Cognitive impairment in Parkinson's disease without dementia

Some degree of cognitive impairment appears frequently in Parkinson's disease (PD) patients, even at the onset of the disease. However, due to the heterogeneity of the patients and the lack of standardized assessment batteries, it remains unclear which capacities are primarily affected by this disease. Fifty PD patients were assessed with 15 tests including executive functions, attention, temporal and spatial orientation, memory, and language tasks. Their results were compared with those of 42 age‐ and education‐matched healthy seniors. Semantic fluency, along with visual search appeared to be the most discriminant tasks, followed by temporal orientation and face naming, as well as action naming and immediate recall. PD patients studied showed an impairment of frontal‐ to posterior‐dependent capacities. Executive functions, attention, and recall tasks appeared to be significantly impaired in the patients. Nevertheless, significantly poor scores in tasks like action and face naming, as well as semantic fluency, also reveal a mainly semantic deficit. © 2010 Movement Disorder Society


Introduction
Though there is some heterogeneity in previous data, it is agreed that some degree of cognitive impairment is present in a considerable number of patients suffering from Parkinson's disease (PD) [1][2][3][4][5][6] . Thus, whereas early studies based on DSM-III criteria had found that 14% of PD patients present dementia 1 , more recent work estimates prevalence rates around 30% [2][3][4][5][6] . Furthermore, when fine-grained neuropsychological and cognitive measures are used, a certain level of cognitive impairment can be identified even in some PD patients who cannot be diagnosed with dementia yet 2,6,7 . The diagnostic label Mild Cognitive Impairment (MCI) 8 , borrowed from the domain of Alzheimer's disease (AD), has been applied to these patients with the intention of describing an intermediate state between preserved cognitive capacity and fully-established dementia in PD 2,6 . Given that MCI is considered to be a precursor of dementia 6 and one of the best predictors of its appearance 9, 10 , it is imperative to conduct a formal neuropsychological examination of PD patients, in order to ascertain their degree and pattern of cognitive preservation over the course of the disease.
However, given that normative data for extensively used scales, like MMSE or the Mattis Dementia Rating Scale, is incomplete in PD populations, along with the fact that the cognitive profile in PD with dementia differs significantly from that in AD 11 , specific tools must be designed to assess the cognitive decline in PD. Several specific tests have reported their usefulness in PD but the lack of concordance 12 and absence of studies comparing the performance of PD without dementia patients with controls lead to question their validity.
The most common deficits described are related to executive functions 3, 4, 6, 13-15 , although visuospatial 14,16,17 , attentional 14,18 , memory 3,4,6,19,20 , and language 15,21 impairments have also been observed in this kind of patients. Some of these disorders have been detected even in the early stages of the disease 22,23 , while others appear in later stages. In an attempt to put order to the cognitive heterogeneity present in PD patients, signs of cognitive impairment have been proposed to cluster into two main groups 24 . Thus, whereas some of the PD patients present reduced attention and executive capacities, others show a pattern of impairment that is primarily characterized by memory loss. These two cognitive impairment profiles would be the result of frontal lobe-and temporal lobe-dependent pathology respectively 24 . However, it has also been suggested that the executive deficit is the primary cause of the other symptoms including the language and memory impairments 17,19 .
The identification of preclinical dementia in PD patients provides an opportunity to understand cognitive decline in PD and is, thus, crucial for accurate management and prompt treatment. Selecting the cognitive functions to be included in an assessment battery is the first step to build a rational screening test. Obviously, this selection should include the ones that have been affected since the beginning of the disease. Moreover, it is important to assess a variety of cognitive abilities that cover a full range of anterior to posterior-dependent functions, in order to disentangle the contribution of executive function to the expressed deficits.
In this study we conducted a systematic evaluation of a range of cognitive functions including orientation, attention, executive function, memory and language. In some cases, several tasks were designed for each capacity in order to be able to establish fine-grained conclusions regarding the exact origin of the deficit. PD patients without dementia were compared to controls of similar demographic characteristics with the aim of detecting cognitive deficits before the dementia stage. were used to discard dementia. A summary of participant's characteristics is provided in table 1. Table 1 here

Stimuli and Procedure
Participants were tested with a battery consisting of 15 tasks. The tests were aimed to study those cognitive abilities that are known to be impaired in PD patients, like executive function and attention, but also other capacities like temporal and spatial orientation; semantic and episodic memory; and language production and comprehension. The tasks were: -Visual search: The participant must mark the number "3" every time it appears on a 12 x 20 grid of the numbers 1 through 9.
-Odd-one-out: Groups of four photographs are presented to the participant, who is asked to indicate which of them is different from the others. Three different criteria are used: category (e.g., three apples and a pear), color (e.g., three yellow flowers and one red flower) and number (e.g., three single gloves and a pair of gloves).
-Temporal Orientation: The participant is presented with five questions regarding placement in time (e.g., What day of the week is it? What year is it?).
-Spatial Orientation: The participant is asked to answer five questions about spatial location (e.g., What city are we in? What street do you live on?).
-Immediate recall: The participant is asked five questions about an imaginary friend (e.g., What's my friend's name? How old is he?) that is previously described to him (My friend's name is Arthur, he is 46 years old, …).
-Delayed recall: The participant is asked the same questions about the imaginary friend several minutes after the immediate recall test.
-Semantic fluency: For 60 seconds, the participant is asked to name as many fruits as he or she can without repeating any.
-Semantic association: The participant is asked to choose which of two drawings presented at the bottom of a page (e.g., a hammer and a padlock) is related to the one presented at the top (e.g., a key).
-Object naming: Ten color drawings of objects are presented, half of them living things and half inanimate objects, for the participant to name.
-Action naming: Ten color drawings of actions are presented for the participant to name.
-Famous faces naming: The participant has to name ten photographs of famous people presented one by one.
-Definition-word pairing: The participant must show which of the four presented words (e.g., vice, insult, swindle, prayer) corresponds to a given definition (e.g., deception with intent to profit).
-Synonymy: A word (e.g., plea) is presented at the top and another three (e.g., prayer, request, fervor) appear at the bottom of a sheet. The participant must indicate which of these three words is synonymous with the word at the top.
-Phonological fluency: For 60 seconds, the participant must name as many words as he or she can that begin with the letter "f" and are not proper nouns.
-Pseudoword repetition: The experimenter articulates ten pseudowords, each with three or four syllables, one by one for the participant to repeat. One point is awarded for each "correctly pronounced" pseudoword.
Each participant was run through the experimental battery individually in two sessions scheduled two or three days apart. Table 2   MANCOVA analyses were followed up by a discriminant analysis (see table 2) conducted with diagnosis as the dependent variable and the different tasks as independent variables. The test battery appeared to correctly distinguish 80.4% of the cases, with positive and negative predictive values of 76% and 86% respectively. ROC curves were also computed for each of the comparisons, yielding the area under the curve values also reported in table 2. Semantic fluency, along with visual search appeared to be the most discriminant tasks, followed by temporal orientation, face naming, as well as action naming and immediate recall.

Discussion
In order to explore the profile of cognitive impairment in PD, a group of 50 nondemented PD patients was assessed with a neuropsychological battery including memory, language, attention, orientation and executive function tasks. In line with previous results in the literature 3, 4, 6, 13-15 the presence of an executive function deficit in PD patients was confirmed in the sample studied here, as significant differences appeared between the scores of the PD and control groups in the odd-one-out test. Yet again, according to previous studies 3, 14, 18 PD patients presented an attention deficit that became apparent in their poor scores in the visual search task. Temporal orientation also appeared to be impaired in the PD group, with patients scoring significantly lower than healthy controls. Spatial orientation, however, was preserved in PD patients, who obtained scores similarly high to those of the control group. Visual search, odd-one-out and temporal orientation turned out to be amongst the four most discriminant tasks in our study. Furthermore, differences between the groups' scores in temporal orientation and visual search appeared to be significant even when global cognitive impairment was controlled entering MMSE scores as a covariable in the analysis. Generally, this pattern of results is congruent with the well established idea 17,19 that PD patients present a deficit in frontal lobe-dependent capacities.
Concerning language, oral production difficulties can be ruled out in the PD group, as no significant differences appeared between their scores and those of control participants either in the phonological fluency or in the pseudoword repetition tasks. Reading comprehension also appeared to be preserved in the patient group, with scores comparable to those of healthy seniors in both the synonymy and definition-word pairing tasks.
The episodic memory of PD patients has previously been compared to that of patients suffering from AD. Immediate recall abilities of PD patients have been shown to be similar to those of AD patients 14 but better preserved in the case of delayed recall 28,29 .
Accordingly, PD patients in our study produced significantly less correct responses than healthy seniors in the immediate recall task, although differences disappeared in the delayed recall test. Due to an alleged preservation of recognition abilities of PD patients 30,31 , their episodic memory impairments have been ascribed to retrieval deficits secondary to a general executive impairment 17 . However there is compelling evidence that PD patients do suffer from a recognition deficit 32 , so coding difficulties cannot be ruled out as a possible cause of the episodic memory impairment of these patients.
With regards to semantic memory, significant differences appeared in the semantic fluency task, suggesting a relative impairment of this capacity that, nevertheless, was not apparent in the semantic association task. This was probably due to a ceiling effect, as both groups obtained scores close to the maximum in the task. Semantic fluency is agreed to be one of the most sensitive task to detect cognitive impairment in PD patients 15,17,19 . This is confirmed in our study, in which semantic fluency together with the visual search task obtained the greatest area under the ROC curve values. Similarly to what happens in the case of recall tasks, even though semantic fluency deficits have repeatedly been reported in previous studies 15,19 , it has been suggested that, rather than having a purely semantic origin, the poor scores obtained by PD samples in this task reveal a deficit of executive functions 14,19 . This assumption usually relies in the presence of deficits in the phonological fluency task, that is very similar to semantic fluency but with no semantic component 17 .
However, there is also evidence that PD patients are more impaired on tests of semantic than phonemic fluency 15,33 . As no significant differences appeared between the scores of the PD and control groups in the phonological fluency task in our study, we are inclined to suggest that there is indeed a memory component in the semantic fluency deficit present in patients suffering from PD. Moreover, there is increasing evidence that the pattern of atrophy present in PD extends to medial temporal regions 34,35 , what would explain the appearance of a memory-specific impairment.
Special attention should be paid to the three naming tasks that allow us to assess different aspects of the semantic memory of the patients. No significant differences appeared between the scores of the PD and control groups in the object naming task.
Nevertheless action and face knowledge turned out to be significantly impaired in PD patients, making face and action naming appear amongst the most discriminant tasks.
Previous studies had noted the appearance of a relative impairment of action knowledge in PD participants 36,37 . This action-verb deficit has been taken as evidence of the crucial role of movement-related neural structures in the representation of semantic knowledge about verbs. The results of our study replicate these findings adding to the hypothesis that frontal lobe deficiencies influence the semantic impairment present in PD patients. On the other hand, PD patients in our study also presented a face naming deficit. Face processing has consistently been associated to the fusiform area, in the ventral region of the temporal lobe [38][39][40][41] , and has been suggested to be a useful predictor of dementia in prodromal and initial stages of Alzheimer disease 42 . The results of our study suggest that assessment of face processing should also be conducted in the PD population in order to fully understand their cognitive profile.
Together with the results of previous studies, the experiment presented here points out the existence of a cognitive deficit in PD patients that affects anterior-to posteriordependent capacities. Although executive functions, attention and recall tasks appear to be amongst the most sensitive tests to detect cognitive impairment in PD patients, also mainly semantic tasks, like action and face naming, as well as semantic fluency, seem to be good indicators of this condition. Since MCI is believed to be a precursor of dementia 6 , and given that different profiles of impairment have been suggested to predict different varieties 43 , a wide range of tasks should be applied to these patients in order to fully understand their cognitive deficit.