Journal of Neurology

, Volume 260, Issue 9, pp 2228–2236

Persistence of associations between cognitive impairment and motor dysfunction in the early phase of Parkinson’s disease

Authors

    • Department of Pharmacology and Clinical Neuroscience, NeurologyUmeå University
  • Lars Forsgren
    • Department of Pharmacology and Clinical Neuroscience, NeurologyUmeå University
  • Eva Elgh
    • Department of Community Medicine and Rehabilitation, Geriatric MedicineUmeå University
Original Communication

DOI: 10.1007/s00415-013-6971-6

Cite this article as:
Domellöf, M.E., Forsgren, L. & Elgh, E. J Neurol (2013) 260: 2228. doi:10.1007/s00415-013-6971-6

Abstract

The relation between cognitive and motor functions in Parkinson’s disease is not fully understood. In an incidence population of newly diagnosed drug naïve patients with Parkinson’s disease, associations were found between the degree of bradykinesia and postural instability and gait disturbances, measured by the Unified Disease Rating Scale, and different types of cognitive functions. To investigate the stability of these associations over time, we explored the association of differences between baseline and 1-year follow-up in 91 incident cases with Parkinson’s disease. The magnitude of change between the two assessments was assessed together with analysis of differences based on which dopaminergic medication was used. Change in bradykinesia was associated with change in working memory and mental flexibility. Changes in postural instability and gait disturbances were associated with change in visuospatial memory. A negative effect of the dopamine agonist pramipexole on phonemic fluency performance was found compared to treatment with other dopaminergic drugs. Change in cognitive and motor functions were associated from time of diagnosis until 1 year after diagnosis. These persisting findings strengthen results from a previous cross-sectional study suggesting similar associations. The effects of dopamine agonists on phonemic fluency should be investigated further.

Keywords

Parkinson’s diseaseCognitionPopulation-basedProspectiveDopamine agonist

Introduction

Parkinson’s disease (PD) is a neurodegenerative disorder where deterioration of dopaminergic neurons in the substantia nigra projecting to the basal ganglia results in two of the classical motor features of the disease: bradykinesia and rigidity. Tremor in PD is proposed to be a result of actions from two circuits, a dopaminergic striatopallidal circuit and a cerebello-thalamo-cortical circuit [1], and may also involve serotonergic dysfunction [2]. Axial impairments in the disease, gait and postural instability may be due to affliction of other neurotransmitter systems and not caused by pure dopaminergic failure [3, 4]. Non-motor symptoms are also frequent in PD. These include disturbances in cognitive, affective, sensory, sleep, mood and autonomic functions. Cognitive decline is common in early stages of the disease and studies report a range of cognitive domains being affected [5, 6]. Different motor phenotypes have been associated to Parkinson's disease dementia (PDD). Postural instability and gait difficulties (PIGD) have been connected to a faster rate of cognitive decline [7], while tremor dominant disease has been connected to a more benign course in early stages of the disease [8].The connections between different motor functions and different cognitive domains have been investigated in cross-sectional studies and some evidence exists for the association between bradykinesia and cognitive domains with a more frontal/sub cortical demand [9] as well as for axial signs and episodic memory and visuospatial function [9, 10]. What is less clear is how these associations develop over time and how they are affected by treatment with dopaminergic medication.

It is well-established that dopamine is involved in cognition [11], especially in working memory-related processes. Dopamine replacement therapy has shown both beneficial and aggravating effects on different cognitive functions [12]. According to Gotham et al. [13], dopaminergic medication improves functions with severe dopaminergic depletion such as in the dorsal striatum, but overdoses in areas with less dopamine depletion in early stages of the disease, such as the prefrontal cortex (PFC) or the ventral striatum. Different types of treatment, i.e. levodopa or dopamine-agonists, have also shown different, sometimes even opposite effects on cognitive functions where short-term use of the dopamine agonist pramipexole has been suggested to cause decline in short-term verbal memory, attentional-executive functions and verbal fluency [14]. Levodopa is converted to dopamine presynaptically with subsequent effects postsynaptically, where it binds to both D1 class receptors (including D1 and D5) and D2 class receptors (including D2, D3 and D4). Dopamine agonists on the other hand act directly on the postsynaptic system. The commonly used non-ergot dopamine agonists, such as pramipexole and ropinirol, have high affinity only for D2 class receptors [15], where pramipexole has a higher affinity for D3 receptors and ropinirol for D2 receptors [16]. Thus, there may be fundamental differences in the functional effects of different dopaminergic drugs. Several studies have evaluated the short-term effect of dopaminergic medication on cognitive function, but what is less investigated is the mid- and long-term effect. It is therefore unclear how continuous treatment over many months with dopaminergic medication affects cognitive functions in PD.

The aims of this paper were to investigate cognitive change after the first year following diagnosis in an incidence cohort of PD; to investigate the effect of different types of dopaminergic medication on cognition; and to explore stability of associations between motor and cognitive functions over time by analyzing changes in motor functions after the first year following diagnosis, in relation to changes in different cognitive functions.

Method

The data used in this study are from the NYPUM-project [NY (new) Parkinsonism in UMeå] [17]. The NYPUM-project is a prospective population-based study on idiopathic parkinsonism in a local catchment area of Umeå University Hospital with around 142,000 inhabitants. All incident cases with suspected parkinsonism were referred to the Department of Neurology at the Umeå University Hospital, where all neurologists in the catchment area worked. To ensure that as many cases would be identified, a letter was written twice yearly to all general practitioners, geriatricians, neurosurgeons, psychiatrists, internists, private practitioners and to company health services in the area informing them of the study and asking for referral of suspected cases. Patients are followed up yearly with repeated examinations, up to 8 years from diagnosis. Neuropsychological assessment is performed within 1–2 months of inclusion (baseline) and after 1, 3, 5 and 8 years. Baseline and 1-year follow-up data for the patients included during the inclusion period (January 1, 2004 to April 30, 2009) are included in the present study.

Procedure

Global cognition was measured with the Mini Mental State Examination (MMSE). Depression was assessed by the Montgomery and Åsberg Depression Rating Scale (MADRS); participants with scores over eight were considered to have a mild depression and severe depression if scores were 18 or over [20]. Dopamine transporter (DAT) single-photon emission computed tomography (SPECT)-imaging using 123I-Ioflupane ([123I]FP-CIT) was performed within 1–2 months following inclusion. The levodopa equivalent dose (LED) was calculated based on the conversion factors suggested by Thomlinson et al [21]. The severity of parkinsonism was measured by the Unified Parkinson’s Disease Rating Scale (UPDRS) [22] and the Hoehn and Yahr staging. Variables of the different motor scores were calculated from the UPDRS: the sum of UPDRS items 20 and 21 for tremor, item 22 for rigidity, the sum of items 24, 25, 26, and 31 for bradykinesia, the sum of items 27, 28, 29, and 30 for the posture and gait score (arising from chair, posture, gait and postural stability) [23], and the sum of items 18 and 19 for the bulbar score (speech and facial expression) [24]. The study was approved by the Ethics Committee of the Faculty of Medicine at Umeå University. Written, informed consent was obtained from all participants.

Participants

Only patients with previously undiagnosed idiopathic parkinsonism were included, i.e., patients with bradykinesia and at least one additional cardinal sign: resting tremor, rigidity or impaired postural reflexes. They were classified against diagnostic criteria for PD, multiple system atrophy, progressive supranuclear palsy and corticobasal degeneration [18] by movement disorder specialists. A total of 186 cases with idiopathic parkinsonism were identified. At follow-up 3–8 years after inclusion, 150 patients fulfilled the diagnostic criteria for PD, 31 had atypical parkinsonism and five were unclassifiable according to the UK Parkinson’s Disease Society Brain Bank criteria (UK PDSBB). Patients with dementia, or cognitive dysfunction (MMSE < 24) at baseline or onset of dementia within 12 months of the onset of parkinsonism were not included [19].

One hundred and twenty-one patients (81 %) of the PD sample agreed to the neuropsychological investigation at baseline. Six patients were excluded due to either not having enough tests performed (n = 1), having dementia at baseline (n = 1), a normal SPECT FP-CIT scan (n = 3) or severe depression according to MADRS (n = 1). Patients not participating in the neuropsychological baseline evaluation were older (mean 79 vs 69 years, p < 0.001) and had higher scores on the UPDRS (mean 35 vs 26, p < 0.001). Of the remaining 115 patients, 88 (76 %) agreed to participate in the follow-up 12 months later. Patients not participating in the neuropsychological follow-up were older (mean: 72.6 vs 67.7 years, p = 0.014) but did not differ significantly in any other way. Three patients participating at baseline died before follow-up. The group was divided into what type of medication was predominantly used, i.e., treatment with the dopamine agonist pramipexole ± levodopa vs treatment without dopamine agonists. Twelve patients had started their dopaminergic drug treatment before the baseline investigation and were excluded from the analysis of effects of different dopaminergic medication on changes in cognitive scores during follow-up. Treatment with dopamine agonists was by pramipexole in all but two patients who received ropinirol. These two patients were also excluded from the medication analysis. Five patients did not receive either levodopa or dopamine agonists at follow-up. Demographics, baseline data and dopaminergic treatment at 12 months for the 88 participants and the 74 drug-naïve patients at baseline are shown in Table 1.
Table 1

Demographics and baseline measures of the total group (n = 88) and the 74 patients drug-naïve at baseline, by type of dopaminergic treatment

 

Total, mean ± SD (n = 88)

Total, mean ± SD (n = 74)

Pramipexole, mean ± SD (n = 34)

No agonist, mean ± SD (n = 40)

p value

Sex (men/women)

57/31

48/26

21/13

27/13

0.607

Age

67.5 (9.3)

67.5 (9.2)

62.0 (7.5)

72.2 (7.8)

<0.0001**

Years of education

10.2 (3.8)

10.3 (3.8)

11.8 (4.0)

8.95 (3.1)

0.001**

MMSE

28.8 (1.2)

28.8 (1.3)

28.9 (1.2)

28.7 (1.3)

0.464

UPDRS III

24.8 (10.2)

24.3 (10.3)

23.7 (10.2)

24.9 (10.5)

0.620

UPDRS II

8.1 (2.8)

7.9 (2.8)

7.5 (2.8)

8.2 (2.7)

0.262

UPDRS I

1.5 (1.4)

1.5 (1.4)

1.5 (1.5)

1.6 (1.3)

0.750

MADRS

4.2 (3.3)

4.4 (3.2)

4.1 (3.7)

4.7 (2.8)

0.408

LED (12 month)

331 (174)

317 (173)

312 (172)

320 (175)

0.486

Medication

  l-dopa

62 (70 %)

50 (68 %)

15 (44 %)

35 (88 %)

 

 Pramipexole

39 (44 %)

34 (46 %)

34 (100 %)

0

 

 Selegeline

6 (7 %)

4 (5 %)

2 (8.8 %)

2 (5 %)

 

MMSE Mini mental state examination, UPDRS Unified Parkinson’s disease rating Scale, MADRS Montgomery and Åsberg depression rating scale, LED Levodopa equivalent dose, MCI Mild cognitive impairment

** p value < 0.001

Neuropsychological assessment

The participants were tested for 2 h each at baseline. At the 12-month follow-up, a shorter battery of tests was administered for approximately 1 h. Tests included in both baseline and the 12-month follow-up were as follows. Free and Cued Selective Reminding test (FCSRT) [26] is a verbal episodic memory test consisting of 16 items with both free and cued recall. The recall session is to be repeated three adjacent times and a delayed recall is performed after 20 min. Each trial is scored for the number of immediate and delayed free and cued recalled items and the sum of these scores. Brief Visuospatial Memory Test—revised (BVMT) [27] measures non-verbal episodic memory. The participant is presented with six items and is asked to draw each item on a blank sheet immediately after the presentation. This procedure is repeated three times. After 25 min the participant is asked to draw the items again and this is followed by a yes/no recognition task. The scoring accounts for the locations as well as the accuracy of figure drawings. Digit Span [28] is a measure of working memory and consists of lists of random digits presented auditorily. The participant is asked to recall the digit lists in correct serial order (forward and backward). Trail Making Test (TMT) [29] tests visual search, attention, mental flexibility and motor function. It consists of two parts, A with numbers and B with numbers and letters. Controlled Oral Word Association Test (COWAT) [30] evaluates spontaneous production of words beginning with a given letter (FAS), phonemic fluency or a given category (fruit, animal and color), and category fluency, within 1 minute. Benton Judgement of Line Orientation Test (JOLOT) [31] tests visuospatial function. Mental Control [28] is a short test of conceptual tracking. Wisconsin card sorting task (WCST), Boston Naming Test (BNT), associative learning and logical memory from Wechsler memory scale were not included in the 12-month follow-up because they did not have alternate forms and therefore were evaluated as likely to show practice effects after 12 months [25].

Statistics

Assessing correlations and linear-regressions for the total group (n = 88)

For correlations, the non-parametric Spearman's rho test was used to investigate associations of the differences between baseline and 12-month follow-up in the motor and cognitive variables. The variables in the correlation analysis were based on the 12-month follow-up score subtracted from the baseline score for motor and cognitive variables. For the cognitive variables, except for TMT A and B, a negative score means improvement. For the motor variables, a negative score means a decline in function. A negative correlation between motor and cognitive scores (except for TMT A and B) suggests that they change in parallel, i.e. an improvement in motor function means an improvement in cognitive function. For TMT A and B a positive correlation means that they change in parallel. The cognitive variables that had shown a cognitive-motor relationship in our previous paper [9] and the cognitive and motor variables that correlated in the difference between baseline and the 12-month follow-up were used as the dependent variables in separate linear regressions. The motor variable that correlated with the cognitive variable of interest was used as independent variable. Age at baseline, years of education, LED and baseline score of the dependent variable were used as covariates. The residuals of the regression models were explored to see if they fulfilled the assumptions of normality, linearity and homoscedasticity. Because of the theory driven comparisons we did not statistically correct for multiple comparisons [32]. P values under <0.05 were considered significant.

Assessing differences between baseline and 1 year follow-up for patients drug-naïve at baseline (n = 74)

Paired sample t tests were conducted between the baseline assessment and the 12-month assessment for motor and cognitive variables. For the variables that did not have normal distribution, Mann–Whitney U-tests were performed. For the dichotomous variables, Chi square or Fisher's exact tests were performed. To explore if the change in performance between assessments differed between groups based on medication (dopamine agonist pramipexole ± levodopadopa vs no dopamine agonists), univariate analyses of variance were conducted for cognitive variables that have been shown to be affected by pramipexole in a previous study (FCSRT recognition, verbal fluency, category fluency, TMT A and TMT B) [14]. The follow-up score was the dominant variable; medication was entered as between subject factors. Finally, the baseline score of the variable of investigation, age, years of education and LED were entered as covariates. The covariates were chosen because they were likely to influence the direction of motor and cognitive score at follow-up or because they differed significantly between groups. Interaction effects between medication and the other covariates were checked before the full factorial models were executed. The proportions that declined in the different medication groups were calculated and a Chi square test was performed to evaluate if the groups differed (two or more words less at follow-up vs those with improvement or smaller change). U-shaped associations between motor and cognitive variables were checked separately for each medication group. The U-shaped association between change in cognitive and motor variables were statistically checked for in a regression model with the cognitive variable as dependent variable, squared change in motor score (motor2) as independent variable and change in motor score as covariate. P-values under 0.05 where considered significant. A p value of 0.05/5 = 0.01 would have been considered significant if a Bonferroni adjustment had been applied.

Results

Results for correlations and linear-regressions for the total group (n = 91)

Differences between baseline and 1-year assessments for bradykinesia correlated with differences in working memory (digit span), mental flexibility (TMTB) and visuospatial memory (BVMT). Change in postural instability and gait problems correlated with changes in visuospatial memory (BVMT) (Table 2). The linear regressions controlling for age, years of education and LED showed that change in postural stability and gait explained change in visuospatial memory. Change in bradykinesia explained change in working memory (Table 3).
Table 2

Correlations (Spearman’s ρ) between changes in motor variables and changes in cognitive variables between baseline and 1-year follow-up in 88 patients with Parkinson’s disease

 

Sex

Age

Educ

LED

MADRS

Tremor

Brady

Rigidity

PIGD

Bulbar

Sex

         

Age

0.01

        

Years of edu

−0.02

−0.59

       

LED

0.05

−0.02

−0.08

 

0.14

0.22

0.23

0.07

0.02

MADRS

−0.05

−0.19

0.13

0.15

0.20

0.11

0.07

0.19

0.07

FCSRT-free recall

0.11

−0.07

0.12

0.04

−0.02

−0.10

−0.08

−0.11

−0.14

−0.04

BVMT total

−0.07

0.30*

−0.11

−0.08

−0.09

−0.12

−0.22

−0.10

−0.28*

−0.22

Line orientation

0.16

0.14

−0.10

−0.14

−0.22

−0.12

−0.12

−0.01

−0.09

−0.09

Digit span

−0.03

0.13

−0.12

−0.02

−0.11

0.13

0.25

−0.07

−0.18

−0.07

TMT A

0.15

0.12

−0.11

−0.02

−0.06

−0.15

0.10

−0.15

−0.00

−0.11

TMT B

−0.05

0.06

−0.11

0.17

0.04

−0.04

0.27*

−0.07

0.11

−0.02

Phonemic f

−0.03

−0.16

0.06

0.02

0.03

−0.09

0.14

−0.09

0.07

0.00

Category f

−0.09

−0.16

−0.03

−0.20

−0.14

0.05

0.09

−0.04

−0.08

−0.02

Mental control

0.00

0.04

−0.05

0.03

0.06

0.06

0.17

−0.10

−0.00

0.17

Edu education, LED levodopa equivalent dose, MADRS Montgomery and Åsberg Depression Rating Scale Brady = bradykinesia, PIGD postural imbalance and gait disturbance, BVMT Brief visuospatial memory test, FCSRT Free and Cued Selective Reminding Test, TMT A Trail Making Test part A, TMT B Trail Making Test part B, f fluency

Correlations in BOLD displays p values < 0.05, * p values < 0.01

Table 3

Linear regression to assess changes in motor variable’s ability to predict changes in cognitive measures after controlling for baseline age, years of education and levodopa equivalent dose

 

Standardized B

p-value

95 % CI for B

R2 for model

p value for model

PIGD

     

 BVMT total

−0.272

0.014*

−1.601 to −0.190

0.082

0.031*

 Line orientation

−0.099

0.381

−0.658 to 0.254

−0.009

0.512

Bradykinesia

     

 Digit span

−0.300

0.007**

−0.311 to −0.050

0.072

0.042*

 TMT B

0.243

0.038*

0.212 to 7.315

0.046

0.105

 BVMT total

−0.160

0.159

−0.496 to 0.082

0.039

0.134

Digit span forward (n = 34)

   

0.195

0.015*

  Bradykinesia2

−0.473

0.014*

−0.047 to −0.006

  

  Bradykinesia

−0.042

0.818

−0.159 to 0.126

  

PIGD postural imbalance and gait disturbance, BVMT Brief visuospatial memory test, FCSRT Free and Cued Selective Reminding Test, TMT B Trail Making Test part B, Bradykinesia2 statistics based on only patients receiving pramipexole

* p value < 0.05, ** p value < 0.01

Differences between baseline and 1-year follow-up for patients drug-naïve at baseline (n = 74)

No significant changes except for a small improvement in MMSE score were found between baseline and follow-up in any of the cognitive variables (Table 4). The motor symptoms improved between the assessments and significantly so for UPDRS III total, tremor and posture/gait sub-scores. Phonemic fluency changed differently between patients receiving the dopamine agonist pramipexole compared to those who did not [F (1.67) = 5.244 p = 0.025, partial eta square = 0.073, ANCOVA]. Patients receiving pramipexole performed worse at the 1-year follow-up (baseline mean: 43.7 words, follow-up mean: 38.5 words) and the patients that did not showed slight improvement between assessments (baseline mean: 36.2 words and follow-up mean: 37.5 words). None of the other cognitive variables changed differently between medication groups. There were no interaction effects between medication group and the covariates (age, education, LED and baseline score of the variable of interest). In the pramipexole ± levodopa group, 74 % produced two or more words less at follow-up, while this occurred in only 31 % of the group not treated with dopamine agonists (Chi square, p < 0.0001). The decline pattern was the same for patients using pramipexole as monotherapy as for those using it together with levodopa (Fig. 1). For patients receiving pramipexole, the correlation between bradykinesia and digit span forward was U-shape. That is, patients with less bradykinesia at follow-up either improved or deteriorated in their working memory performance during the same period (Fig. 2). Bradykinesia2 was significantly associated with change in digit repetition forward in patients receiving pramipexole (table 2), which supports the U-shaped relationship.
Table 4

Pair-wise comparisons between assessments at baseline and 12 months for cognitive and motor measures in patients with PD who were drug-naïve at baseline (n = 74)

 

Baselinea

Follow up a

p value

MMSE

28.8 (1.25)

29.2 (1.04)

0.025*

MADRS

4.37 (1.04)

3.34 (2.87)

0.017*

FCSRT free recall

25.3 (7.03)

25.8 (7.46)

0.293

BVMT total

18.5 (6.94)

17.7 (7.75)

0.141

Digit span total

13.9 (3.49)

13.7 (3.13)

0.563

TMT A(s)

56.6 (26.3)

54.3 (24.3)

0.336

TMT B (s)

151.7 (76.9)

156.8 (82.7)

0.469

Mental control

6.2 (1.87)

6.1 (2.10)

0.503

Phonemic fluency

39.7 (15.7)

37.9 (15.1)

0.078

Animal fluency

17.1 (5.30)

18.0 (5.41)

0.061

Line

24.1(3.42)

24.4 (3.70)

0.409

UPDRS III

24.4 (10.4)

22.0 (9.69)

0.014*

Tremor

2.50 (2.11)

1.49 (1.67)

<0.001**

Bradykinesia

8.95 (4.40)

8.68 (4.21)

0.543

Rigidity

5.73 (4.21)

5.43 (3.57)

0.313

Posture/gait

2.36 (1.62)

1.88 (1.57)

0.008**

Bulbar

1.77 (1.31)

1.70 (1.25)

0.665

MMSE Mini Mental State Examination, UPDRS Unified Parkinson’s Disease Rating Scale, MADRS Montgomery and Åsberg Depression Rating Scale, LED Levodopa equivalent dose, BVMT Brief Visuospatial Memory Test, FCSRT Free and Cued Selective Reminding Test, TMT A & B Trail Making Test part A & B

*p value < 0.05, **p value < 0.01

a mean (SD)

https://static-content.springer.com/image/art%3A10.1007%2Fs00415-013-6971-6/MediaObjects/415_2013_6971_Fig1_HTML.gif
Fig. 1

Error bars for different groups based on dopaminergic medication at baseline (bold bars) and follow-up (thinner bars) for verbal fluency (FAS). Y axis 95 % Confidence Interval of Verbal fluency (FAS), X-axis grouping of sample based on medication

https://static-content.springer.com/image/art%3A10.1007%2Fs00415-013-6971-6/MediaObjects/415_2013_6971_Fig2_HTML.gif
Fig. 2

Scatter plot of the difference between baseline and follow-up for 74 patients with PD (drug-naïve at baseline). x and dashed line represents patients receiving the dopamine agonists pramipexole (DA ± l-dopa), + and full line represents patients not receiving dopamine agonists. The X axis represents change in digit span where positive scores mean decline and negative scores mean improvement. The Y axis represents changes in bradykinesia where positive score means improvement and negative score indicates more bradykinesia

Discussion

We found that change in postural instability and gait between baseline and 1-year follow-up was associated with change in visuospatial memory. Change in bradykinesia was associated with change in working memory and mental flexibility. This association differed between patients treated with pramipexole with and without levodopa, compared to those not treated with dopamine agonists. For all significant associations, improvements in motor function were associated with an improvement in cognitive function. Further, we found that treatment with the dopamine agonist pramipexole resulted in decline in phonemic fluency after 1 year.

The significant correlation found between PIGD and visuospatial memory supports the result from our baseline study [9] suggesting a connection between postural instability/gait disturbances and visuospatial memory. By showing that they also are correlated in their change in performance 1 year following diagnosis, we strengthen the assumptions of common mechanisms or pathology behind postural instability/gait and visuospatial memory in PD. Postural and gait disturbances in PD have been connected to cholinergic denervation of the pedunculopeptine nucleus (PPN) [3]. The same authors suggest that cortical cholinergic denervation is related to decreased performance of executive function and attention [33]. In our study the only cognitive domain that was associated with postural instability and gait in the change between baseline and follow-up was measure of visuospatial memory. There is evidence of the importance of visual information in generating motor plans [34] and in the control of locomotion [35] that could partly explain the association between postural instability and gait and visuospatial function. An alternative view on the associations between motor and cognitive functions can be that they reflect widespread neurodegeneration that affects structures related to movement and cognition. Countering this view is the idea that different types of motor functions seem to be associated with different types of cognitive functions, both cross-sectional and in evolvement over 1 year.

Change in bradykinesia was associated with change in working memory and mental flexibility, which strengthens the baseline association between bradykinesia and working memory. We found a U-shaped relationship between bradykinesia and working memory in patients receiving pramipexole. Patients with less bradykinesia at the 1-year follow-up either performed worse or better on digit span forward. The relationship between working memory and dopamine level is suggested to follow an inverted U-shape, where both excessive and insufficient dopamine levels impair working memory performance [36]. Dopaminergic medication has been suggested to lead to decline in working memory performance for patients with well-preserved dopamine-levels in the ventral striatum and PFC and improvements in working memory performance for those with lower dopamine levels in the same areas. COMT genotype as well as other relevant genes involved in DA transmission might be contributing factors for the variable relationship between bradykinesia and working memory performance [37]. Bradykinesia has been described as the best measure of the nigrostriatal lesion and is related to dopaminergic depletion in the dorsal striatum that is severely dopamine depleted early in the disease [38]. The U-shaped association between digit span forward and bradykinesia for patients receiving pramipexole could support the overdose hypothesis and perhaps indicate that pramipexole is more likely to impair working memory by “overdosing” than other types of medication. Pramipexole has a high affinity for D3 receptors that have a high density in ventral striatum [39]. Ventral striatum is less dopamine depleted early in the disease, which could be a possible explanation for the discrepancy between medication groups.

A possible explanation for the negative effect of pramipexole on verbal fluency could be that D3 auto receptors contribute to the presynaptic regulation of tonically released dopamine. This leads to a moderate inhibitory action on locomotion [16] and possibly even on cognitive performance. Since we did not compare different dopamine agonists, we cannot tell whether our finding is a specific effect of pramipexole or an effect that also can be found with other dopamine agonists. In line with our results, Brusa et al. [14] showed that in 20 right-handed patients with PD, pramipexol produced significant impairment of short-term verbal memory, attentional-executive functions and verbal fluency. Levodopa on the other hand has been suggested to have a positive effect on planning ability [12] and verbal fluency [40]. Previous studies have mostly focused on the acute effect of dopaminergic medication on cognition. There are a few studies investigating a more prolonged effect of dopaminergic medication on cognitive function (6–24 months), none of which had patients that received pramipexole as monotherapy. In our study the decision on the type of dopaminergic medication used in individual patients was made by the treating physician based on personal experiences and the common practice to primarily use dopamine agonists in younger onset patients (before 65–70 years of age). Therefore, patients receiving pramipexole differed from the other patients in age and baseline performance. We admit the possibility that the negative effect of pramipexole on phonemic fluency may be spurious; following a Bonferroni adjustment the difference between the groups would not remain significant. Nonetheless, a big proportion of patients with pramipexole declined in verbal fluency. Thus, the uncertainty of the effect of pramipexole on phonemic fluency needs to be investigated further.

Some methodological limitations of this study should be recognized, such as the 24 % loss to follow-up. Patients not participating in the follow-up were older, which makes the sample less representative. This biased loss to follow-up is a problem for most longitudinal studies and is hard to resolve. Also, using neuropsychological tests repeatedly renders the risk of practice effects. In an attempt to minimize practice effects, episodic memory tests and verbal naming tests that did not have alternate forms were excluded from the 12-month follow-up [25]. Still, practice effects can be present even with alternate forms, especially for non-verbal memory. Another problem is the lack of control group at the 12-month follow-up. We wanted to investigate whether motor and cognitive variables that were associated in the baseline study also developed in parallel and if different types of medication affected cognition differently for patients with PD. To answer these questions a healthy control group is desirable but not mandatory. Finally, we made a lot of comparisons that influence the risk of type I errors. We did not adjust for alpha error inflation in this study to avoid exclusion of possible true relationships [32].

The strengths of the study are that it is based on a fairly large community-based study population followed longitudinally with repeated assessment of multiple cognitive domains. The use of established diagnostic criteria for PD by movement disorder specialists and long-time follow-up, together with abnormal SPECT FP-CIT in all cases, makes the diagnosis likely to be correct despite the lack of autopsy verification. Another strength is the prospective identification of incident cases with the vast majority of individuals being drug-naïve at baseline.

In summary, we found that during a period of 1 year from diagnosis, change in bradykinesia was associated with change in working memory and mental flexibility, and change in postural instability/gait disturbances were associated with change in visuospatial memory. These findings strengthen the results from our previous study on baseline data suggesting similar associations. The negative effect found for pramipexole on phonemic fluency should be further investigated.

Acknowledgments

This study was supported by grants from The Swedish Medical Research Council, The Parkinson Foundation in Sweden, The Swedish Association of Persons with Neurological Disabilities, Umeå University, Västerbotten County Council (ALF), King Gustaf V and Queen Victoria Freemason Foundation, the Swedish Brain Foundation and Lions Clubs Sweden’s Foundation for Research in Age-related Diseases.

Conflicts of interest

Magdalena Eriksson Domellöf reports no disclosures. Dr Eva Elgh reports no disclosures. Dr Lars Forsgren serves on scientific advisory boards for Pfizer and UCB; and receives research support from the Parkinson Foundation, King Gustaf V and Queen Victoria Freemason Foundation, the Kempe Foundation, Västerbotten County Council, Umeå University and the Swedish Medical Research Council.

Copyright information

© Springer-Verlag Berlin Heidelberg 2013