Dopaminergic modulation of cognitive function-implications for l-DOPA treatment in Parkinson's disease

Abstract

It is well recognised that patients with Parkinson's disease exhibit cognitive deficits, even in the earliest disease stages. Whereas, l-DOPA therapy in early Parkinson's disease is accepted to improve the motor symptoms, the effects on cognitive performance are more complex: both positive and negative effects have been observed. The purpose of the present article is to review the effects of l-DOPA medication in Parkinson's disease on cognitive functions in the broad domains of cognitive flexibility and working memory. The review places the effects in Parkinson's disease within a framework of evidence from studies with healthy human volunteers, rodents and non-human primates as well as computational modeling work. It is suggested that beneficial or detrimental effects of l-DOPA are observed depending on task demands and basal dopamine levels in distinct parts of the striatum. The study of the beneficial and detrimental cognitive effects of l-DOPA in Parkinson's disease has substantial implications for the understanding and treatment development of cognitive abnormalities in Parkinson's disease as well as normal health.

Introduction

Parkinson's disease (PD) is a progressive neurodegenerative disorder, characterised primarily by motor symptoms such as tremor, rigidity and bradykinesia. In addition to an increased risk for clinical dementia and clinical depression (Brown and Marsden, 1984), non-demented and non-depressed PD patients exhibit subtle cognitive problems, even in the earliest disease stages, which predict incident pathological dementia (Woods and Troster, 2003) and quality of life (Schrag et al., 2000). These cognitive difficulties resemble, but are not identical to those observed in patients with frontal lobe damage and mainly include so-called executive deficits (Brown and Marsden, 1988a, Cooper et al., 1991, Dubois and Pillon, 1997, Lees and Smith, 1983, Owen et al., 1992, Owen et al., 1993, Partiot et al., 1996, Taylor and Saint-Cyr, 1995, Taylor et al., 1986). Whilst medication with l-DOPA (l-3,4-dihydroxyphenylalanine) is well known to improve the motor symptoms, effects on cognitive functions are more complex: both positive as well as negative effects have been observed. In the present article, I put forward the hypothesis that these contrasting effects of l-DOPA reflect the spatio-temporal progression of dopamine (DA) depletion, which, in the earliest disease stages, is most severe in the dorsal striatum and progresses only later to the ventral striatum (Fig. 1) (Bernheimer et al., 1973, Kish et al., 1988). In brief, l-DOPA in early PD may improve certain cognitive functions that are associated with the severely depleted dorsal striatum, whilst at the same time impairing (by ‘over-dosing’) other cognitive functions, associated with the relatively intact ventral striatum.

The core pathology underlying PD is degeneration of the DA cells in the midbrain, leading to severe DA depletion in the striatum (Dauer and Przedborski, 2003). Accordingly, the motor symptoms and some of the cognitive deficits may be alleviated by replenishment of striatal DA through the oral administration of the DA precursor l-DOPA or synthetic DA receptor agonists (Hornykiewicz, 1974). Surgical treatments including pallidotomy and deep brain stimulation techniques targeting the globus pallidus or the subthalamic nucleus have also been found to improve in particular the motor symptoms of the disease as well as some cognitive functions (e.g. Fukuda et al., 2002, Obeso et al., 2000). The renewed interest in the effects of surgical treatments on cognitive function is not the focus of the present article and the reader is referred to recent reviews (Morrison et al., 2004, Pillon et al., 2003).

Most severe in PD is the DA cell loss in the ventrolateral tier of the substantia nigra pars compacta, which projects primarily to the dorsal striatum (i.e. the dorsolateral putamen and the dorsal parts of the caudate nucleus). The dorsal striatum in turn projects predominantly to a selective set of cortical structures, including the motor and premotor cortex, the supplementary motor area as well as the dorsolateral PFC (Fig. 1) (Alexander et al., 1986). Much less affected are cells in the dorsal tier of the midbrain, including the ventral tegmental area (VTA), which project to the ventral striatum (i.e. the ventral putamen, the ventral caudate nucleus and the nucleus accumbens). This part of the striatum is strongly connected to the amygdala, the orbitofrontal cortex (OFC), the anterior cingulate cortex and the inferotemporal cortex (Middleton and Strick, 1996, Selemon and Goldman-Rakic, 1985). Although there may be some degeneration of serotonergic cells in the dorsal raphe nucleus and cholinergic neurons in the substantia innominata (Fig. 1) (Agid et al., 1987, Paulus and Jellinger, 1991) as well as formation of Lewy Bodies (Gibb et al., 1989), these non-dopaminergic abnormalities are thought to develop only in the later stages of the disease, so that pathology is relatively restricted to DA depletion in the dorsal striatum, at least early in the course of the disease (Kish et al., 1988, Dauer and Przedborski, 2003). In the PFC, DA function is relatively unaffected early on (Agid et al., 1987) and may even be upregulated (Kaasinen et al., 2001, Rakshi et al., 1999).

l-DOPA is a widely used and effective treatment for PD and has been shown to benefit certain cognitive functions, although detrimental effects can also develop following l-DOPA therapy. In addition to the occasionally observed severely disabling side effects of psychosis and addiction (Lawrence et al., 2003), the drugs can cause subtle cognitive deficits. At first sight, the effects of l-DOPA appear unpredictable: l-DOPA impairs some, but improves other complex cognitive abilities. However, the relationship between performance and neurotransmission is clearly not random. In the present article I hope to clarify that the apparent unpredictability of the cognitive effects of l-DOPA in PD may be resolved to some extent when the functional and neural heterogeneity of complex cognitive function as well as the disease stage of patients are taken into account.

The procedure most commonly employed to assess effects of l-DOPA in PD is the controlled l-DOPA withdrawal procedure. It requires patients to abstain from their l-DOPA for a period of between 12 and 18 h prior to the neuropsychological assessment. Performance in this OFF state is compared with performance on a separate testing session during which patients take their medication as usual, the ON state. This procedure is less prone to the confounds of differences in disease severity than the alternative procedure in which procedure, performance of de novo, never-medicated patients is compared with performance of the same patients at a later stage after l-DOPA administration or of a different already-treated group.

In most reported medication withdrawal studies, at least some patients also take and withdraw from DA receptor agonists and/or catechol-O-methyltransferase (COMT) inhibitors in addition to their l-DOPA, suggesting that the here-reviewed effects may stem partly from withdrawal from these alternative DA enhancers. However, the drug most consistently manipulated in withdrawal studies is l-DOPA and the majority of selected patients are on l-DOPA only, partly because its shorter half-life renders l-DOPA more suitable to the withdrawal procedure than receptor agonists. Therefore, the present review is restricted to discussion of l-DOPA effects in PD. The controlled study of COMT inhibitors and DA receptor agonists, especially those with greater receptor specificity, is an important direction for future research.

Particularly vulnerable to PD are a set of complex cognitive control mechanisms, often referred to as ‘executive functions’, which contribute to the continuous orchestration of well-adapted habitual and goal-directed behaviour. Cognitive control mechanisms are generally accepted to comprise two mutually opponent computations: (i) the stable maintenance of cognitive representations, rendering them robust, stable and not easily degradable, even in the face of intervening distractors and (ii) the flexible alteration of these representations in response to changing environmental demands. I have chosen to restrict the review to these processes because they (i) have been examined in experimental animals and computational models, thereby directly guiding hypotheses regarding deficits in PD, (ii) are sensitive to manipulations of DA and (iii) are vulnerable in mild, early-in-the-course PD.

Much of our understanding about the role of DA in cognitive functioning has been elucidated by research with experimental animals, healthy human volunteers or computational modeling work. Therefore, the sections on l-DOPA effects in PD are preceeded by summaries of the progress made from these alternative approaches, culminating in explicit hypotheses regarding impairment in PD. Particular emphasis has been placed on functional differences between the striatum and the PFC (Bilder et al., 2004, Crofts et al., 2001, Frank et al., 2001) as well as those between the dorsal and ventral striatum (Robbins and Everitt, 1992, Voorn et al., 2004). More specifically, I suggest that the complex effects of PD and l-DOPA be understood from the perspective of the hypothesis that the striatum and the PFC play distinct roles in the plasticity and stability of cognitive representations, respectively. In addition, the dorsal and ventral striatum may subserve the plasticity of dissociable abstract, stimulus-response and more concrete, stimulus-outcome associations. Within this framework, the effects of l-DOPA in PD patients are considered. This approach highlights the consistency of behavioural patterns observed in PD patients with studies with experimental animals, healthy human volunteers and computational modeling work.