this task, subjects are required to search through a number of colored boxes on the monitor screen (by touching each one) in order to find blue tokens, which are hidden inside (Owen et al. 1990). On any trial, only a single token is hidden in one of the boxes. Once found, the next token is hidden. The key instruction is that once a token is found within a particular box, then that box will not be used again to hide a token. Two types of error are possible. First, a subject may return to a box in which a token has already been found (a between-search error). Second, a subject may return to a box already opened and shown to empty in the same trial (a within-search error). There are four trials with each of four, six and eight boxes. The task is scored according to the number of betweenand within-search errors at each level of difficulty and for the use of an effective search strategy (Owen et al. 1990). Both error scores are a measure of spatial working memory, but the between-search error is more stringent as the subject is required to remember across searches which boxes contained blue tokens while conducting a new search (Joyce and Robbins 1991). Visual discrimination learning/attentional set shifting This visual discrimination learning/attentional set-shifting test (Downes et al. 1989) is one of attentional set shifting based in part on the Wisconsin Card Sort Test (WCST) (Milner 1964). There are nine stages in which a subject has to learn a visual discrimination to criterion (six consecutive trials correct). After reaching criterion, the subject passes to the next stage. The task starts with a simple discrimination and its reversal for stimuli varying in just one dimension (e.g. two different white line configurations). A second alternative dimension is then introduced (purple-filled shapes) and compound discrimination and its reversal are tested. To succeed, subjects must continue to respond to the previously relevant dimension whilst ignoring the presence of the new, irrelevant dimension. At the intradimensional shift stage, novel exemplars of each of the two dimensions are introduced and subjects must continue to respond to one of the two exemplars from the previously relevant dimension. Following another reversal, the extradimensional shift and its reversal are presented, again using novel exemplars of each stimulus dimension. In order to succeed at the extradimensional shift stage, the subject must shift attentional or response set to the previously irrelevant stimulus dimension whilst ignoring the previously relevant stimulus dimension. The extradimensional shift stage is akin to a change in category in the WCST. For each stage of the test, the computer calculates the number of trials to criterion, number of errors made, and the latency to complete each discrimination. Decision-making Gamble (Rogers et al. 1999a) This decision-making task was based on the Iowa gambling task (Bechara et al. 1999). The subject is told that the computer has hidden at random a yellow token inside one of the ten red or blue boxes arrayed at the top of the screen. The likelihood of each choice being correct is therefore indicated to the subject on each trial by the ratio of red to blue boxes displayed on each trial. This produces a range of situations from one in which one outcome is much more likely (9:1) to those in which the two outcomes are almost equally likely (6:4). The subject is then asked whether the token is hidden in a red or a blue box. After making this initial choice, the subject attempts to increase a total points score by placing a bet on this choice being correct. The available bets appear in sequence, one after another. The subject performs the task in two separate conditions. In the ascending condition, the first bet offered is small and increases in size. In order to make a large bet, the subject has to wait until the computer displays it. The three principal measures in this task are (i) the quality of decision making, i.e., what proportion of the time the subjects indicate that the token is in the color of box that it is the most likely to be in (ii) risk taking, i.e., the percentage of points the subject is prepared to bet on the token being in the color of box they said it was and (iii) the speed of decision making, i.e., how long it takes the subject to decide which color of box is hiding the token.

37 Table 2 Details of the ten subjects that participated in the trial Patient Age (years) Sex National adult reading test estimated IQ Mini mental state examination Behavioral measures Neuropsychological measures 1 65 Male Male Male Male Male Male Male x Female x 9 65 Female x Female 15 3 All bar pattern recognition memory, decision making, digit span 403 Paired-associates learning In this task (Sahakian et al. 1988; Fowler et al. 1995), six white boxes appear in a circle around the screen. These open for 3 s and in a random order. Some of the boxes open to reveal one of a number of abstract color stimuli, others may be empty. A probe stimulus appears at the center after all the boxes have been opened, and the subject is required to touch the box where it has previously appeared. At first, only two stimuli are hidden. An incorrect response results in the whole cycle of stimulus presentation being repeated for a maximum of ten cycles in total. At this point, the test would terminate. Otherwise, the stimulus set size is increased from two to three to six and finally eight different stimuli (two extra boxes being added to the display on these occasions). Results Subject details Twelve subjects were initially approached to be included in the trial. Two subjects withdrew their consent before measurements could be taken on the first drug session. The clinical details from all ten subjects are shown in Table 2. Note that incomplete data were available from four subjects, two from each counterbalancing condition, who did not wish to complete the testing. Caregiver interviews and questionnaires Behavioral data were collected during caregiver interviews using the Neuropsychiatric Inventory (NPI) (Cummings et al. 1994). This instrument assesses ten behaviors, namely: delusions, hallucinations, agitation, depression, anxiety, euphoria, apathy, disinhibition, irritability and aberrant motor behavior (including pacing rummaging and compulsions). A frequency rating (1 4) multiplied by a severity rating (1 3) produces a subscale score for each behavior and the summation of subscale scores produces the total NPI score. The NPI has shown to be valid when compared with a variety of other diagnostic approaches and to have a high inter-rater and testretest reliability (Cummings et al. 1994). The Cambridge Behavioral Inventory (CBI), validated through previous work by Bozeat et al. (2000) and Perry and Hodges (1999), is a questionnaire containing 81 items. As well as changes in behavior, personality and mood (disinhibition, agitation, aggression, eating/appetite, sleep motivation and stereotypic behaviors) also included are questions that probe memory, orientation and activities of daily living. Caregivers were asked to fill out one questionnaire on every test session. Paired-associates learning The numbers of errors made at the six-box level of difficulty during paroxetine and placebo are shown in Fig. 1. Only four of the subjects completed the eight-box task and so these data were unavailable for analysis. Seven of the eight subjects made more errors at the sixbox stage while taking paroxetine than when taking placebo; only one subject made fewer errors on paroxetine. The overall group difference neared significance (n=8) (F 1,6 =5.586, P=0.056). In summary, paroxetine increased the number of errors made on the pairedassociates learning task. Data analysis The experimental data were analyzed using the Statistical Package for the Social Sciences Version [SPSS V9.0.1] (SPSS inc. Chicago, Ill., USA). The order that the drug was received in was included as a between-subjects factor of no interest. The data shown in the figures always represent untransformed values. Fig. 1 Paired-associates learning: errors at the six pattern stage

38 404 Decision making One subject was not willing to complete both arms of the test on one occasion and therefore the data from the ascend condition of the task were excluded. Quality of decision making There were no differences in the quality of decision making across the two different sessions, whether considered over all ratios (n=7) (F 1,5 =1.760, P=0.242) or decisions made in response to particular ratios of boxes (n=7) (F 3,15 =0.720, P=0.527). Risk taking FvFTD subjects did not bet in a significantly different fashion on paroxetine relative to placebo overall (n=7) (F 1,5 =0.011, P=0.920) or in response to specific ratios (F 3,15 =0.889, P=0.469). It can be seen from Table 3 that the average percentage bet did not differ between the conditions. The subjects did not significantly adjust their bets in response to the different ratios of red and blue boxes presented to them (n=7) (F 3,15 =1.378, P=0.296). Deliberation time The time needed to come to a decision about what color box the token was hidden behind was not affected by treatment (n=7) (F 1,5 =0.878, P=0.392) or in response to the different ratios of red and blue boxes (n=7) (F 3,15 = 0.863, P=0.451). In summary, there was no effect of paroxetine on any of the measures in the decision-making task. Visual discrimination learning/attentional set shifting Table 3 Means (standard errors) of psychological measures under paroxetine and placebo Paroxetine Placebo Paired-associates learning 6 Pattern errors 13 (1.1) 7.3 (2.3) Gamble (descend condition only) Quality of decision making 0.92 (0.05) 0.89 (0.09) Percentage bet 78 (4.3) 78 (5.4) Deliberation time (ms) 8.8 (3.0) 6.5 (1.4) Intradimensional/extradimensional set shifting Intradimensional proportion correct 0.65 (0.15) 0.74 (0.12) Extradimensional proportion correct 0.32 (0.11) 0.52 (0.09) Reversal proportion correct 0.55 (0.07) 0.64 (0.07) Delayed pattern recognition Number correct (out of 24) 13.6 (1.1) 16.2 (1.1)* Latency (ms) 4.1 (0.7) 4.1 (0.4) Immediate pattern recognition Number correct (out of 24) 15.0 (1.4) 16.7 (1.6) Latency (ms) 5.2 (1.1) 3.7 (0.4) Spatial span Span 4.1 (0.4) 4.6 (0.4) Errors 10.3 (1.2) 12.3 (1.4) Spatial recognition Latency (ms) 4.9 (0.8) 4.1 (0.5) Percentage correct 55 (5) 64 (6) Spatial working memory Between-search errors 49.9 (9.3) 49.8 (10.3) Within-search errors 6.1 (3.5) 5.0 (2.7) Strategy 37.8 (1.6) 36.8 (1.3) Neuropsychiatric inventory NPI (7.2) 28.8 (4.8) NPI (3.3) 15.5 (3.86) Cambridge behavioral inventory 41.1 (3.6) 38.5 (4.3) Verbal fluency 5.4 (1.7) 4.9 (1.2) Digit span Forwards 5.1 (0.9) 6.3 (0.9) Backwards 3.8 (1.2) 4.8 (1.0) Rahman et al. (1999) used number of errors as an assay of performance. This measure discriminated between patient and controls at reversal stages. Given that many of the subjects in the present study did not complete the extradimensional shift stage, this would not be an appropriate measure here. Instead, performance was measured by the percentage of trials in each stage that received a correct response; in this way a subject who failed to reach a given stage was scored with a zero for that stage. This pattern was investigated by using an ANOVA on the transformed proportion correct scores. The proportion scores were arcsine transformed [2 arcsine ( p x)]. There was no overall effect of paroxetine (n=8) (F 1,6 =2.052, P=0.202) and a similar decline in performance over the stages of the task (F 8,48 =9.780, P<0.001). There was, however, a significant interaction of session by stage (F 8,48 =3.446, P=0.003), indicating that paroxetine was having specific effects on particular stages. To investigate this effect further, we compared the transformed proportion correct scores for the intra-dimensional shift stage, the extra-dimensional shift stage and the combined proportion correct for the four reversal stages in three different repeated-measures ANOVAs. Paroxetine did not significantly affect the ability of the subjects to perform an intra-dimensional shift (F 1,6 =0.574, P=0.477) or an extra-dimensional shift (F 1,6 =4.776, P=0.072), but paroxetine tended to impair a subject s ability to perform a reversal shift (F 1,6 =5.960, P=0.050) (see Fig. 2). In summary, paroxetine impaired reversal shifts on the attentional set-shifting task.

39 Immediate pattern recognition memory 405 The deficit in pattern recognition after a delay was not seen if subjects were asked to recognize patterns immediately, neither in terms of accuracy (n=7) (F 1,5 =0.68, P=0.45) nor latency (n=7) (F 1,5 =2.429, P=0.180). Spatial span length and errors There was no effect of paroxetine on span length (n=8) (F 1,6 =3.429, P=0.114) or on the number of errors made (n=8) (F 1,6 =1.391, P=0.283). Spatial recognition memory and response latency Fig. 2 Visual discrimination/attentional set shifting: proportion of correct reversal trials Paroxetine did not have a significant impact on spatial recognition memory (n=8) (F 1,6 =4.592, P=0.076) or on the transformed response latency (n=8) (F 1,6 =0.594, P=0.470). Spatial working memory Between search errors There was no effect of paroxetine on the number of between-search errors (n=8) (F 1,6 =0.001, P=0.978). Not surprisingly, the number of between-search errors increased with increasing level of difficulty from four to six to eight box problems (n=8) (F 2,10 =90.435, P<0.001). There was no interaction of difficulty with session on the number of between-search errors (F 2,10 =0.125, P=0.841). Within-search errors Fig. 3 Delayed pattern recognition: accuracy Delayed pattern recognition memory Subjects were significantly less accurate under the influence of paroxetine than placebo (n=7) (F 1,5 =11.2, P=0.020). As can be seen from Fig. 3, one subject scored the same when taking paroxetine as he did when taking placebo, but the other six subjects all performed more poorly when taking paroxetine. There were no significant differences in latency (n=7) (F 1,5 =0.024, P=0.884). In summary, paroxetine impaired accuracy but not speed on the delayed pattern recognition task. ANOVA was used to compare the test session across the level of order and the three levels of search difficulty. There was no effect on the number of within-search errors found by comparing the drug and placebo conditions (n=8) (F 1,6 =0.164, P=0.700). Subjects in our study did not make more within-search errors with increasing level of difficulty from four to six to eight box problems (n=8) (F 2,12 =2.798, P=0.101). There was no interaction of difficulty with session on the number of within-search errors (F 1,7 =0.100, P=0.784). Strategy The strategy scores are very consistent between paroxetine and placebo sessions (F 1,6 =3.097, P=0.129).

40 406 Neuropsychiatric inventory The NPI assesses 12 behavioral domains based on an interview with the caregiver. As well as the total NPI-12 score we also used a preselected subscore consisting of four symptoms that are particularly prominent in fvftd (disinhibition, apathy, aberrant motor behavior and euphoria) (Levy et al. 1996). The technique of selecting a cluster of NPI scores was used after a study that used a similar method four items that mark out Lewy body dementia (McKeith et al. 2000). Total NPI score Nine caregivers were willing to be interviewed on both drug conditions. There was no significant effect of paroxetine relative to placebo on the total NPI score (n=9) (F 1,7 =0.119, P=0.740). NPI4 There was no significant effect of paroxetine relative to placebo on the NPI-4 score (F 1,7 =0.013, P=0.913). It can be seen from Table 3 that both the NPI4 and the total NPI did not change a great deal after treatment with paroxetine. Cambridge Behavioral Inventory The CBI was filled in by the caregivers on day of the interview with the subject or the days preceding it. There was no difference in the scores when the subjects were taking paroxetine relative to when the subjects were taking placebo (n=9) (F 1,7 =2.25, P=0.177). Verbal fluency and digit span There was no effect of paroxetine relative to placebo on verbal fluency (n=7) (F 1,5 =0.220, P=0.659) or forward or backward digit span (F 1,6 =1.568, P=0.266; F 1,5 =1.149, P=0.333). Discussion Paroxetine did not significantly improve scores on any of the subjective measures taken. Despite using two methods of recording caregiver s assessment of the patient (the NPI is completed during an interview with the caregiver by the interviewer and the CBI is a questionnaire filled in by the caregiver), no improvements were seen. The NPI is well validated for assessment of psychiatric symptoms in dementia, including those particularly prevalent in fvftd such as disinhibition. Even when subscores specific to fvftd were considered, neither trends nor significant differences were observed. These results conflict with the generally positive results from recent studies (Swartz et al. 1997; Moretti et al. 2003); Swartz et al. treated 11 patients with FTD; over half were reported to have improvements in behavioral symptoms such as disinhibition, depressive symptoms, carbohydrate cravings and compulsions. The discrepancy between our results and those of Swartz et al. (1997) and Moretti et al. (2003) could be accounted for by a variety of factors: some patients with advanced FTD may have severe depletion of the 5-HT receptor-containing neurons, removing the substrate for paroxetine s effects. If so, it is possible that SSRIs may only be useful early in the course of this dementia. Moretti et al. (2003) used a somewhat smaller maximum dose of paroxetine (20 mg compared with 40 mg). It is conceivable that a lower dose of paroxetine could give the optimal balance between therapeutic and side effects. The discrepancy could also be due to procedural differences; as in previous studies, raters and patients were aware of the treatment, with no placebo control. Our study included a placebo control; and the patient, caregiver and experimenter were all blind to the order of treatment conditions. Despite the lack of effect of paroxetine on subjective ratings, paroxetine did produce effects on performance on a variety of neuropsychological tests: the reversal component of the visual discrimination task, the errors in the paired-associates learning task and delayed pattern recognition memory. Paroxetine did not, however, improve performance in these tasks; in fact the effect was to increase the number of errors made (an increase in error rates on cognitive tests after paroxetine therapy was also reported by Moretti et al. 2003). These three tasks have been previously shown to be sensitive to another serotonergic manipulation, namely tryptophan depletion (Park et al. 1994; Rogers et al. 1999a, 1999b; Rubinsztein et al. 2001). Spatial recognition memory, spatial working memory and immediate pattern recognition were not significantly affected by paroxetine therapy, although some non-significant trends toward impairment were observed in the ED-shift and spatial recognition memory tasks. The tasks not significantly affected have also been shown to be relatively insensitive to tryptophan depletion (Park et al. 1994). There was only one task sensitive to tryptophan depletion (Rogers et al. 1999b), which was not affected in the same way by paroxetine the decisionmaking task. The small number of subjects involved in the neuropsychological arm of this study may have limited the power to detect changes in the quality of decision making, although there was sufficient power to detect changes in the other tests. Other than the lack of effect of paroxetine on the quality of decision making, the general pattern of deficits is strikingly similar to those seen after tryptophan depletion and does not mimic the pattern of deficits shown in other psychopharmacological manipulations, for example: methylphenidate (Elliott et al. 1997; Rogers et al. 1999a).

41 It is generally assumed that chronically administered SSRIs cause an increase in releasable pools of 5-HT through reuptake inhibition and desensitization of 5-HT autoreceptors, (although this has recently been challenged on the basis of microdialysis studies in the macaque (Smith et al. 2000). While it is possible that any interference with the serotonergic system would disrupt performance on these tasks, if a chronic course of paroxetine increased 5-HT, it would generally have effects opposite to those seen in tryptophan depletion. However, it could be that the course of paroxetine given was not long enough for the desensitization to be complete in those areas of the frontal lobe that have been particularly implicated in reversal learning and delayed pattern recognition. Since it has been suggested that the OFC is critically involved in both reversal learning and delayed pattern recognition (Rogers et al. 1999a; Rubinsztein et al. 2001) and that this is assumed to be a key site of dysfunction in fvftd (Rahman et al. 1999), it is important to attempt to understand the changes effected by paroxetine in this area. Blier and de Montigny (1998) found that the evoked release of tritiated 5-HT in the guinea pig was significantly enhanced in the OFC after an 8-week course of high-dose paroxetine or fluoxetine but not after 3 weeks. Although a 4- to 6-week study period is conventional for testing the efficacy of a treatment for depression, 8 12 weeks is typically used in assessment of SSRI treatment of obsessive compulsive disorder (Montgomery 1996). It is possible that a longer course of paroxetine could produce different results on neuropsychological measures and perhaps even subjective measures. The difficulties in understanding the potential changes in the serotonergic system in different parts of the brains of our subjects after a 6-week course of paroxetine are further complicated by the fact that serotonergic function is disrupted in FTD. It is therefore of interest to know whether similar effects would be observed in normal individuals after a similar course of paroxetine. It is possible that paroxetine is unable to exert an effect on the damaged OFC, and that serotonergic influences on the striatum or other serotonergically innervated regions connected to the OFC produce the changes observed. In summary, the data from this study offer no evidence to support the use of paroxetine as a treatment for fvftd. Paroxetine treatment did not affect abnormal behaviors as rated by caregivers and caused a decrease in accuracy in paired-associates learning, reversal learning and delayed pattern recognition. The pattern of deficits of subjects with fvftd after a 6-week course of paroxetine closely resembles the pattern of deficits observed in normal individuals who have undergone tryptophan depletion. Acknowledgements Funding was provided by a Wellcome Trust programme grant (019407) to Prof. T.W. Robbins, Prof. B.J. Everitt, Dr. A.C. Roberts and Prof. B.J. Sahakian and carried out within a MRC Center for Behavioural and Clinical Neuroscience. J.B.D. was funded by Merck Sharp and Dohme, a James Baird Award from the University of Cambridge School of Clinical Medicine and an Oon Khye Beng studentship. S.R. was funded by the MRC and the James Baird Fund. John Hodges funding was provided by a MRC program grant. The authors would like to thank Dr. S. Bozeat for help with collecting the data, Dr. M. Aitken for statistical advice and Prof. Robbins for his comments on the results. References 407 Bechara A, Damasio H, Damasio AR, Lee GP (1999) Different contributions of the human amygdala and ventromedial prefrontal cortex to decision-making. J Neurosci 19: Benton AL (1968) Differential behavioural effects in frontal lobe disease. Neuropsychologia 6:53 60 Blier P, de Montigny C (1998) Possible serotonergic mechanisms underlying the antidepressant and anti-obsessive-compulsive disorder responses. Biol Psychiatry 44: Blier P, de Montigny C (1999) Serotonin and drug-induced therapeutic responses in major depression, obsessive-compulsive and panic disorders. Neuropsychopharmacology 21:91S 98S Bozeat S, Gregory CA, Ralph MA, Hodges JR (2000) Which neuropsychiatric and behavioural features distinguish frontal and temporal variants of frontotemporal dementia from Alzheimer s disease? J Neurol Neurosurg Psychiatry 69: Cummings JL, Mega M, Gray K, Rosenberg-Thompson S, Carusi DA, Gornbein J (1994) The Neuropsychiatric Inventory: comprehensive assessment of psychopathology in dementia. Neurology 44: Dolan M, Anderson IM, Deakin JF (2001) Relationship between 5- HT function and impulsivity and aggression in male offenders with personality disorders. Br J Psychiatry 178: Downes JJ, Roberts AC, Sahakian BJ, Evenden JL, Morris RG, Robbins TW (1989) Impaired extra-dimensional shift performance in medicated and unmedicated Parkinson s disease: evidence for a specific attentional dysfunction. Neuropsychologia 27: Elliott R, Sahakian BJ, Matthews K, Bannerjea A, Rimmer J, Robbins TW (1997) Effects of methylphenidate on spatial working memory and planning in healthy young adults. Psychopharmacology 131: Fowler KS, Saling MM, Conway EL, Semple JM, Louis WJ (1995) Computerized delayed matching to sample and paired associate performance in the early detection of dementia. Appl Neuropsychol 2:72 78 Gregory CA (1999) Frontal variant of frontotemporal dementia: a cross-sectional and longitudinal study of neuropsychiatric features. Psychol Med 29: Gregory CA, Serra-Mestres J, Hodges JR (1999) Early diagnosis of the frontal variant of frontotemporal dementia: how sensitive are standard neuroimaging and neuropsychologic tests? Neuropsychiatry Neuropsychol Behav Neurol 12: Grundman M, Corey-Bloom J, Thal LJ (1998) Perspectives in clinical Alzheimer s disease research and the development of antidementia drugs. J Neural Transm Suppl 53: Hodges JR, Miller B (2001) The classification, genetics and neuropathology of frontotemporal dementia. Introduction to the special topic papers: part I. Neurocase 7:31 35 Joyce EM, Robbins TW (1991) Frontal lobe function in Korsakoff and non-korsakoff alcoholics: planning and spatial working memory. Neuropsychologia 29: Levy ML, Miller BL, Cummings JL, Fairbanks LA, Craig A (1996) Alzheimer disease and frontotemporal dementias. Behavioral distinctions. Arch Neurol 53: Manes F, Sahakian B, Clark L, Rogers R, Aitken M, Robbins T (2002) Decision-making processes following damage to the prefrontal cortex. Brain 125: Mann JJ (1999) Role of the serotonergic system in the pathogenesis of major depression and suicidal behavior. Neuropsychopharmacology 21:99S 105S

42 408 Mavaddat N, Kirkpatrick PJ, Rogers RD, Sahakian BJ (2000) Deficits in decision-making in patients with aneurysms of the anterior communicating artery. Brain 123: McKeith I, Del Ser T, Spano P, Emre M, Wesnes K, Anand R, Cicin-Sain A, Ferrara R, Spiegel R (2000) Efficacy of rivastigmine in dementia with Lewy bodies: a randomised, double-blind, placebo-controlled international study. Lancet 356: Milner B (1964) Effects of different brain lesions on card sorting. Arch Neurol 9: Montgomery SA (1996) Long-term management of obsessivecompulsive disorder. Int Clin Psychopharmacol 11[Suppl 5]:23 29 Moretti R, Torre P, Antonello RM, Cazzato G, Bava A (2003) Frontotemporal dementia: paroxetine as a possible treatment of behavior symptoms. A randomized, controlled, open 14-month study. Eur Neurol 49:13 19 Neary D, Snowden JS, Gustafson L, Passant U, Stuss D, Black S, Freedman M, Kertesz A, Robert PH, Albert M, Boone K, Miller BL, Cummings J, Benson DF (1998) Frontotemporal lobar degeneration: a consensus on clinical diagnostic criteria. Neurology 51: Neary D, Snowden JS, Mann DM (2000) Classification and description of frontotemporal dementias. Ann New York Acad Sci 920:46 51 Orrell MW, Sahakian BJ (1991) Dementia of frontal lobe type (editorial). Psychol Med 21: Owen AM, Downes JJ, Sahakian BJ, Polkey CE, Robbins TW (1990) Planning and spatial working memory following frontal lobe lesions in man. Neuropsychologia 28: Park SB, Coull JT, McShane RH, Young AH, Sahakian BJ, Robbins TW, Cowen PJ (1994) Tryptophan depletion in normal volunteers produces selective impairments in learning and memory. Neuropharmacology 33: Perry RJ, Hodges JR (1999) Attention and executive deficits in Alzheimer s disease. A critical review. Brain 122: Plaisted KC, Sahakian BJ (1997) Dementia of the frontal lobe typeliving in the here and now. Aging Mental Health 1: Rahman S, Sahakian BJ, Hodges JR, Rogers RD, Robbins TW (1999) Specific cognitive deficits in mild frontal variant frontotemporal dementia. Brain 122: Ratnavalli E, Brayne C, Dawson K, Hodges JR (2002) The prevalence of frontotemporal dementia. Neurology 58: Robbins TW (2000) Chemical neuromodulation of frontal-executive functions in humans and other animals. Exp Brain Res 133: Rogers RD, Robbins TW (2001) Investigating the neurocognitive deficits associated with chronic drug misuse. Curr Opin Neurobiol 11: Rogers RD, Blackshaw AJ, Middleton HC, Matthews K, Hawtin K, Crowley C, Hopwood A, Wallace C, Deakin JF, Sahakian BJ, Robbins TW (1999a) Tryptophan depletion impairs stimulusreward learning while methylphenidate disrupts attentional control in healthy young adults: implications for the monoaminergic basis of impulsive behaviour. Psychopharmacology 146: Rogers RD, Everitt BJ, Baldacchino A, Blackshaw AJ, Swainson R, Wynne K, Baker NB, Hunter J, Carthy T, Booker E, London M, Deakin JF, Sahakian BJ, Robbins TW (1999b) Dissociable deficits in the decision-making cognition of chronic amphetamine abusers, opiate abusers, patients with focal damage to prefrontal cortex, and tryptophan-depleted normal volunteers: evidence for monoaminergic mechanisms. Neuropsychopharmacology 20: Rubinsztein JS, Rogers RD, Riedel WJ, Mehta MA, Robbins TW, Sahakian BJ (2001) Acute dietary tryptophan depletion impairs maintenance of affective set and delayed visual recognition in healthy volunteers. Psychopharmacology 154: Sahakian BJ, Morris RG, Evenden JL, Heald A, Levy R, Philpot M, Robbins TW (1988) A comparative study of visuospatial memory and learning in Alzheimer-type dementia and Parkinson s disease. Brain 111: Smith TD, Kuczenski R, George-Friedman K, Malley JD, Foote SL (2000) In vivo microdialysis assessment of extracellular serotonin and dopamine levels in awake monkeys during sustained fluoxetine administration. Synapse 38: Sparks DL, Markesbery WR (1991) Altered serotonergic and cholinergic synaptic markers in Pick s disease. Arch Neurol 48: Swartz JR, Miller BL, Lesser IM, Darby AL (1997) Frontotemporal dementia: treatment response to serotonin selective reuptake inhibitors. J Clin Psychiatry 58:

43

44 Dement Geriatr Cogn Disord 1999;10(suppl 1):15 28 Comparative Cognitive Neuropsychological Studies of Frontal Lobe Function: Implications for Therapeutic Strategies in Frontal Variant Frontotemporal Dementia S. Rahman T.W. Robbins B.J.Sahakian Departments of Psychiatry and Experimental Psychology, University of Cambridge, UK Key Words Comparative studies W Dopamine W Dorsolateral W Frontotemporal dementia W Idazoxan W Noradrenaline W Orbitofrontal W Prefrontal W Serotonin W Therapeutic strategies Abstract Patients with mild frontal variant frontotemporal dementia (fvftd) who attend the clinic are usually unaware of the pervasive changes in their personality and behaviour, despite the fact it is these changes which have prompted the referral from the patient s spouse or carer. Comparative studies across various species offer unique insights into the heterogeneous structure and functions of the prefrontal cortex, and can allow a novel approach to the precise identification of the neuropsychological deficits present in these patients. We have found that they may show marked deficits on tests sensitive to ventromedial prefrontal or orbitofrontal function, in the relative absence of impairments on tests sensitive to dorsolateral prefrontal function. We highlight important differences in the neurocognitive profile of these patients with that of patients with other neurodegenerative conditions, including basal ganglia diseases and dementia of the Alzheimer type. The specific nature of these neuropsychological deficits, together with converging evidence from clinical and neuropathological studies, may provide useful clues about the predominant locus of dysfunction in the early stages of fvftd and possible underlying neurotransmitter abnormalities. This is important for the successful development of therapeutic intervention strategies for both cognitive and behavioural symptoms in fvftd. Finally, we evaluate critically the rationales for therapeutic modulation of noradrenergic, serotonergic and dopaminergic neurotransmitter systems at various stages of disease. Introduction Two key issues concerning frontal variant frontotemporal dementia (fvftd) include the potential contribution of cognitive neuropsychology to the diagnosis of this clinical disorder, and its differential diagnosis from other neurodegenerative diseases (in particular, dementia of the Alzheimer type, DAT). Another major consideration has been the need to identify any neurotransmitter systems that might be compromised early in disease, and the implications of this for the successful development of rational strategies for the treatment of behavioural and ABC Fax karger@karger.ch S. Karger AG, Basel /99/ $17.50/0 Accessible online at: Dr. B.J. Sahakian Department of Psychiatry, University of Cambridge School of Clinical Medicine Addenbrooke s Hospital, Hills Road Cambridge CB2 2QQ (UK) Tel , Fax , bjs1001@cam.ac.uk

45 cognitive symptoms. This particular issue shall be the focus of the latter part of this article. To address the general question of differential diagnosis, and in particular the discrimination from DAT, it is useful to note that fvftd affects younger patients, with its peak onset in the presenium [1]. Patients with fvftd present clinically with early changes in personality and loss of social awareness [2]. Clinically, the majority of patients with fvftd who attend the clinic are usually unaware of the pervasive changes in their personality and behaviour, despite the fact it is these changes which have prompted the referral from the patient s spouse or carer. Patients may appear apathetic or withdrawn, or alternatively may become socially disinhibited with facetiousness and inappropriate jocularity. Mental rigidity and an inability to appreciate the subtler aspects of language such as irony are common. Stereotypical or ritualistic behaviour are also common. There is often indifference to domestic and occupational responsibilities, a lack of empathy for family and friends, and a gradual withdrawal from all social interactions. Many of these behavioural symptoms (such as the loss of personal awareness, hyperorality, stereotyped and perseverative behaviour, a progressive reduction of speech and preserved spatial orientation) may all be useful in differentiating patients with fvftd from those with DAT [3]. This observation is particularly useful in the context of a previous study which had revealed that whilst personality change, unconcern and socially inappropriate behaviour in frontotemporal dementia were prominent in fvftd, disturbance in memory and topographical orientation tended to be prominent in patients with DAT [4]. As in other dementias, neuroimaging may, in addition, contribute significantly to the diagnosis. In one HMPAO- SPECT study [5], blood flow was found to be significantly lower in the frontal lobes, anterior temporal cortex, and the basal ganglia, in patients with frontal lobe dementia, compared to patients with DAT. Within the frontal lobe dementia group, blood flow was significantly lower in the orbital than in the dorsal frontal cortex, and in the anterior temporal than in the dorsal temporal cortex. In considering specifically the potential role of cognitive neuropsychology to the diagnosis of fvftd, it is important to recognize that fundamental to our understanding of the prefrontal cortex have been exquisite neuroanatomical studies which have revealed that there appears to be an overall correspondence among all constituent areas in the brains of humans and monkeys [6]. Also important has been the recognition that the prefrontal cortex has developed substantially in primates through evolution [7], and this may have, as a consequence, resulted in the prefrontal cortex in both monkeys and in man being heterogeneous in its functions and its distinct patterns of cortical and subcortical structural connectivity. On the basis of anatomical and neurophysiological evidence, the concept of parallel and segregated cortico-striato-pallidal-thalamocortical circuit loops has evolved which emphasizes the functional interrelationships between the neocortex and striatum [8, 9]. In humans, the functional roles of various frontal-subcortical loops have been successfully identified, on the basis of clinical syndromes and neuroimaging studies. Sophisticated methods for neuropsychological assessment, as well as a comprehensive understanding of functional neuroanatomy, may therefore allow considered appraisal of the nature of cognitive deficits in frontotemporal dementia. Whilst there are clear methodological considerations (for example, the dementing process may extend beyond a particular focus and hence cognitive deficits may relate to dysfunction of an extended neural network), it is still important that attempts should be made to isolate the anatomical region involved to direct the selection of further cognitive and behavioural investigations [10]. The importance of the consequences of lesions of the frontal lobe in terms of changes in behaviour and personality in animals was described elegantly by Ferrier [11] in Animals with bilateral frontal lobe ablations were described as hyperactive, impulsive, and lacking affection and socialization; apathy and inattention were also emphasized by Ferrier. Parallels in the human literature have been subsequently found, with Blumer and Benson [12] in 1975 describing patients with frontal lobe lesions as pseudodepressed (patients appearing slow and lacking in initiative or concern), or pseudopathic (patients demonstrating restlessness and impulsivity). Our understanding of the effects of lesions of distinct areas of the frontal cortex may also be clarified further within the context of the concept of the frontal-subcortical loops described earlier. Two circuits which may be hypothesized to be affected by the pathological process include the dorsolateral prefrontal circuit (which includes the dorsolateral convexities of the prefrontal cortex and the dorsolateral portion of the caudate), and also the lateral orbitofrontal circuit (which includes the orbitofrontal cortex and the ventromedial portion of the caudate nucleus). To identify the major domains of neuropsychological deficit in patients with mild fvftd and the putative underlying locus of pathology, it is useful to note that the patients in the early stages of fvftd clinically are somewhat similar to those with defined neurosurgical lesions of 16 Dement Geriatr Cogn Disord 1999;10(suppl 1):15 28 Rahman/Robbins/Sahakian

46 the ventromedial prefrontal cortex. Damasio and coworkers [13] have described how, in general, patients with ventromedial frontal lobe lesions exhibit an inability to initiate, organize, and carry out normal activities, and also demonstrate poor decision-making, financial mistakes, breakup of relationships, perseverative activities, and decreased spontaneity. Patients in the early stages of fvftd also bear some resemblance to other patient groups with dysfunction principally of the orbitofrontalsubcortical circuit, for example those with ruptured anterior communicating aneurysms, orbitofrontal tumours, and inferior frontal lobe infarction [14 16]. In contrast, patients with dorsolateral prefrontal-subcortical dysfunction are predominantly characterized by impairments in executive function [17]. Major insights into the functions of the orbitofrontal cortex have been provided by the studies of Bechara et al. [18]. They have found that patients with damage to the orbitofrontal cortex are very prone to make bad decisions in gambling games because, hypothetically, they do not develop a gut feeling indexed for example by changes in GSR. Recently, Bechara et al. [18] have made a significant contribution to advancing our understanding of dissociations of prefrontal cortical function. In their study, all subjects with ventromedial lesions were impaired on their gambling task, whereas only a subset of these subjects with the most anteriorly placed lesions were normal on the working memory tasks; however, subjects with right dorsolateral/high mesial lesions were impaired on the working memory tasks but not on the gambling task. Recent results from the primate literature also detail other cognitive deficits consequent upon orbitofrontal damage. Dias et al. [19, 20] have reported a double dissociation in the prefrontal cortex of affective and attentional shifts in the marmoset following excitotoxic lesions to the lateral and orbitofrontal cortex. Whereas damage to the lateral prefrontal cortex in monkeys causes a loss of inhibitory control in attentional selection (an extradimensional shift), damage to the orbitofrontal cortex in monkeys causes a loss of inhibitory control in affective processing (a reversal), thereby impairing the ability to alter behaviour in response to fluctuations in the emotional significance of stimuli. Neither lesion affected learning at the intradimensional stage. In this paper, we shall describe the neurocognitive deficits in patients with mild fvftd that we have found, and outline differences with other neurodegenerative conditions (such as DAT and basal ganglia disorders) and patients with focal neurosurgical lesions (in particular, frontal and temporal lobe excisions). In the neuropsychological assessment of frontal lobe function in fvftd, various tests may be applied to demonstrate that these patients show distinct deficits in comparison to other neurodegenerative conditions. Therefore, we shall describe the performance of these patients and other patient groups on computerized neuropsychological assessment, which includes tests from the Cambridge Neuropsychological Test Automated Battery (CANTAB): (a) pattern recognition; (b) spatial recognition; (c) a planning task (the one-touch Tower of London ); (d) a spatial working memory task; (e) an attentional set-shifting task, and (f) a decision-making task. These tests allow detailed automatic recording of response accuracy and speed via a touchscreen apparatus, and have been designed to reflect theoretical developments in the cognitive neuroscience of memory, attention, and executive functions [21]. This particular type of neuropsychological assessment also has the further advantages of facilitating cross-species comparisons between rats, primates, and humans, being also sensitive to a variety of brain pathologies and defined pharmacological challenges. Neuropsychological Assessment Pattern and Spatial Recognition The CANTAB contains a test of visual pattern recognition memory using a serial recognition task analogous to other paradigms to test visual memory for sets of objects in nonhuman primates [22, 23], and a complementary test of visuospatial recognition memory. These tests have now been used in a variety of neurodegenerative diseases, and in patients with selective neurosurgical lesions. Performance on the pattern recognition test has been shown to be impaired in patients with temporal lobe excisions and amygdalo-hippocampectomy, whilst the spatial recognition test is sensitive to frontal lobe lesions [24]. The use of visual, nonverbal material enables comparisons to be made with the extensive nonhuman literature defining the neural and neurochemical substrates of visual learning and memory. The recognition tests appear extremely sensitive to the cognitive deficits in probable dementia of the Alzheimer type (pdat), patients with both mild and moderate pdat being impaired on pattern and visuospatial recognition [25 27], presumably reflecting early temporal lobe pathology seen in DAT [28]. In a cross-sectional study of patients with Parkinson s disease (PD) classified as either nonmedicated, medicated with mild disability, or medicated with severe disability [29], it was found that no Cognition and Drug Therapy in Frontotemporal Dementia Dement Geriatr Cogn Disord 1999;10(suppl 1):

47 a b Fig. 1. Two tests of cognition to assess fronto-subcortical function. a One-touch Tower of London the subject is asked to work out the minimum number of moves that would be required to achieve the goal arrangement presented in the upper half of the screen from the arrangement in the lower half, and to make their response by pressing one of several choices offered on the screen. b Spatial working memory a token can be revealed by touching the boxes: it will not be in the same box twice (see text for details). patient group was impaired on the test of pattern recognition memory, and only the group with severe clinical disability were impaired on spatial recognition. In a study of patients with mild-moderate pdat and patients with Huntington s disease (HD) matched for severity of dementia [30], both groups were impaired on pattern and spatial recognition, with the HD group performing significantly worse on both tests. In contrast to other patients with neurodegenerative conditions, patients with fvftd appear unimpaired on the pattern and spatial recognition task. Tower of London The Tower of London task was modified by Shallice [31] from the earlier Tower of Hanoi task to stress the capacity for mental planning with only a single goal outcome specified. In one form of the task, after considering the starting and goal configurations of a set of coloured balls on the screen, the subject maneuvres these balls between the locations simply by touching them and their desired destinations. However, this nature of the task does not avoid completely the on-line editing problems noted by Goel and Grafman [32], as subjects may have to stop to think during performance of the sequence. In a variant of this task (the one-touch Tower of London ), the subject is asked to consider an arrangement of coloured balls hanging in socks or pockets to match a goal arrangement presented in the top half of the screen. Subjects are then simply asked to work out the minimum number of moves that would be required to achieve the goal arrangement, and to make their response by pressing one of several choices offered on the screen. This task provides a purer measure of the look ahead function in planning and its important memory component. A typical appearance of the computer screen during this task is shown in figure 1a. Using the one-touch Tower of London, patients with fvftd are found to be unimpaired compared to controls in the accuracy of their solutions. A possible explanation for this may be derived from what is known about the precise neural substrates underlying successful performance in this task. In a recent positron emission tomography (PET) neuroimaging study [33] using the one-touch Tower of London, regional increases in regional cerebral blood flow (rcbf) were produced in a distributed cortical network that included the superior occipito-parietal cortex and three main zones in the frontal cortex: the premotor area, a band of activation including the dorsolateral prefrontal cortex bilaterally and the frontopolar cortex on the right. Patients with mild fvftd may therefore be relatively unimpaired in the one-touch Tower of London task because, at that stage of disease, the site of pathology does not predominantly involve the dorsolateral prefrontal cortex. Spatial Working Memory Primates and humans have developed the capacity to process information on line, a capacity that is considered to underlie comprehension, thinking, and executive functions [34]. The spatial working memory task is essentially a modification of one used to examine the effects of dorsolateral prefrontal cortex lesions in primates [35], and is conceptually similar to the radial arm maze which has been successfully used to assess the role of the hippocampus in working memory in rats [36]. The test is openended in the sense that the subject is free to produce his or her own self-ordered sequences of responses. 18 Dement Geriatr Cogn Disord 1999;10(suppl 1):15 28 Rahman/Robbins/Sahakian

48 Fig. 2. Performance on the spatial working memory test, as measured by between search errors. Mild fvftd, mild frontotemporal dementia; Frontal, frontal lobe excision patients; Temporal, temporal lobe excision patients; PD, Parkinson s disease; MSA, multiple system atrophy (of striatonigral predominance); SRO, Steele-Richardson-Olzsewski syndrome; HD, Huntington s disease. The HD group were relatively late in the course of the disease [30; data taken from 24, 37, 38]. Control groups (control) were matched by age and premorbid intelligence with their respective patient groups Mild fvftd Mild fvftd (control) Frontal Frontal (control) Temporal PD (medicated severe) PD (medicated mild) PD (nonmedicated) PD (control) MSA MSA (control) SRO SRO (control) HD HD (control) In our version of the task, adapted for humans, subjects are required to search through a number of red boxes presented on the computer screen (by touching each one) in order to find blue tokens that were hidden inside. Figure 1b shows the typical appearance of a screen from this task. The object of the task is to avoid those boxes in which a token had already been found. Importantly, the subjects are able to search through the boxes in any order that they wish, although the number of boxes visited before a token was found was determined by the computer. As can be seen in figure 2, neurosurgical patients with frontal lobe damage are significantly impaired on this task, making more returns to boxes ( between search errors) in which a token had previously been found, at all levels of task difficulty as determined by the number of boxes employed (2, 3, 4, 6, or 8). This task may also be sensitive to deficits in patients with temporal lobe damage, although only at the most extreme level of task difficulty (i.e. 8 boxes); the task has also proved sensitive to deficits in medicated PD patients with both mild and severe clinical symptoms [37], HD patients [31], and also in patients with multiple system atrophy and Steele- Richardson-Olszewski syndrome [38]. Interestingly, we found that patients with fvftd, even with their presumed frontal lobe pathology, are also relatively unimpaired on the test of spatial working memory. Again, a possible reason for this may be derived from what is known about the precise neural substrates underlying successful performance in the task. In recent PET studies, components of working memory have been investigated using a close analogue of the spatial working memory task [39]. These studies indicated two major areas of activation in the dorsolateral and ventrolateral prefrontal cortex. Further data suggest that the ventrolateral area may play a role in the passive receipt of items for memory (as in spatial span performance), whereas the dorsolateral prefrontal cortex is important for the monitoring role by which candidate sequences are compared with the goal sequence. A recent meta-analysis of functional neu- Cognition and Drug Therapy in Frontotemporal Dementia Dement Geriatr Cogn Disord 1999;10(suppl 1):

49 1 2 Simple discrimination and reversal Compound discrimination (separate) Compound discrimination and reversal problem of this type by which possible solutions are considered and either rejected or accepted [41]. Furthermore, the lack of impairment in the spatial working memory task, in addition to the relative lack of impairment in the one-touch Tower of London task, lends further support to the hypothesis that the dorsolateral prefrontal cortex is not predominantly affected in the mild stages of fvftd. We now examine the evidence for the hypothesis that a major locus of dysfunction relatively early in the course of fvftd is, rather, the ventromedial prefrontal (or orbitofrontal) cortex Intradimensional shift and reversal Extradimensional shift and reversal Fig. 3. Stages of the attentional set-shifting task. Subjects have to choose between the two stimuli to receive positive feedback according to a learned rule. When they reach a criterion of six consecutively correct answers at each stage, they move onto the next stage unless 50 trials have occurred in which case the test is terminated. Reversal refers to a reversal of the contingencies, such that the correct and incorrect stimuli are unpredictably swapped. The stimuli are presented randomly with respect to spatial location. 1: Simple discrimination between shapes; 2 and 3: compound discrimination, in which the white lines are introduced as distractors; 4 and 5: dimensional shifts are implemented with unexpected changes to novel exemplars of both shape and line; with intradimensional shift, shape remains relevant (one of the new shapes is correct, regardless of its pairing with line); with extradimensional shift, line becomes relevant (one of the lines is correct, regardless of its pairing with shape). roimaging studies indeed suggests that the specific regions of the lateral prefrontal cortex (dorsolateral or ventrolateral) prefrontal cortex make identical contributions to both spatial and nonspatial working memory [40]. The observation that these patients with fvftd are unimpaired on the spatial working memory task is useful since, arguably, successful planning in the Tower of London task places a significant load on spatial working memory not only in terms of the storage of a correct sequence but also in the search processes required in any analytical Attentional Set-Shifting The Wisconsin Card Sorting Test (WCST) is the classical test of cognitive set-shifting. However, it is a deceptively complex task which may be solved using a number of different approaches. We have devised a computerized test of attentional set-shifting ability which deconstructs the WCST into its constituent elements. A detailed account of this attentional set-shifting paradigm has been described elsewhere [42, 43], but, in brief, the paradigm is based on studies of intradimensional and extradimensional shifts used to examine attentional set in humans [44] and other animals [45]. In this paradigm, initially the subject is required to learn a series of discriminations in which one of two stimuli was correct and the other was not, using feedback provided automatically by the computer. The test is composed of nine stages presented in the same fixed order, represented pictorially in figure 3. The test begins with a simple discrimination (sd) and reversal (sd_r) for stimuli varying in one dimension (i.e. two purple filled shapes). A second, alternative dimension is then introduced (white line configurations) and compound discrimination (cd) and reversal (cd_r) were tested. To succeed, subjects have to continue to respond to the previously relevant stimuli (i.e. purple shapes), ignoring the presence of the new, irrelevant dimension (lines). At the intradimensional shift (id) stage, new exemplars are introduced from each of the two dimensions (new lines and new shapes) and subjects are required to transfer the previously learned rule to a novel set of exemplars from the same stimulus dimension. Thus, to succeed, they have to continue to respond to one of the two exemplars from the previously relevant dimension (shapes). Following another reversal of contingencies (id_r), the extradimensional shift (ed) and reversal (ed_r) are presented and again, novel exemplars from each of the two dimensions are introduced. However, at this stage, the subject is required to shift response set to the alternative (previously irrelevant) stimulus dimension and ig- 20 Dement Geriatr Cogn Disord 1999;10(suppl 1):15 28 Rahman/Robbins/Sahakian

50 nore the previously relevant dimension. At each stage, a change in contingencies only occurs once the subject has learnt the current rule to a criterion of six consecutive correct responses. The subject is only allowed to proceed to each successive stage of the test if he or she reached criterion at the previous stage. This permits a clear and simple method of analysing and presenting the main results. The proportion of subjects reaching criterion at each stage of the CANTAB attentional set-shifting paradigm is shown in figure 4. Patients with neurosurgical excisions of the frontal lobes are specifically impaired in their ability to shift response set to the previously irrelevant stimulus dimension (i.e. at the ed stage of learning), but not to shift attention to new exemplars of a previously relevant dimension (i.e. at the id stage of learning) [46]. The ed shift is a core component of the WCST, and the pattern of deficits reported in patients with frontal lobe damage on this test may be partly explained in these terms [47]. By comparison, patients with temporal lobe excisions and amygdalo-hippocampectomy patients have been found to be unimpaired in their ability to perform either shift [46], supporting the anatomical specificity of the ed shift. This observation concurred with previous reports that patients with unilateral or bilateral temporal lobe damage and a patient with bilateral hippocampal damage were unimpaired on the WCST [48, 49]. The ed shift stage has also been shown to pose particular difficulties for medicated and nonmedicated PD patients [43] and in patients in the preclinical and early clinical stages of HD [50, 51]. Figure 5 shows the number of errors made at the id and ed stages in the attentional set-shifting paradigm, for subjects who attempt both of these stages. Patients with frontal lobe excisions generally make more errors at the ed shift stage, compared to the id shift stage, and this is borne out as a significant difference in the patients with frontal lobe damage requiring more trials to criterion specifically at the ed shift stage. Figure 5 also demonstrates graphically the lack of deficit at the ed shift stage for temporal lobe excision patients [41] and for patients with DAT [after 52]. It is particularly interesting to observe that patients with mild pdat, who exhibit significant memory deficits, are no worse than age- and IQ-matched controls at negotiating the attentional set-shifting test [52]. This is a very significant finding, as it implies that the deficits in extradimensional shifting are found patients with basal ganglia disorders. In contrast, the deficits present in early DAT resemble those produced by posterior cortical, especially temporal lobe damage. A key observation regarding patients with mild fvftd is that, compared to matched controls, they are unimpaired on the ed s s s s s Controls Mild pdat Mild fvftd FLE s TLE nmedpd PSP HD(mod) s_d s_r cp_d cd c_r Stage id id_r ed ed_r Fig. 4. Performance on the attentional set-shifting paradigm, assessed in terms of the proportion of subjects reaching each stage of the test by producing six consecutively correct responses. Stages: s_d, simple discrimination; s_r, simple reversal; cp_d, compound discrimination, spatially discontiguous elements; cd, compound discrimination; c_r, compound discrimination reversal; id, intradimensional shift; id_r, intradimensional reversal; ed, extradimensional shift; ed_r, extradimensional reversal. Key as shown [data taken from 30, 37, 38, 46, 52]. shift, both in terms of the number of errors made compared to the id shift and also in terms of the proportion of subjects reaching criterion. However, if the total number of errors on the nonreversal stages (cd+id+ed) are compared with the total number of errors on the corresponding reversal stages (cd_r+ id_r+ed_r), patients with fvftd are found to show a deficit specific to the reversal stages [U = 8, W = 92.0, P = ; significant as p! ]. In preliminary analysis, including only subjects who attempt all stages of the task, this deficit has not been demonstrated in other patient groups (see fig. 6) (FLE: U = 71.5, W = 193, P = 0.216; TLE: U = 114, W = 400, P = 0.080; DAT: U = 4, W = 22, P = 0.248; all n.s. as p ). The specific nature of the reversal deficit in patients with fvftd is shown clearly in figure 6a d. As the extradimensional shift is known to be more difficult than reversal shifting for normal subjects [42], it is unlikely that the deficit in reversal shifting (or indeed the above-mentioned deficit in extradimensional s s s s s s s Cognition and Drug Therapy in Frontotemporal Dementia Dement Geriatr Cogn Disord 1999;10(suppl 1):

51 Fig. 5. Errors at the intradimensional (id) and extradimensional (ed) shift stages of the attentional set-shifting paradigm. Panels show errors for (a) frontal variant frontotemporal dementia (fvftd); (b) frontal lobe excision (FLE); (c) temporal lobe excision (TLE) and (d) mild dementia of the Alzheimer type (DAT) patients. Control groups (Controls) were matched by age and premorbid intelligence with their respective patient groups [data from 41, 52]. Errors Errors fvftd Controls id Stage TLE Controls id Stage ed ed a c Errors Errors FLE Controls id Stage DAT Controls id Stage ed ed b d fvftd Controls FLE Controls Errors 15 Errors Nonreversals Stage Reversals a 0 Nonreversals Stage Reversals b Fig. 6. Cumulative errors at the nonreversal (cd+id+ed) and their corresponding reversal stages (cd_r+id_r+ed_r) of the attentional set-shifting paradigm. Panels show errors for (a) frontal variant frontotemporal dementia (fvftd); (b) frontal lobe excision (FLE); (c) temporal lobe excision (TLE) and (d) mild dementia of the Alzheimer type (DAT) patients. Control groups (Controls) were matched by age and premorbid intelligence with their respective patient groups [data from 41, 52]. Errors TLE Controls Nonreversals Stage Reversals c Errors DAT Controls Nonreversals Stage Reversals d 22 Dement Geriatr Cogn Disord 1999;10(suppl 1):15 28 Rahman/Robbins/Sahakian

52 shifting) is simply due to differences in task sensitivity. A recent PET imaging study using an adapted version of the attentional set-shifting paradigm revealed that, compared to the intradimensional shift scans, the extradimensional shift activated the frontal pole on the left side and areas 9/46 on the right, whereas reversal learning engaged neural circuitry associated with the ventromedial prefrontal cortex (e.g. the ventral caudate) [53]. In relation to this work and the nonhuman primate studies by Dias et al. [19, 20] described earlier, the specific reversal learning deficit of patients with fvftd early in the course of the disease is consistent with the hypothesis that the pathology predominantly affects the orbitofrontal cortex or its connections, rather than the dorsolateral prefrontal cortex. One intriguing alternative is that the deficits in reversal learning may reflect pathology affecting the caudate nucleus. Divac et al. [54] originally showed in primates that a lesion involving the ventrocaudal neostriatum impaired visual discrimination reversal. The hypothesis has therefore emerged that the ventrocaudal striatum is a critical link in the stimulus response, or habit learning circuit [55]. Neurones which reflect the responses of orbitofrontal neurones considered are found in the ventral head of the caudate nucleus and the ventral striatum, which receive input from the orbitofrontal cortex [56]. The behaviour of patients with fvftd may also be in keeping with the few reports of the effects of focal caudate lesions. Mendez et al. [57] reported that patients with dorsal caudate lesions exhibited confused and disinterested behaviour, in contrast to patients with ventral caudate lesions who exhibited more disinhibited, euphoric and inappropriate behaviour. The existence of severe atrophy of the caudate in frontotemporal degenerative dementia has long been recognized [58], and one may speculate that it is not merely coincidental that neurones in the orbitofrontal cortex and the ventral head of the caudate may be affected, given the nature of the fronto-subcortical loops. In particular, neuronal degeneration within the orbitofrontal cortex might affect its predominant efferent connections as elucidated in the primate literature [59], for example the temporal lobe and the lateral hypothalamus. Decision-Making Although patients with mild fvftd appear relatively unimpaired on certain probes of dorsolateral prefrontal function (spatial working memory and the one-touch Tower of London planning task), these patients do show deficits in an aspect of the attentional set-shifting task sensitive to orbitofrontal or ventromedial function. Another test which can be used to examine cognition in patients with fvftd is a decision-making task [described in detail in 60], which has previously been found to be sensitive to ventromedial or orbitofrontal function, rather than dorsolateral prefrontal function. The three principal measures in this task are: (a) the speed of decision-making, i.e. how long it takes the subject to decide which colour of box is hiding the token as measured by the mean deliberation time; (b) the quality of decisions: as an ineffective approach is to bet continuously on the least likely of the two possible outcomes, one measure was how much of the time the subject chose the most likely outcome, and (c) risk adjustment, i.e. the rate at which a subject increases the percentage of the available points bet in response to more favourable ratios of red:blue boxes (e.g. 9 red:1 blue vs. 4 red:6 blue). The differences between the patients with mild fvftd and their age- and IQmatched controls in these three measures are shown in figure 7a c. Using this decision-making task, we have shown that patients with fvftd are different from their age- and IQmatched controls in that they are willing to bet a much higher proportion of their accumulated reward at all ratios. Furthermore, they show no significant difference from controls in their tendency to choose the most likely outcome, and exhibit increased deliberation times. Therefore, to summarize, although patients with fvftd may make accurate probability judgements, they are unable to adjust the levels of their bets accordingly. As they are not merely impulsive in making their decisions, they appear to be true risk-takers. There are broad similarities in the performance of the patients with fvftd and the patients with ventromedial lesions studied by Damasio and coworkers [18, 61]. Both groups of patients make abnormal decisions that are no longer personally advantageous, and in particular have difficulty planning their future over immediate, medium and long ranges. In general, neither the group described by Damasio and co-workers, nor our patient group, is able to adjust their behaviour successfully appropriately to the task opportunities available. One explanation that has been proposed for the poor performance in the gambling task of patients with ventromedial lesions described by Bechara et al. [62] is that there exists a failure to anticipate future outcomes, and this may also be a possible reason for the poor performance of the patients with fvftd in the decision-making task. Of added interest is the observation that similar increases in deliberation times in the decision-making task may be produced in those normal volunteers with acutely reduced tryptophan and reduced 5-HT function who demonstrate the most severe deficits in decision-making [60]. Cognition and Drug Therapy in Frontotemporal Dementia Dement Geriatr Cogn Disord 1999;10(suppl 1):

53 100 8,000 % bets Deliberation times (ms) 6,000 4,000 2,000 0 Patients Controls a 0 Patients Controls b 100 Fig. 7. Decision-making task. (a) % bets; (b) deliberation times (ms) and (c)% choice of most likely outcome, shown for mild frontal variant frontotemporal dementia patients and control subjects, matched for age and premorbid intelligence level. % choice of most likely outcome Patients Controls c In Summary: Patients with mild fvftd appear unimpaired on neuropsychological assessment on tests of visuospatial mnemonic function (pattern and spatial recognition), which are known to be sensitive to deficits in patients with mild DAT. They may show most marked deficits on tests sensitive to ventromedial or orbitofrontal function (for example, the decision-making task and the reversal stages of the attentional set-shifting task), and be relatively unimpaired on tests sensitive to dorsolateral prefrontal function (for example, the spatial working memory task and the one-touch Tower of London ). These neuropsychological findings may bear some relevance to the behavioural presentation of these patients in the clinic. Certain behavioural symptoms are in particular associated with dysfunction of the orbitofrontal-subcortical circuit, for example disinhibited behaviour, obsessive and compulsive behaviour, and changes in eating behaviour [63]. The presence of these same symptoms in fvftd, disinhibition [64], compulsive behaviours (in particular repetitive checking activities) [65], and increased sweet and carbohydrate preference [66], is therefore of considerable theoretical interest. Therefore, specific deficits found upon formal neuropsychological assessment can potentially provide important clues as to the neural networks that are compromised in disease, and therefore the neurotransmitter systems involved. This, in the longer term, may help to identify any neurotransmitter systems principally affected by disease, and lead to developments in therapeutic strategies. Therapeutic Strategies There is currently a need for full-scale studies of the effects of putative cognitive enhancers in fvftd, in stark contrast to the magnitude of trials currently being carried out with DAT patients. The evidence implicating cholinergic neurones in the decline in memory and cognition in ageing and in DAT has been reviewed elsewhere [67]. However, neurochemical studies concerning fvftd are rather limited. Francis et al. [68] reported normal levels of noradrenaline (NA), serotonin (5-HT) and dopamine (DA) metabolites in a patient with fvftd, but low release values of NA (with high release values of 5-HT and DA from autopsy tissue). Coull et al. [69] therefore considered one method of increasing noradrenergic activity by giving patients a moderate dose of an 2 antagonist, idazoxan (IDZ), which acts presynaptically to elevate synaptic con- 24 Dement Geriatr Cogn Disord 1999;10(suppl 1):15 28 Rahman/Robbins/Sahakian

54 centrations of NA [70]. They found that IDZ produced dose-dependent improvements in performance, particularly on tests of planning, sustained attention, verbal fluency and episodic memory. In contrast, IDZ produced deficits in performance on a test of spatial working memory, which we have previously described as being sensitive to dorsolateral prefrontal function. The results also indicated that if patients were impaired on the working memory task, the deficit might be further exacerbated by IDZ. It is of particular interest, as can be seen from figure 2, that patients with fvftd may be relatively unimpaired on the spatial working memory task. This may have some relevance to the detrimental effect of IDZ in fvftd patients on this task, in that IDZ may somehow have an overdose effect upon performance, and highlights further the neural specificity of the deficits in fvftd. It is notable that IDZ has also recently been found to produce deficits in spatial working memory in patients with pdat [71]. It may be of some relevance that in aged monkeys with naturally occurring catecholamine depletion, 2 adrenergic agonists such as guanfacine have repeatedly been shown to improve dorsolateral prefrontal cortical function assessed using a spatial delayed response task [72]. The cognitive-enhancing properties of these agonists may be considered to be a consequence of their action at postjunctional noradrenergic receptors. Also, in aged monkeys, improvements in performance on the delayed response task induced by a variety of 2 agonists can be blocked by 2 antagonists such as yohimbine and IDZ, but not by the 1 antagonist, prazosin [73]. However, the detrimental effects of IDZ upon spatial working memory may also be explained in terms of recent developments in elucidating how IDZ interacts with other other neurotransmitter systems. It is now known that IDZ is an agonist at 5-HT 1A autoreceptors modulating 5-HT synthesis in the rat brain in vivo, causing overall a reduction in the synthesis of 5-HT in the cerebral cortex [74]. It is widely recognized that 5-HT constrains the activity of DA, resulting in an opposing relationship between DA and 5- HT [75], and so a decrease in activity in the serotonergic activity potentially could lead to overdosing of the dopaminergic neurotransmitter system. Consistent with this hypothesis, Matsumoto et al. [76] have recently demonstrated that IDZ can induce an increase in DA release in the prefrontal cortex. This is important as excessive DA activity in the prefrontal cortex has previously found to be detrimental to many cognitive functions mediated by the prefrontal cortex, including working memory [77]. The specific nature of cognitive deficits seen in patients with mild fvftd do, however, provide some important insights into the nature of the neurotransmitter systems which may be compromised early in the disease process. Deficits in reversal learning similar to those seen in patients with fvftd can be seen with dysfunction of serotonergic neurotransmitter systems in the prefrontal cortex. Normal subjects with low levels of 5-HT centrally induced by a low tryptophan drink, when required to perform a simple rule reversal in the presence of an irrelevant dimension (the cd_r stage), needed more trials to learn this rule, in comparison to subjects receiving placebo [78]. Nomura [79] has reported an analogous effect in rodents: rats given a low tryptophan diet showed impaired learning in an operant type discrimination learning paradigm. In the decision-making task, patients with fvftd show marked increases in deliberation times, akin to those seen in normal volunteers with acutely reduced 5-HT function. Miller et al. [66] have recently suggested that some of the symptoms associated with FTD are related to serotonergic dysfunction. Weight gain exceeding 4.5 kg occurred in 64% of FTD patients, and a change in food preference to carbohydrates occurred in 75%. Also, severe compulsions were found in 64% of individuals with FTD. There is indeed currently some neurochemical evidence from postmortem studies to suggest that there exist profound postsynaptic serotonergic receptor abnormalities in patients with FTD in comparison to controls [80]. It is noteworthy that, in this study, reduced total 5-HT receptor binding was found in samples of tissue from autopsyproven cases of FTD from the frontal pole, temporal pole and the hypothalamus, but not in the cholinergic cells of the nucleus basalis of Meynert, in comparison to controls. In any case, various other studies [68, 81] converge upon the notion that the neurochemical changes of FTD have a profile distinct from that of DAT. Many of the symptoms of fvftd are indeed currently treated using serotonin-boosting compounds, including impulsive behaviour [82], depression [83], alterations in eating habits [84], and obsessions and compulsions [85]. Swartz et al. [86] have demonstrated in preliminary studies that it is possible to ameliorate many of these specific symptoms, including impulsivity, depression, carbohydrate craving and compulsions, in fvftd patients by using 5-HT-boosting compounds. Mild improvements in eating disorders have also been noted. However, as the authors themselves acknowledged, this study suffered from an uncontrolled and unblinded design and the use of three different SSRIs. Clearly, more studies have to be Cognition and Drug Therapy in Frontotemporal Dementia Dement Geriatr Cogn Disord 1999;10(suppl 1):

55 done in this area to examine objectively the effects of 5- HT-boosting compounds upon behavioural symptoms, and in particular the effect upon these compounds upon cognition given the potential problems that may occur, described earlier for IDZ. It shall also be useful to consider in particular whether a specific pattern of presenting clinical symptoms may predict response to therapy. The accurate diagnosis of symptoms in patients is therefore extremely important. For example, Levy et al. [87] recently highlighted that apathy did not correlate with depression in a combined sample of patients, including 28 patients with FTD, and hence that distinguishing the two syndromes might have therapeutic implications. This is important given that apathy may be a later presentation in fvftd [64]; furthermore, apathy may be associated with significantly reduced rcbf in the dorsolateral (rather than the orbitofrontal) prefrontal cortex [88, 89]. Agents used to treat apathy behaviourally have previously included drugs that modify dopaminergic neurotransmission including amphetamine, bromocriptine and methylphenidate [90]. These drugs may also be of some use in treatment of cognitive deficits consequent upon dysfunction of the dorsolateral prefrontal cortex later in the disease process. A landmark study by Brozoski et al. [91] demonstrated that 6-hydroxydopamine lesions of the dorsolateral prefrontal cortex in rhesus monkeys was as devastating to memory performance as removing the cortex itself, and that, importantly, this impaired performance could be reversed by DA agonists such as L-dopa and apomorphine. In relation to this, it is useful to note that Elliott et al. [92] have demonstrated significantly enhanced performance in tests of spatial working memory and planning in healthy young adults, using methylphenidate, a stimulant related to amphetamine [93]. It is therefore clear that much can be accrued from a comparative approach to frontal lobe function not only for our understanding of the neuropsychological deficits in patients with fvftd, but also for the development of successful therapeutic strategies. Conclusions Whilst it is understood that various divisions of the prefrontal cortex appear to have different patterns of anatomical connectivity and function, the precise ways in which they interact to control cognition and behaviour in social and nonsocial contexts is still relatively poorly understood. A precise understanding of the nature of frontal lobe function in humans and nonhuman primates, and a precise characterization of cognition and behaviour in patients with early frontal variant frontotemporal dementia, can complement each other not only in addressing this problem, but also in the development of novel rational therapeutic strategies for intervention in patients in the mild stages of disease. It is hoped that this will result in substantial benefits for these patients in terms of their cognition and behaviour. Acknowledgements We thank Dr. A.M. Owen, Dr. R.D. Rogers and Prof. J.R. Hodges for discussion. The majority of the work described in this paper was supported by a Programme Grant from the Wellcome Trust to Prof. T.W. Robbins, Dr. B.J. Sahakian, Prof. B.J. Everitt and Dr. A.C. Roberts. S.R. is funded by a Medical Research Council Research Studentship. Clinical aspects of the work with frontotemporal dementia patients was funded by a Medical Research Council Programme Grant to Prof. J.R. Hodges. References 1 Neary D, Snowden JS: Frontotemporal dementia: Nosology, neuropsychology, and neuropathology. Brain Cogn 1996;31: Hooten WM, Lyketsos CG: Differentiating Alzheimer s disease and frontotemporal dementia: Receptor operator characteristic curve analysis of four rating scales. Dement Geriatr Cogn Disord 1998;9: Miller BL, Ikonte C, Ponton M, Levy M, Boone K, Darby A, Berman N, Mena I, Cummings JL: A study of the Lund-Manchester research criteria for frontotemporal dementia: Clinical and single-photon emission CT correlations. Neurology 1997;48: Barber R, Snowden JS, Craufurd D: Frontotemporal dementia and Alzheimer s disease: Retrospective differentiation using information from informants. J Neurol Neurosurg Psychiatry 1995;59: Starkstein SE, Migliorelli R, Teson A, Sabe L, Vazquez S, Turjanski M, Robinson RG, Leiguarda R: Specificity of changes in cerebral blood flow in patients with frontal lobe dementia. J Neurol Neurosurg Psychiatry 1994;57: Petrides M, Pandya DN: Association fiber pathways to the frontal cortex from the superior temporal region in the rhesus monkey. J Comp Neurol 1988:273: Preuss TM, Kaas JH: Human brain evolution; in Zigmond, Bloom, Landis, Roberts, Squire (eds): Fundamental Neuroscience, chapt 50. New York, Academic Press, Alexander GE, DeLong MR, Strick PL: Parallel organization of functionally segregated circuits linking basal ganglia and cortex. Annu Rev Neurosci 1986;9: Dement Geriatr Cogn Disord 1999;10(suppl 1):15 28 Rahman/Robbins/Sahakian

56 9 Alexander GE, Crutcher MD: Functional architecture of basal ganglia circuits: Neural substrates of parallel processing. Trends Neurosci 1990;13: Stuss DT: Assessment of neuropsychological dysfunction in frontal lobe degeneration (review). Dementia 1993;4: Ferrier D: The Functions of the Brain. Smith & Elder, Blumer D, Benson D: Personality changes with frontal and temporal lobe lesions; in Benson DF, Blumer D (eds): Psychiatric Aspects of Neurologic Disease. New York, Grune & Stratton, 1975, pp Anderson SW, Damasio H, Tranel D, Damasio AR: Cognitive sequelae of focal lesions in ventromedial frontal lobe. J Clin Exp Neuropsychol 1992;14: Hunter R, Blackwood W, Bull J: Three cases of frontal meningiomas presenting psychiatrically. Br Med J 1968;iii: Logue V, Durward M, Pratt TRC, Piercy M, Nixon WL: The quality of survival after an acute anterior cerebral aneurysm. Br J Psychiatry 1968;114: Bogousslavsky J, Regli F: Anterior cerebral artery territory infarction in the Lausanne Stroke Registry. Arch Neurol 1990;47: Cummings JL: Frontal-subcortical circuits and human behavior (review). Arch Neurol 1993; 50: Bechara A, Damasio H, Tranel D, Anderson SW: Dissociation of working memory from decision making within the human prefrontal cortex. J Neurosci 1998;18: Dias R, Robbins TW, Roberts AC: Dissociation in prefrontal cortex of affective and emotional shifts. Nature 1996;380: Dias R, Robbins TW, Roberts AC: Dissociable forms of inhibitory control within prefrontal cortex with an analog of the Wisconsin Card Sort Test: Restriction to novel situations and independence from on-line processing. J Neurosci 1997;17: Fray PJ, Robbins TW, Sahakian BJ: Neuropsychiatric applications of CANTAB. Int J Geriatr Psychiatry 1996;11: Gaffan D: Recognition impaired and association intact in the memory of monkeys after transection of the fornix. J Comp Physiol Psychol 1974;86: Mishkin M: A memory system in the monkey. Philos Trans R Soc Lond 1982;298: Owen AM, Sahakian BJ, Semple J, Polkey CE, Robbins TW: Visuo-spatial short-term recognition memory and learning after temporal lobe excisions, frontal lobe excisions or amygdalohippocampectomy in man. Neuropsychologia 1995;33: Sahakian BJ, Morris RG, Evenden JL, Heald A, Levy R, Philpot M, Roberts AC, Robbins TW: A comparative study of visuospatial memory and learning in Alzheimer-type dementia and Parkinson s disease. Brain 1988;111: Sahgal A, Sahakian BJ, Robbins TW, Wray C, Lloyd S, Cook JH, McKeith IG, Disley J, Eagger S, Boddington S, Edwardson JA: Detection of visual memory and learning deficits in Alzheimer s disease using the Cambridge Neuropsychological Test Automated Battery (CAN- TAB). Dementia 1991;2: Sahgal A, Lloyd S, Wray C, Galloway P, Robbins TW, Sahakian BJ, McKeith IG, Cook JH, Disley JCA, Edwardson JA: Does visuospatial memory in Alzheimer s disease depend on the severity of the disorder? Int J Geriatr Psychiatry 1992;7: Van Hoesen GW, Solodkin A: Cellular and systems neuroanatomical changes in Alzheimer s disease. Ann NY Acad Sci 1994;747: Owen AM, Roberts AC, Hodges JR, Robbins TW: Contrasting mechanisms of impaired attentional set-shifting in patients with frontal lobe damage or Parkinson s disease. Brain 1993;116: Lange KW, Sahakian BJ, Quinn NP, Marsden CD, Robbins TW: Comparison of executive and visuospatial memory function in Huntington s disease and dementia of Alzheimer-type matched for degree of dementia. J Neurol Neurosurg Psychiatry 1995;58: Shallice T: Specific impairments of planning. Philos Trans R Soc Lond [B] 1982;298: Goel V, Grafman J: Are the frontal lobes implicated in planning functions? Interpreting data from the Tower of Hanoi. Neuropsychologia 1995;33: Baker RC, Rogers RD, Owen AM, Frith CD, Dolan RJ, Frackowiak RS, Robbins TW: Neural systems engaged by planning: A PET study of the Tower of London task. Neuropsychologia 1996;34: Goldman-Rakic PS: Regional and cellular fractionation of working memory (review). Proc Natl Acad Sci USA 1996;26: Passingham RE: Memory of monkeys (Macaca mulatta) with lesions in prefrontal cortex. Behav Neurosci 1985;99: Olton DS, Becker JT, Handelman GE: Hippocampus, space and memory. Behav Brain Sci 1979;2: Owen AM, James M, Leigh PN, Summers BA, Marsden CD, Quinn NP, Lange KW, Robbins TW: Fronto-striatal cognitive deficits at different stages of Parkinson s disease. Brain 1992; 115: Robbins TW, James M, Owen AM, Lange KW, Lees AJ, Leigh PN, Marsden CD, Quinn NP, Summers BA: Cognitive deficits in progressive supranuclear palsy, Parkinson s disease and multiple system atrophy in tests sensitive to frontal lobe dysfunction. J Neurol Neurosurg Psychiatry 1994;57: 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: Owen AM: The functional organization of working memory processes within human lateral frontal cortex: The contribution of functional neuroimaging. Eur J Neurosci 1997;9: Owen AM, Downes JJ, Sahakian BJ, Polkey CE, Robbins TW: Planning and spatial working memory following frontal lobe lesions in man. Neuropsychologia 1990;28: Roberts AC, Robbins TW, Everitt BJ: The effects of intra-dimensional and extra-dimensional shifts on visual discrimination learning in humans and non-human primates. Q J Exp Psychol 1988;40: Downes JJ, Roberts AC, Sahakian BJ, Evenden JL, Morris RG, Robbins TW: Impaired extradimensional shift performance in medicated and unmedicated Parkinson s disease: Evidence for a specific attentional dysfunction. Neuropsychologia 1989;27: Eimas PD: Effects of overtraining and age on intradimensional and extradimensional shifts in children. J Exp Child Psychol 1966;3: Schepp BE, Schrier AM: Consecutive intradimensional and extradimensional shifts in monkeys. J Comp Physiol Psychol 1969;67: Owen AM, Roberts AC, Polkey CE, Sahakian BJ, Robbins TW: Extra-dimensional versus intra-dimensional set-shifting performance following frontal lobe excisions, temporal lobe excisions or amygdalo-hippocampectomy in Man. Neuropsychologia 1991;29: Grant DA, Berg EA: A behavioural analysis of degree of reinforcement and ease of shifting to new responses in a Weigl-type card sorting problem. J Exp Psychol 1948;38: Milner B, Corkin S, Teuber HL: Further analysis of the hippocampal amnesic syndrome: 14- year follow-up study of HM. Neuropsychologia 1968;6: Haaland KY, Vranes LF, Goodwin JS, Garry PJ: Wisconsin Card Sort Test performance in a healthy elderly population. J Gerontol 1987; 42: Lawrence AD, Hodges JR, Rosser AE, Kershaw A, ffrench-constant C, Rubinsztein DC, Robbins TW, Sahakian BJ: Evidence for specific cognitive deficits in preclinical Huntington s disease. Brain 1998;121: Lawrence AD, Sahakian BJ, Hodges JR, Rosser AE, Lange KW, Robbins TW: Executive and mnemonic functions in early Huntington s disease. Brain 1996;119: Sahakian BJ, Downes JJ, Eagger S, Evenden JL, Levy R, Philpot MP, Roberts AC, Robbins TW: Sparing of attentional relative to mnemonic function in a subgroup of patients with dementia of the Alzheimer type. Neuropsychologia 1990;28: Rogers RD, Andrews TC, Grasby PM, Brooks D, Robbins TW: Contrasting cortical and subcortical PET activations produced by reversal learning and attentional set-shifting in humans. Soc Neurosci Abstr 1997;23:87.1. Cognition and Drug Therapy in Frontotemporal Dementia Dement Geriatr Cogn Disord 1999;10(suppl 1):

57 54 Divac I, Rosvold HE, Szwarcbart MK: Behavioural effects of selective ablation of the caudate nucleus. J Comp Neurol 1967;334: Mishkin M, Malamut B, Bachevalier J: Memories and habits: two neural systems; in Lynch G, McGaugh JL, Weinberger NM (eds): Neurobiology of Learning and Memory. Guildford, 1984, Williams GV, Rolls ET, Leonard CM, Stern C: Neuronal responses in the ventral striatum of the behaving monkey. Behav Brain Res 1993; 55: Mendez MF, Adams NL, Lewandowski KS: Neurobehavioral changes associated with caudate lesions. Neurology 1989;39: Filley CM, Kleinschmidt-De Masters BK, Gross KF: Non-Alzheimer fronto-temporal degenerative dementia. A neurobehavioural and pathologic study. Clin Neuropathol 1994;13: Leichnetz GR, Astruc J: Efferent connections of the orbitofrontal cortex in the marmoset. Brain Res 1975;84: Rogers RD, Everitt BJ, Baldacchino A, Blackshaw AJ, Swainson R, Wynne K, Baker NB, Hunter J, Carthy T, Booker E, London M, Deakin JFW, Sahakian BJ, Robbins TW: Dissociable deficits in the decision-making cognition of chronic amphetamine abusers, opiate abusers, patients with focal damage to prefrontal cortex, and tryptophan-depleted normal volunteers: Evidence for monoaminergic mechanisms. Neuropsychopharmacology 1999;20: Bechara, A, Damasio, AR, Damasio, H, Anderson, SW: Insensitivity to future consequences following damage to human prefrontal cortex. Cognition 1994;50: Bechara A, Damasio H, Tranel D, Damasio AR: Deciding advantageously before knowing the advantageous strategy. Science 1997;275: Masterman DL, Cummings JL: Frontal-subcortical circuits: The anatomic basis of executive, social and motivated behaviours. J Psychopharmacol 1997;11: Gustafson L, Brun A, Passant U: Frontal lobe degeneration of non-alzheimer type. Ballières Clin Neurol 1992;1; Mendez MF, Perryman KM, Miller BL, Swartz JR, Cummings JL: Compulsive behaviors as presenting symptoms of frontotemporal dementia. J Geriatr Psychiatry Neurol 1997;10: Miller BL, Darby AL, Swartz JR, Yener GG, Mena I: Dietary changes, compulsions and sexual behavior in frontotemporal degeneration. Dementia 1995;6: Lawrence AD, Sahakian BJ: The cognitive psychopharmacology of Alzheimer s disease: Focus on cholinergic systems. Neurochem Res 1998;23: Francis PT, Holmes C, Webster MT, Stratmann GC, Procter AW, Bowen DM: Preliminary neurochemical findings in non-alzheimer dementia due to lobar atrophy. Dementia 1993; 4; Coull JT, Sahakian BJ, Hodges JR: The 2 antagonist idazoxan remediates certain attentional and executive dysfunction in patients with dementia of frontal type. Psychopharmacology (Berl) 1996;123: Dennis T, L Heureux R, Carter C, Scatton B: Presynaptic alpha-2 adrenoreceptors play a major role in the effects of idazoxan on cortical noradrenaline release (as measured by in vivo dialysis) in the rat. J Pharmacol Exp Ther 1987; 241: Goodwin GM, Conway SC, Peyro-Saint-Paul H, Glabus MF, O Carroll RE, Ebmeier KP: Executive function and uptake of 99m Tc-exametazime shown by single photon emission tomography after oral idazoxan in probable Alzheimer-type dementia. Psychopharmacology (Berl) 1997;131: Franowicz JS, Arnsten AF: The alpha-2 -noradrenergic agonist, guanfacine, improves delayed response performance in young adult rhesus monkeys. Psychopharmacology (Berl) 1998;136: Arnsten AFT: Catecholamine regulation of the prefrontal cortex (review). J Psychopharmacol 1997;11: Llado J, Esteban S, Garcia-Sevilla JA: The alpha-2-adrenoreceptor antagonist idazoxan is an agonist at 5-HT 1A autoreceptors modulating serotonin synthesis in the rat brain in vivo. Neurosci Lett 1996;218: Luciana M, Collins PF, Depue RA: Opposing roles for dopamine and serotonin in the modulation of human spatial working memory functions. Cereb Cortex 1998;8: Matsumoto M, Yoshioko M, Togashi H, Mori K, Ueno K, Saito H: Effects of idazoxan on dopamine release in the prefrontal cortex of freely moving rats. Eur J Pharmacol 1998;343: Murphy BL, Arnsten AF, Goldman-Rakic PS, Roth RH: Increased dopamine turnover in the prefrontal cortex impairs spatial working memory performance in rats and monkeys. Proc Natl Acad Sci 1996;93: Park SB, Coull JT, McShane RH, Young AH, Sahakian BJ, Robbins TW, Cowen PJ: Tryptophan depletion in normal volunteers produces selective impairments in learning and memory. Psychopharmacology 1994;33: Nomura M: Effects of bifemalane on discrimination learning of serotonergic dysfunction in rats. Pharmacol Biochem Behav 1992;42: Sparks DL, Markesbery WR: Altered serotonergic and cholinergic synaptic markers in Pick s disease. Arch Neurol 1991;48: Qume M, Mistra A, Zeman S, Muthu J, Cross AJ, Francis PT, Procter AW, Neary D, Bowen DM: Serotonergic profiles of lobar atrophies. Biochem Soc Trans 1995;23:600S. 82 Coccaro EF: Central serotonin and impulsive aggression (review). Br J Psychiatry 1989;8: Goodwin GM: How do antidepressants affect serotonin receptors? The role of serotonin receptors in the therapeutic and side effect profile of the SSRIs. J Clin Psychiatry 1996;57(suppl 4): Yager J: The treatment of eating disorders. J Clin Psychiatry 1988;49: Hollander E: Obsessive-compulsive disorderrelated disorders: The role of selective serotonergic reuptake inhibitors. (review). Int Clin Psychopharmacol 1996;11(suppl 5): Swartz R, Miller BL, Darby A, Schuman S: Frontotemporal dementia: Treatment response to serotonin-selective reuptake inhibitors. J Clin Psychiatry 1997;58: Levy ML, Cummings JL, Fairbanks LA, Masterman D, Miller BL, Craig AH, Paulsen JS, Litvan I: Apathy is not depression. J Neuropsychiatry Clin Neurosci 1998;10: Craig AH, Cummings JL, Fairbanks L, Itti L, Miller BL, Li J, Mena I: Cerebral blood flow correlates of apathy in Alzheimer disease. Arch Neurol 1996;53: Okada K, Kobayashi S, Yamagata S, Takahashi K, Yagmuchi S: Poststroke apathy and regional cerebral blood flow. Stroke 1997;28: Marin RS, Fogel BS, Hawkins J, Duffy J, Krupp B: Apathy: A treatable syndrome. J Neuropsychiatry Clin Neurosci 1995;7: Brozoski T, Brown RM, Rosvold HE, Goldman PS: Cognitive deficit caused by regional depletion of dopamine in prefrontal cortex of rhesus monkeys. Science 1979;205: Elliott R, Sahakian BJ, Matthews K, Bannerjea A, Rimmer J, Robbins TW: Effects of methylphenidate on spatial working memory and planning in healthy young adults. Psychopharmacology (Berl) 1997;131: Scheel-Kruger J: Comparative studies of various amphetamine analogues demonstrating different interactions with the metabolism of the catecholamines in the brain. Eur J Pharmacol 1971;14: Dement Geriatr Cogn Disord 1999;10(suppl 1):15 28 Rahman/Robbins/Sahakian

58 Review TRENDS in Cognitive Sciences Vol.5 No.6 June 2001 Decision making and neuropsychiatry Shibley Rahman, Barbara J. Sahakian, Rudolf N. Cardinal, Robert D. Rogers and Trevor W. Robbins 271 Abnormal decision making is a central feature of neuropsychiatric disorders. Recent investigations of the neural substrates underlying decision making have involved qualitative assessment of the cognition of decision making in clinical lesion studies (in patients with frontal lobe dementia) and neuropsychiatric disorders such as mania, substance abuse and personality disorders. A neural network involving the orbitofrontal cortex, ventral striatum and modulatory ascending neurotransmitter systems has been identified as having a fundamental role in decision making and in the neural basis of neuropsychiatric diseases. This network accounts for the dissociations among decision-making deficits in different clinical populations. Ultimately, a more refined and sophisticated characterization of such deficits might guide the early diagnosis and cognitive and therapeutic rehabilitation of these patients. S. Rahman B.J. Sahakian* Dept of Psychiatry, University of Cambridge, Level E4, Addenbrooke s Hospital, Hills Road, Cambridge, UK CB2 2QQ. * msexc.addenbrookes. anglox.nhs.uk R.N. Cardinal T.W. Robbins Dept of Experimental Psychology, University of Cambridge, Downing Street, Cambridge, UK CB2 3EB. R.D. Rogers Dept of Psychiatry, University of Oxford, Warneford Hospital, Oxford, UK OX3 7JX. Abnormal decision making has been demonstrated to lie at the core of several common neuropsychiatric disorders ranging from substance abuse to mania. Crucially, these deficits can all be explained within a unified theoretical framework. Reports of the detrimental effects of brain lesions upon decision making have existed since the classic neurological case of Phineas Gage, arguably the earliest recorded case of acquired sociopathy 1,2. Patients with injury to the prefrontal cortex (PFC) engage in decisions and behaviours that have repeated negative consequences for their wellbeing 3 5. They act as though they have lost the ability to ponder different courses of action and to select the one that promises the best balance of short- and long-term benefit. They also have difficulty in planning and organizing their own lives and tend to make poor decisions about friends, business associates and day-to-day activities. Specifically, patients with lesions of the orbitofrontal region of the prefrontal cortex (OFC) exhibit profound personality changes, problems with self-conduct, difficulties with emotional reactions to sensory stimuli, difficulties with social interactions, and problems making decisions within the context of their everyday lives. All this is in the conspicuous absence of the marked cognitive deficits more frequently shown by patients with damage to dorsal areas of PFC (Refs 6 8). Recent research has been advanced significantly by the demonstration that patients exhibiting acquired sociopathy following orbitofrontal damage also show consistent deficits on a laboratory gambling task involving choices between actions that differ in terms of the size and probabilities of their associated punishments and rewards 9,10. It has, however, only recently been realized that an understanding of the neural substrates can be applied to clinical disorders, and attempts are now being made to compare the nature of the decision making in these disorders. The complexity of decision-making processes means that deficits in decision-making cognition can be manifested in several ways. An individual might take a protracted time to make a decision, allocate an inappropriate amount of resources to a given decision or tend to make decisions that are unlikely to produce the desired outcome. For example, one deficit demonstrated by patients with orbitofrontal lesions is a very large increase in deliberation times associated with their decisions 11. These data reflect the clinical observation that such patients take protracted times to make decisions in their day-to-day lives, and confirm that damage to the OFC is especially associated with decision-making deficits when there is limited contextual information to assist the identification of the optimal response. The slow, ineffective deliberation between choices suggests that the OFC does not mediate a simple inhibitory mechanism. Significantly, in a decision-making task that we have developed, in which subjects place bets with different odds of success (Box 1), OFC-lesioned patients risk significantly less of their accumulated reward than controls at the more favourable odds. Damage to the OFC can thus lead to a pattern of conservative behaviour rather than a pattern of gambling and risk-taking decisions; these patterns are therefore dissociable. A key point regarding this result is that patients with lesions to more dorsal areas of PFC appear to be unaffected in such decision making. Instead, the dorsolateral PFC has repeatedly been shown to mediate important aspects of the executive control of behaviour, such as working memory, planning and attention These findings emphasize functional as well as structural differences between different areas of PFC. The decision-making task shown in Box 1 has been applied successfully to several different neuropsychiatric disorders and has the particular advantage of having been validated through functional neuroimaging, as well as lesion studies. A recent PET study of decision-making cognition required subjects to gamble accumulated reward on predictions about which of two mutually exclusive outcomes would occur 15. Critically, the largest reward was always associated with the least likely of the two outcomes, ensuring an element of conflict inherent in risk-taking. Resolving these decisions was associated with three distinct foci of activation /01/$ see front matter 2001 Elsevier Science Ltd. All rights reserved. PII: S (00)

59 272 Review TRENDS in Cognitive Sciences Vol.5 No.6 June 2001 Box 1. Neurocognitive assessment of decision making To assess decision making in the laboratory, we have developed a computerized decision-making task a (Fig. I). Subjects are told that the computer has hidden a yellow token at random inside one box of a configuration of red and blue boxes. The subjects first have to decide in what colour box the token has been hidden, before attempting to increase their points by betting on whether or not they believe their choice to be correct. There are two betting conditions: (1) the bet on offer either ascends or descends, and (2) the subjects touch the screen when they are happy with the bet on offer. One of the boxes at the top of the display then opens to reveal the actual location of the yellow token and the chosen bet is added to or subtracted from the total points score, according to whether the initial colour decision was correct. Subjects are given 100 points at the beginning of each block of trials and, although no real monetary significance is attached to the points accumulated by the end of the task, subjects are encouraged to treat the points as valuable and to accumulate as many as possible. If a subject s score falls to just 1 point, the current block terminates and the next begins. This computerized task provides explicit information about the relative attractiveness of two mutually exclusive response options. It has a lighter load on working memory than other decisionmaking tasks because it is less dependent on the outcome of previous trials. By using information presented in a readily comprehensible visual format, this task allows subjects to choose what they POINTS 100 RED BLUE 25 TRENDS in Cognitive Sciences Fig. I. View of the screen in a task used to assess decision-making cognition in healthy volunteers and patient groups. (See text for details.) perceive to be the most likely outcome and, importantly, to state how much they are prepared to bet that they are correct. In real life, this relates to the need to weigh-up available opportunities, judge the relative probabilities of a successful outcome and then choose how much current resources or reward should be com mitted to the chosen strategy. Reference a Rogers, R.D. et al. (1999) Dissociable deficits in the decision-making cognition of chronic amphetamine abusers, opiate abusers, patients with focal damage to prefrontal cortex, and tryptophan-depleted normal volunteers: evidence for monoaminergic mechanisms. Neuropsychopharmacology 20, within the right inferior and orbital PFC (Fig. 1): (1) laterally in the anterior part of the middle frontal gyrus, (2) medially in the orbital gyrus and (3) posteriorly in the anterior portion of the inferior frontal gyrus. By contrast, increases in the degree of conflict associated with these decisions were associated with increased activity within the anterior cingulate cortex, together with only limited, mainly left-sided, changes in orbital PFC activity. Choices in this study were not associated with any significant changes in neural activity within dorsolateral prefrontal areas, consistent with the findings from brain-damaged patients 11. Elucidation of the neural substrates underlying decision making using these different neuroscientific techniques is clearly essential to understand decision-making abnormalities in neuropsychiatric disorders. Somatic markers To explain the dissociation between personal decision-making ability and other aspects of cognition, markers or biasing factors in humans have been postulated to act in normal cognition to enhance decision making 10. For the brain to compute expected utilities accurately (as a computer might) would take a finite time; it is better to make an imperfect decision quickly than eventually to make what would have been the perfect decision. Damasio has argued that somatic markers provide a way of speeding up decision making 7. Somatic markers are signals relating to body states (in other words, representations of the body itself) that are acquired early in the sampling of novel action outcome contingencies. Once retrieved, these markers influence the processes of responding to stimuli in several ways; some markers act consciously ( in mind ) and others covertly, in a non-minded manner. The markers pre-bias cognitive systems, preventing them from considering particularly bad courses of action. Somatic markers, therefore, constitute a rapidly retrieved signal that improves performance by removing options from the consideration of a computationally intensive cognitive process. One example of a somatic marker is the skin conductance response (SCR) induced by sympathetic nervous system activity (and thus is an index of autonomic arousal). This marker is probably to be

60 Review TRENDS in Cognitive Sciences Vol.5 No.6 June (a) (b) (c) Sagittal Fig. 1. Neuroimaging of the decision-making task. Peaks of activityassociated performance in the decision-making task illustrated in Box 1 (compared with a visuomotor control task) rendered onto the averaged MRI scans of eight volunteer subjects used in the study 15 (threshold, P < 0.01). (a) Peak activation in orbitomedial PFC (BA 11). (b) Peak activation within orbitolateral PFC (BA 10). (c) activation within the inferior convexity (BA 47). (Reproduced by kind permission of the Society for Neuroscience.) Transverse Z Value Z Value Z Value both sensed and generated by the PFC. There is fmri evidence that neural activity involving both medial prefrontal cortex and the right OFC accompanies the generation and afferent representation of discrete SCRs (Ref. 16). The marker can reach the OFC directly, through the somatosensory cortex, or indirectly, through the interactions of the ascending somatic sensory system and the ascending, chemically defined neurotransmitter systems of the isodendritic core. Learned versions of somatic responses might also be reproduced within the somatosensory cortex and relayed to the OFC in a manner that is sometimes described as an as if loop (the somatosensory information originates in the cortex but as if it had been produced in the body itself). By contrast, the medial network of the PFC appears to be the primary source of visceromotor outputs to the hypothalamus and brainstem. Central autonomic effectors, including the amygdala, are capable of activating somatic responses in the viscera and endocrine systems 17. Like subjects with damage to the OFC, subjects with somatosensory cortical damage or damage to the amygdala are impaired in the gambling task of Bechara et al. 18,19, suggesting a role for these structures (and the peripheral nervous system) in decision making. This is in keeping with the somatic marker hypothesis outlined above. Current lesion and neuroimaging studies therefore identify the OFC (Refs 20,21), and its functional TRENDS in Cognitive Sciences interactions with related structures such as the amygdala and somatosensory cortices, as being critical to many aspects of decision making and social/emotive cognition (see Fig. 2). Clinical disorders disrupting these systems include substance abuse, ruptured anterior communicating artery aneurysms, frontal-variant frontotemporal dementia, bipolar and unipolar depression and personality disorders. As deficits in decision making are now both qualifiable and quantifiable, it is essential to define, clinically and neuropsychologically, the meaning of commonly used terms such as impulsivity, risk-taking and disinhibition. Decision making in substance abusers Examining the decision-making behaviour of substance abusers provides a useful starting point, because drug abuse could reflect a breakdown of the ability to evaluate potential reward against harm from drug self-administration. Activity in the OFC and its connections have now been found to play a role in several components of the maladaptive behaviour of substance abuse, including expectancy, craving and impaired decision making 22. The importance of the ventral striatum and the amygdala in humans is being further highlighted in novel functional neuroimaging studies by Breiter and Rosen 23. These structures are part of an extended neural network involved in processing the features of rewards and assigning value to the goals of behaviour in the context of motivational states. Recent findings indicate that chronic amphetamine abusers show a pattern of decision-making deficit closely resembling those shown by OFC-lesioned patients. This suggests that decision-making Striatum Pallidum Thalamus Body SI, SII and insular cortices Ascending neuromodulatory systems Orbitofrontal cortex Amygdala Hypothalamus TRENDS in Cognitive Sciences Fig. 2. The neural substrates of decision making. Postulated key structures involved in decision making, and their inter-relationships.