The nosology and treatment options for cognitive impairment in Parkinson disease (PD) remains perhaps the greatest unmet clinical need.1 The reported frequency of mild cognitive impairment (MCI) in newly diagnosed PD has ranged from 22% to 43%.2,3 A proportion of those with PD-MCI progress to PD dementia (PDD), the prevalence of which is variously estimated between 24% and 31%, but is as high as 75% in longitudinal studies.4 Dementia is associated with increased disability, nursing home placement, mortality, and caregiver stress. Global cognition and executive/working memory deficits also occur in individuals at risk for PD.5 Which features of MCI are predictive of development of dementia remain a matter of debate.
It may be tempting to place these impairments on a spectrum of cognitive dysfunction, but this conceptualization is not in keeping with the growing evidence on cognitive symptom heterogeneity in PD. One hypothesis distinguishes between 2 independent but partially overlapping cognitive syndromes associated with PD.6 Frontostriatal network dysfunction is an early, dopamine-modulated syndrome that manifests with difficulties in working memory, attention, planning, and response inhibition. PDD is a syndrome of progressively worsening impairments in verbal fluency, verbal/visual memory, and visuospatial skills, associated with more posterior cortical degeneration.6 A major, unanswered question regards the relationship between frontostriatal deficits and PDD: semantic fluency and visuoconstructional, but not spatial planning deficits, are predictive of progression to dementia, supporting at least a degree of independence between the 2 syndromes.7
In this issue of Neurology®, Niccolini et al.8 report on the relationship between brain regional expression of phosphodiesterase 4 (PDE4) and performance on the Cambridge Neuropsychological Test Automated Battery spatial working memory task in 12 levodopa-treated patients with PD, free of MCI or dementia. PDE4 is an intracellular enzyme abundant throughout the brain. It regulates the cyclic adenosine monophosphate cascade and modulates protein transcription involved in synaptic plasticity. It may also provide neuroprotection by increasing the expression of growth factors including brain-derived neurotrophic factor.9 PDE4 expression was evaluated using [11C]rolipram PET. Cognition was assessed with tests of psychomotor speed, attention, verbal and spatial episodic memory, and spatial working memory. Compared to healthy controls, patients performed worse on tests of processing speed, spatial episodic memory, and spatial working memory, and showed substantial reductions in PDE4 expression in subcortical and frontal cortical regions. An intriguing relationship emerged between performance on the spatial working memory task and a cognitive (nonmotor) striato-thalamo-cortical pattern of PDE4 loss. Several findings contribute to the poignancy of this pattern.
Reductions in patients’ PDE4 expression of up to 32% compared to controls occurred in brain regions that only partially overlap with the primary PD pathophysiology in the nigrostriatal pathway. PDE4 loss was observed in precentral gyrus, dorsomedial frontal cortex, supplementary motor area, caudate, accumbens, thalamus, and hypothalamus, but not putamen, globus pallidus, and substantia nigra. Cortical or subcortical volume did not account for these decreases. Remarkably, loss of PDE4 in dorsomedial frontal cortex, precommissural caudate, and thalamic nuclei showed a strong inverse relationship with errors (greater errors with PDE4 loss) on the working memory task, explaining 35%–55% of its variance. The scatterplots in figure 1C show most bivariate data points falling strikingly near the regression line. Even more remarkably, PDE4 loss did not correlate with any other function affected by PD, including processing speed, episodic memory, motor function, and neuropsychiatric burden. Niccolini et al. suggest that PDE4 loss is a neurochemical change in PD associated specifically with cognitive symptoms and a potential target for their treatment.
The discovery in patients with PD of altered brain levels of an enzyme involved in synaptic plasticity is exciting, but its interpretation challenging given the limited research on human clinical populations. The current study is the first to use [11C]rolipram in PD. It is not known whether PD-related α-synuclein aggregation and dopaminergic dysfunction drive PDE4 loss, with cognitive dysfunction resulting, or whether PD-unrelated processes including normal aging play a role. The current study is exploratory and provides a number of null findings that must be interpreted with caution. PDE4 loss was not associated with disease duration (average 11 years, range 3–25) or with levodopa equivalent daily and lifetime dose. No age-related differences emerged in the healthy controls. At face value, these findings suggest that PDE4 loss may occur very early in PD, and introduce the possibility that this loss is a predictive marker for cognitive decline and a potential target for future therapies. It would be interesting to use [11C]rolipram PET to investigate longitudinal and treatment-related changes in brain PDE4 expression. This could strengthen the case for this enzyme’s role in PD pathophysiologic and clinical progression, and shed light on novel mechanisms by which therapeutic effects are achieved.
Whether disease-specific or not, the relationship between PDE4 loss and spatial working memory deficits is striking because of its magnitude, and because the regional distribution of PDE4 loss clearly tracks striato-thalamo-cortical circuitry. Spatial working memory deficits are part of the frontostriatal dysexecutive syndrome in PD.6 The emergence of a novel neurochemical correlate of these deficits might extend the understanding of the neural underpinnings beyond dopamine dysmodulation. It will be important to determine whether PDE4 loss is specific to working memory or also plays a role in other executive processes characteristic of this syndrome. Equally important will be the systematic evaluation of the relationship between PDE4 expression and the prediction and development of the cognitive impairments associated with PDD. For now, we note that the current finding that PDE4 loss is not related to episodic memory is in line with the widely accepted view that working and episodic memory have different cellular and neural substrates. If this pattern of findings persists in larger-scale studies, a more thorough evaluation of working memory than currently recommended10 may be needed in PD-MCI diagnosis. It may be that this distinction will guide the development of different treatment strategies.