The long-term effect of cognitive training in patients with Parkinson’s disease, a blinded single group study

(1)

Bachelor Thesis

The long-term effect of cognitive training in patients with Parkinson’s disease, a

blinded single group study

Nina Samoei

12091413

Supervisor:

Tim van Balkom

The Department of Anatomy & Neurosciences at the VU medical center,

Amsterdam

(2)

Abstract

Parkinson’s disease (PD) is characterized by motor and cognitive symptoms. Cognitive complaints have a great impact on the lives of patients with PD. Pharmacological solutions have been proven to help with motor complaints, but solutions for cognitive complaints are relatively new. Cognitive training (CT) can help to relieve cognitive complaints of PD

patients on different cognitive domains. In this study, we investigated the effect of a cognitive training with a follow-up after six months and one year. 140 subjects in a mild to moderate disease stage underwent several neuropsychological assessments. We performed a latent class growth analysis (LCGA) with three tasks from different domains and found variation in cognitive change over time. On the basis of the LCGA, there were groups formed with different characteristics. Crucially, the relatively older, more severely affected patients show less progress in cognitive tasks compared to healthier and younger patients.

Keywords: Parkinson’s disease, cognitive training, follow-up, latent class growth analysis, cognitive improvement.

Introduction

PD is a common disease which is caused by the degeneration of dopaminergic neurons (Pfeiffer et al., 2015). PD is particularly known as a motor disorder; therefore, the non-motor complaints are often overlooked. The most obvious motor symptoms of PD are resting tremor, rigidity, postural instability and bradykinesia (Bosboom et al., 2004). Cognitive dysfunction, depression and hyposmia are examples of non-motor symptoms of PD, which are often developed years before diagnosis (Schapira et al., 2017). These non-motor symptoms mainly contribute to a poorer quality of life.

Cognitive impairment is the loss of cognitive abilities in several domains. Attention, executive functions and memory are the domains which are most affected with cognitive dysfunction (Muslimovic et al., 2005) and these are often damaged earlier in the course of the disease (Emre et al, 2007). The study of Muslimovic (2005) showed that 24% of their subjects with PD already had cognitive dysfunction at diagnosis and only 4% of their healthy subjects were cognitive impaired. Mild cognitive impairment (MCI) is a diagnosis for decline in cognitive skills and is a precursor to dementia. The prevalence of MCI is 14-18% (Goldman and Litvan, 2011). Aarsland and colleagues found that 25.8% of 1000 PD patients without dementia, suffered from MCI. MCI often develops in PD dementia (PDD). The incidence of PDD varies from 31.4 to 122.5 cases per 1000 patients (Bosboom et al., 2004). The

(3)

Furthermore, cognitive impairment increases as the disease progresses. The cognitive

complaints can cause patients to struggle with daily activities (Klepac et al., 2008). Until this day, levodopa and dopamine agonists are proven to be the most effective drugs for motor symptoms. But there is still no pharmacological solution for decreasing cognitive decline. Cholinesterase inhibitors could help with PDD by reducing the breakdown of acetylcholine but the response in patients can be unpredictable (Connolly & Lang, 2014). Rasagiline, a monoamine oxidase type-B inhibitor, has improved a few cognitive functions in PD patients (Connolly & Lang, 2014), but there has not been a longitudinal study with a large sample yet. Altogether, the effects of pharmacological studies for cognitive complaints are relatively small (Connolly & Lang, 2014).

Cognitive training (CT) is a non-pharmacological method to relieve the cognitive impairment of PD patients on different cognitive domains. There is not one general CT program, but they all practice cognitive skills. Strategy-based approach, computerized or pencil-and-paper exercises are different forms of CT, and these can also be combined (Walton et al., 2017). Using computer programs in CT is the most common method at the moment. This method is useful because computer programs can be adapted to the level of the test subject. There is sufficient evidence that CT can help with cognitive impairment. For example, Walton and colleagues (2017) demonstrated that patients with MCI and at risk of dementia showed medium effect size development in several cognitive domains. The meta-analysis of Leung and colleagues (2015) showed that the overall effect of CT was significant in seven studies with 272 patients. This applies to patients with mild to moderate PD (Leung et al., 2015). Regardless of proof that CT is efficient, it is still unclear whether the training can actually prevent dementia.

There has been previous research about CT in PD, but only a few studies have reported long term follow-up data. For example, the study of Petrelli and colleagues (2015) showed that CT may be effective to reduce cognitive decline in PD. They compared a structured CT program with an unstructured CT program in forty-seven non-demented PD patients and a control group. The CT was given two times a week, six weeks long and they concluded that CT may prevent cognitive decline (Petrelli et al., 2015). And CT can also play a role in preventing the beginning of MCI in PD patients (Petrelli et al., 2015). Díez‐ Cirarda and colleagues (2018) have demonstrated cognitive rehabilitation in fifteen PD patients. They followed a program which lasted three months with a follow-up period of 18 months. This study focused mostly on neuroimaging data and concluded that cognitive rehabilitation program might help with preserving cognitive functions in PD (Díez-Cirarda et al., 2018).

(4)

As mentioned earlier, PD is characterized by motor symptoms. Disease progression affects both cognitive symptoms and motor symptoms, and these may be interrelated. The clinical diagnosis is based on a set of motor symptoms: bradykinesia, rigidity, and rest tremor (Xia and Mao, 2012). However, not every patient experiences the whole set of symptoms. There is a heterogeneity among patients, and this is also seen in the progression of the disease. Xia and Mao (2012) stated that progression of motor symptoms shows a nonlinear pattern and that the decline is more rapid for patients at the early state of PD than patients with a longer disease duration. There is also a difference in the relationship between motor and non-motor symptoms in patients at the early state or an advanced state of the disease. Politis and colleagues (2010) investigated this phenomenon in 265 PD patients and found that the early group (<6 years) suffered more from the cardinal motor symptoms. The late group (≥6 years) was more affected by the motor aspects of PD. Xia and Mao (2012) concluded that non-motor symptoms can precede the non-motor symptoms, years before the PD diagnosis. Which indicates that the disease process starts years before diagnosis. Cognitive problems are under reported and patients are often not aware of these complaints (Politis et al., 2010). Ba and colleagues (2016) showed in their cross-sectional chart review, that PD patients in a more advanced disease stage scored higher on a non-motor symptoms questionnaire, including attention and memory concerns. Along with a correlation between postural instability gait difficulty and the burden that patients experience by cognitive complaints (Ba et al., 2016). In this study, we investigated the effect of a cognitive training after six months and one year on patients with PD. We looked at different subgroups of cognitive changes over time and the characteristics of these groups. Our research question was whether the CT effect is maintained for a year after the intervention along with the association between decline in motor functions and cognitive functions, looking at the different cognitive classifications. We hypothesized that the cognitive functions maintain or improve and that motor functions decrease as the disease progresses. An early intervention with CT may be able to reduce cognitive decline. Eventually, this can help patients with their daily difficulties, and it can improve the overall wellbeing of PD patients. This research is innovative because we used a sizable sample of 140 PD patients.

This is investigated by administering neuropsychological examination with various

questionnaires and cognitive tests. We predict stable or higher scores on the cognitive tasks, in this case the Stroop Color-Word test card I (word-reading), the Letter fluency task and the Rey Auditory Verbal Learning Test. We chose these tests because they measure three

(5)

most damaged by PD and they are trained with the CT. We expect lower scores on the scale of motor functions, the UPDRS-III, over time.

(6)

Materials and Methods Trial design

In this blinded longitudinal single group study, 140 subjects were random and evenly divided in a CT group and an active control group (AC). Thus, seventy patients in the CT group and seventy in the AC. The groups were investigated simultaneously because the researchers were blinded. Neuropsychological examination was executed at four different moments (T0 – baseline, T1 – after training, T2 – after six months, T3 – after one year, T4 – after two years). The location of this study was the Amsterdam University Medical Centers (Amsterdam UMC), location VUmc.

Participants

The participants were 140 Dutch-speaking PD patients with subjective cognitive complaints. They have shown their interest in participation through academic hospitals, a union for PD patients, advertisements and individuals registered in a database where they have shown their interest to participate in PD related research. Before the research started, participants had to sign an informed consent and they had to have access to a computer with internet.

Additionally, the patients should have enough vitality to endure extensive neuropsychological assessments. The inclusion criteria for subjects were that they suffered from significant subjective cognitive complaints (Parkinson’s Disease Cognitive Rating Scale, score > 3) (Kulisevsky et al., 2013). Also, the patients were in a mild to moderate disease stage (Hoehn & Yahr disease stage, score < 4) (Hoehn & Yahr, 1967). An exclusion criterion was that patients should not suffer from traumatic brain injury. Presence of one or more impulse control disorders (ICD) was defined by a positive screening with the ICD criteria interview. Psychotic symptoms were also defined by a positive screening but with the Schedule for Assessment of Positive Symptoms – PD (Andreasen, 1984). Benign hallucinations did not count as an exclusion criterium. Current drug- or alcohol abuse was determined by the CAGE AID-interview, score > 1 (Brown & Rounds, 1995) (Ewing, 1984). Another exclusion

criterium was moderate to severe depressive symptoms (Beck depression inventory, score > 18) (Beck et al., 1961). And finally, the implication for dementia syndrome

(Self-administered Gerocognitive Examination, score < 14 (Scharre et al. 2010) and the Montreal Cognitive Assessment (MOCA), score < 22 (Nasreddine et al., 2005).

(7)

Procedure

First of all, the subjects underwent pre-screening that includes self-administered cognitive screening, questionnaires and a phone interview. When subjects had passed pre-screening, they were asked to visit for a screening at location. Once again, they underwent interviews, questionnaires and this time also extensive neuropsychological assessment. The participants also had to sign an informed consent. Their level of cognitive dysfunction was tested by the MOCA (Nasreddine et al., 2005). Motor dysfunction was tested by the Unified Parkinson’s Disease – Rating Scale part III (UPDRS-III) (Fahn et al., 1987). The cognitive training lasted for eight weeks, three times a week, 45 minutes long. The training consists of thirteen games (See Appendix A for description of the games), the active control consists of three games that engage participants cognitively without a training component. The subjects could plan for themselves when they performed this training during the week. During the post-intervention and follow-up moments, the participants went through several neuropsychological tests and questionnaires (Figure 1).

Figure 1. Timeline of the various measurement points.

Pre-screening - Informed consent - Questionnaires (at home) - Phone interview Screening + randomization - Motor and neuropsychological assessment - Interviews - Questionnaires Pre-intervention - Neuropsychological assessment - Questionnaires Follow-up after 6 months (T2) - Neuropsychological assessment - Questionnaires Follow-up after 1 year (T3) - Motor and neuropsychologcail assessment - Questionnaires

(8)

Outcomes

Different tests were used for the neuropsychological examination, based on Litvan and colleagues (2012). For each cognitive domain, there were two neuropsychological outcomes (See Appendix B for a detailed description of the outcomes). The motor symptoms were tested by the UPDRS-III (Fahn et al., 1987). The neuropsychological outcomes were

essentially used to classify PD-NC, PD-MCI and PD-D (Table 1). The MOCA (Nasreddine et al., 2005), Tower of London (Shallice, 1982) and the Pentagon Copy from the Mini-Mental State Examination (Folstein et al., 1975) are tests that were used as well in the

neuropsychological assessment (See Appendix B for a detailed description). A t-score was used to demonstrate to what extent a patient deviated from the mean in a healthy population. Normal cognition was classified by a t-score higher than 40. To classify MCI, there must be aggravation in either two neuropsychological tests in one cognitive domain or one impaired test in two different cognitive domains (Reuter et al., 2012). This has been established by a t-score between 30 and 40. PDD was diagnosed when a patient deviated from the mean by two times the standard deviation in neuropsychological tests (Figure 2). In other words, the participant had a lower t-score than 30.

For this research, the Stroop Color-Word test card I (word-reading) (SCWT-I), the Letter fluency task and the Rey Auditory Verbal Learning Test (RAVLT) were used to measure change in cognitive function. The independent variables were time (baseline - T2 - T3) and the classification of PD-NC vs. PD-MCI vs. PD-D (Figure 2). The dependent variables were the SCWT-I, the Letter fluency task, the RAVLT and the motor symptoms measured by the UPDRS-III. This way, we looked at the relationship between motor symptoms and cognitive decline.

(9)

Table 1. The outcomes for classifying PD-NC, PD-MCI and PD-D along with the corresponding cognitive

domain.

Domain Outcome

Attention/processing speed Stroop card I (word-reading) Digit span - forward

Executive function Letter fluency – total score

Stroop card III (color-word) corrected for card II speed (color-reading)

Episodic memory Rey Auditory Verbal Learning Test - recall

Location Learning Test – recall

Visuospatial/visuoconstructive function Rey Complex Figure Test – total score Benton Visual Form Discrimination Test – total score

Language Boston Naming Test – total score

Category fluency – total score

Figure 2. The classification of PD-NC, PD-MCI and PD-D displayed in a normal distribution. On the x-axis, the

t-score is presented with the mean and the standard deviations.

Statistical method

For this research, we used a form of Growth Mixture Modeling (GMM), the Latent Class Growth analysis (LCGA), to determine heterogeneity of cognitive decline over time with a random intercept. This means that the individuals are measured multiple times to create different developmental trajectories (Cole et al., 2012). The whole sample was divided in multiple latent subgroups that were formed by indicator variables (Cole et al., 2012). The variables were measured from different time points. This method can be used for longitudinal data and the outcome will presumably be a linear function of time. But the outcome can also be a quadratic function. In our study, the main effect was the course of cognitive change over

PD-MCI

PD-D

 - 2  -    + σ  + 2

(10)

time in the dependent variables, the SCWT-I, the Letter fluency task, the RAVLT and the UPDRS-III. Time was the fixed effect. The interaction effect was the comparison between time and cognitive classification. And the comparison between cognitive functions and motor functions was a separate effect, analyzed by a repeated measures correlation. We performed the LCGA by first looking at the model without the latent classes. Thereafter, we added multiple latent classes until we found the best fitting best fitting model. The model searches for different subgroups of change over time. The number of latent classes was chosen by looking at the Bayesian information criterion (BIC) and Akaike information criterion (AIC) values, and also the clinical relevance. Then we examined whether there were clinical or demographic differences between these groups. The data analysis was performed in RStudio, using the “lcmm” package.

(11)

Results

Participants characteristics

We included 82 males and 51 females in the data analysis (Table 2). The mean age of the participants was 63 years (SD=7.47). Participants who only had one assessment at baseline, were excluded (N=7). At six-months follow-up there were 10 missing assessments and at one-year follow-up 31 missing assessments.

Table 2. The participants characteristics at baseline. Participants who only had one assessment at baseline, were

already excluded. Variables Value Age (years) Mean (SD) 63 (7.47) Sex (N) Male (%) Female (%) 82 (61.65%) 51 (38.35%)

Disease duration (years)

Mean (SD) 6.5 (5.1)

Verhage Educational Level

Median (Range) 6 (3-7)

Education (years)

Mean (SD) 16 (3.93)

MOCA score

Mean (SD) 26.14 (2.13)

UPDRS total score On (N)

Mean (SD)

124

20.68 (8.77)

UPDRS total score

Off (N) Mean (SD)

9

24.67 (10.25)

UPDRS Hoehn and Yahr Scale

Median (Range) 2 (1-3)

BDI total score

Mean (SD) 8.02 (4.13)

PAS total score

Mean (SD) 10.12 (6.73) PD-CFRS total score Mean (SD) 8.89 (4.46) Cognitive classification (N) Normal Cognition SD-MCI MD-MCI PDD 28 16 66 23

Abbreviations: MOCA – Montreal Cognitive Assessment; UPDRS – Unified Parkinson's disease rating scale;

BDI – Beck Depression Inventory; PAS – Parkinson Anxiety Scale; PD-CFRS – Parkinson's Disease - Cognitive Functional Rating Scale.

Latent class growth analyses

To analyze latent groups of cognitive change across PD patients we performed latent class growth analyses on the three cognitive tasks, the SCWT-I time, the total correct words of the

(12)

Letter Fluency and the total correct words of the RAVLT.For the SCWT-I, the three latent class model was the statistically most relevant model with the lowest BIC and AIC value (BIC=2654.84, AIC=2616.6). Nevertheless, the number of participants in the latent classes were highly unevenly distributed. Class 1 contained 5.7% of the participants, class 2

contained 91.4% of the participants and class 3 contained 2.9% of the participants. As showed in Figure 3, class 2 consisted of the large majority of the participants. For each latent class, there was a significant effect of time relative to baseline (Table 3). There was a decline in the SCWT-I time for each latent class and therefore improvement on the attention domain. Class 1 and class 3 seemed to have been formed on the basis of a few outliers. That is why we separated class 2 by performing another LCGA on this group only. However, this did not provide relevant information (See Appendix C).

Figure 3. The LCGA results of the SCWT-I, separated in a three-class model. On the x-axis, time is displayed.

(13)

Table 3. Fixed effects in the longitudinal model of the SCWT-I.

Coefficient Standard error p-value

Time effect at T2 Class 1 (N=5) -24.25 2.45 < .0001 Class 2 (N=115) -1.59 0.57 0.0052 Class 3 (N=13) -13.25 3.79 0.0005 Time effect at T3 Class 1 -29.50 2.71 < .0001 Class 2 -2.72 0.61 < .0001 Class 3 -14.27 3.83 0.0002

The LCGA of Letter fluency performance showed that the 2-class model had the best model fit (BIC=2735.639, AIC=2709.16). Groups were evenly distributed across latent classes. Class 1 contained 37.1% of the participants and class 2 contained 62.9% of the participants.

There was a significant effect of time, except for class 1 at time 2 (Table 7).So, there was an improvement seen in the total correct words compared to baseline. Class 2 had a coefficient of 8.04 after six months and a coefficient of 2.66 after one year. The performance of this class improved and then decreased. For class 2 (Mdn=27), the MOCA total score was significantly higher than class 1 (Mdn=25.5), W=1400, p=.0025. And, class 1 (Mdn=24) scored

significantly higher than class 2 on the UPDRS score (Mdn=19), W=2710, p=.0017. This was also seen at the UPDRS Hoehn & Yahr scale, where class 1 (Mdn=2.5) had a higher score than class 2 (Mdn=2), W=2492, p=.0236). The ratio in cognitive classification differed as well between the two classes, χ2=13.79, p=.0032. Therefore, we performed a post hoc comparison with Bonferroni adjustment. There were significantly more participants with PDD in class 1 than in class 2, p=.0019.

(14)

Figure 4. The LCGA results of the Letter fluency, separated in a two-class model.

Table 4. Fixed effects in the longitudinal model of the Letter fluency.

Time effect at T2 Class 1 1.25 1.74 0.4726 Class 2 8.04 1.83 < 0.0001 Time effect at T3 Class 1 4.36 1.58 0.0059 Class 2 2.66 1.23 0.0311

(15)

Table 5. The characteristics of the 2-class model of the Letter fluency.

Class 1 Class 2 p-value

N 49 84 Age (years) 65.60 (5.90) 62.48 (7.33) .3188 Sex Male Female 35 14 47 37 .0685 Disease duration (years) 7.04 (4.69) 6.27 (5.34) .1071 Verhage Educational Level 6 (5-7) 6 (3-7) .3868 Education (years) 15.33 (3.62) 16.58 (4.05) .0676

MOCA total score 25.39 (2.23) 26.57 (1.97) .0025

UPDRS total score 23.94 (9.31) 19.20 (8.20) .0017

UPDRS Hoehn and Yahr Scale

2.5 (1-3) 2 (1-3) .0236

BDI total score 8.16 (3.93) 7.94 (4.52) .5188

PAS total score 11.41 (7.83) 9.36 (5.90) .3040

PD-CFRS total score 9.35 (4.59) 8.63 (4.38) .4608 Cognitive classification Normal cognition SD-MCI MD-MCI PDD 7 5 21 16 21 11 45 7 .0032

Abbreviations: MOCA – Montreal Cognitive Assessment; UPDRS – Unified Parkinson's disease

rating scale; BDI – Beck Depression Inventory; PAS – Parkinson Anxiety Scale; PD-CFRS – Parkinson's Disease - Cognitive Functional Rating Scale

For the RAVLT, the three-class model showed the best fit considering the even distribution. Even though the BIC and AIC values were larger compared to the two-class model

(BIC=2699.87, AIC=2661.63), but the two-class model was more disproportionate. In the three-class model, 30% of the participants were in class 1, 37.9% of the participants were in class 2 and 32.1% of the participants were in class 3. As shown in figure 6, there was a difference in score between the three latent classes. Time was a significant predictor (Table 9), except for class 3 at time point 2 and 3. Overall, there was an improvement seen in the RAVLT performance in comparison to baseline, especially for latent class 1. But the

performance of class 3 decreased after six months. There was a significant difference in age between the classes, F(2, 130)=13.64, p<.0001. The Tukey multiple comparisons of means displayed that the participants in class 1 were older than the participants in class 2 (p<.0001). Additionally, the participants in class 3 were older than the participants in class 2 (p=.0002). There was a significant difference in sex, χ2=27.26, p<.0001. In class 2, there were

(16)

were significantly more males compared to class 1 and 2 (p=.0018). A significant difference was found on the total score of the MOCA, F(2, 130)=22.55, p<.0001. Class 2 had a higher score compared to class 1 (p=.0003), and class 3 (p<.0001). Regarding the cognitive

classifications, the ratio differed between the three latent classes, χ2=38.18, p<.0001. There were significantly more participants with normal cognition in class 2 (p=.0001). And class 3 obtained significantly more participants with PDD (p<.0001).

Figure 5. The LCGA results of the RAVLT, separated in a three-class model.

Table 6. Fixed effects in the longitudinal model of the RAVLT.

Time effect at T2 Class 1 10.72 1.98 < 0.0001 Class 2 3.26 1.38 0.0185 Class 3 -2.08 2.03 0.3057 Time effect at T3 Class 1 13.54 2.27 < 0.0001 Class 2 5.01 1.44 0.0005 Class 3 3.20 1.76 0.0692

(17)

Table 7. The characteristics of the 3-class model of the RAVLT.

Class 1 Class 2 Class 3 p-value

N 39 52 42 Age (years) 65.77 (6.54) 59.12 (6.43) 64.67 (7.70) <.0001 Sex Male Female 29 10 19 33 34 8 <.0001 Disease duration (years) 6.36 (5.11) 6.56 (5.21) 6.74 (5.10) .6986 Verhage Educational Level 6 (4-7) 6 (3-7) 6 (3-7) .5069 Education (years) 16.06 (3.95) 17.05 (4.29) 15.02 (3.16) .2820

MOCA total score 25.72 (1.92) 27.40 (1.76) 24.95 (1.94) <.0001

UPDRS total score 21.49 (10.20) 19.67 (8.15) 22.02 (8.48) .2924

UPDRS Hoehn and Yahr Scale

2 (1-3) 2 (1-3) 2 (1-3) .2840

BDI total score 7.44 (3.52) 8.92 (4.35) 7.45 (4.26) .4792

PAS total score 10.79 (6.59) 10.08 (6.53) 9.57 (7.19) .8243

PD-CFRS total score 9.20 (4.84) 8.65 (3.97) 8.92 (4.74) .9996 Cognitive classification Normal cognition SD-MCI MD-MCI PDD 7 5 23 4 20 7 22 3 1 4 21 16 <.0001

Furthermore, we compared the LCGA test results per neuropsychological task with each other to see how the participants were divided in the several latent classes per task (Table 8). There was a connection in the latent class solutions between the tasks (p=.0292). The post hoc comparison with Bonferroni adjustment did not show a significant result. But when looking at the distribution, there are more participants from class 3 of the RAVLT that were also in class 1 of the Letter fluency. This group made less progression than the other groups. And there were more participants class 1 and class 2 of the RAVLT, that were also in the Letter fluency class 2. These were the relatively healthier groups that made more progress.

(18)

Table 8. The distribution of the Letter fluency and the RAVLT. The latent classification of the fluency test is

shown in the columns and the classification of the RAVLT is displayed in the rows.

RAVLT

Class 1 Class 2 Class 3

Letter fluency

Class 1 14 13 22

Class 2 25 39 20

Finally, we performed a repeated measures correlation with the change on the UPDRS–III score along with the three tests and the corresponding latent classes. There were no significant associations between the change in motor symptoms and the cognitive tests (Table 9). PD medication could affect motor skills. We compared the levodopa equivalent daily dose

(LEDD) score from baseline with the LEDD score after one year, but there was no significant difference (p=.1977).

Table 9. The outcome of the repeated measures correlation of the UPDRS-III with change on

neuropsychological test performance.

r p-value 95% confidence interval SCWTC-I Class 1 Class 2 Class 3 -0.0538 -0.1420 -0.0500 -0.2093 0.5839 0.5201 0.6381 0.5138 -0.2439, 0.1402 -0.4672, 0.2091 -0.2556, 0.1600 -0.5412, 0.4841 Letter fluency Class 1 Class 2 -0.1027 -0.2294 -0.0086 0.2971 0.1784 0.9437 -0.2905, 0.0927 -0.5267, 0.1180 -0.2465, 0.2303 RAVLT Class 1 Class 2 Class 3 -0.0781 -0.1406 -0.1188 -0.1051 0.4281 0.4753 0.4264 0.5670 -0.2676, 0.1171 -0.5007, 0.2609 -0.3984, 0.1810 -0.2652, 0.4484

(19)

Discussion

The aim of this study was to test whether there were different subgroups of cognitive changes over time and what the characteristics were of these groups. Our research question was

whether the CT effect is maintained for a year after the intervention along with the association between decline in motor functions and cognitive functions, looking at the different cognitive classifications. We hypothesized that the cognitive functions maintain or improve and that motor functions decrease as the disease progresses. In other words, participants who deteriorate in motor symptoms, may show other cognitive improvement. We tested this by performing a LCGA with three cognitive tasks from three different cognitive domains. The main outcome of this study was that there was an effect of time and there was improvement seen on all three cognitive tasks after six months and one year. This refers to improvement on the attention domain, executive functions and memory. For the SCWT-I, most of the

participants were in the same performance range of score and showed a homogeneous curve in improvement. Therefore, the distribution was based on outliers. The result on the Letter fluency and the RAVLT showed that there was global progress after six months and one year. This was in line with the study of Petrelli and colleagues (2015), which demonstrated that CT may be effective to reduce cognitive decline in PD. The results also corresponded to the conclusion of Díez-Cirarda and colleagues (2018), that cognitive rehabilitation program might help with preserving cognitive functions in PD. The underlying cause for progress may have been that the participants trained the several domains with CT. But it may also be that the test-retest effect has caused this progression. Since only different versions of the cognitive tests were taken at each time point, and not a different test.

We examined whether there were clinical or demographic differences between the latent classes. And the result was that the relatively young and healthy test subjects not only scored higher on the cognitive tasks, but also showed more progression over time. Thus, the older, more severely affected patients showed less progress compared to healthier and younger patients. The study of Park and Bischof (2013) showed evidence that older adults have less neuroplasticity than younger participants. So, younger participants were most likely to have an increase in neural capacity caused by CT (Park & Bischof, 2013) which leads to more improvement on the tasks. Another difference is that for relatively healthier participants, CT can improve memory and decrease cognitive decline (Naismith et al., 2013). While CT in more affected patients ensures the maintenance of cognitive skills. Even though there was no significant difference found in education between the groups, it has been proven that

(20)

reason for this is the ceiling effect (Kwok et al., 2013). But since there was little variation in education level among the participants in this study, this phenomenon was not applicable. And CT can be adapted to the level of the test subject. There were also more males than females in classes that deteriorated more. This may be due to the fact that males are more affected by PD (Mayeux et al., 1992). And the study of Rahe and colleagues (2015) with MCI patients have shown that females demonstrated more improvement in CT compared to males. The repeated measures correlation with the UPDRS–III score along with the three tests

showed no significant associations between the change in motor symptoms and the change on cognitive tests performance. This indicated that there is no relationship between motor

symptoms and cognitive progression. There was also no difference in the mean score of the UPDRS–III at baseline and the follow-up after one year. Hence, no link has been found in the repeated measures correlation in this case. Which could have been caused by the fact that a few participants displayed progress, and a few declined, whereby the average remained the same. Also, we performed a cognitive and not a motor intervention. The cognitive

improvement was thus not expected on a motor level, and therefore no association was seen. Another influencing factor could have been the PD medication, which could have a positive effect on motor skills. This refers to the fact that there is a treatment for motor symptoms but not for cognitive decline. But there was no difference in PD medication between baseline and after one year so no effect on motor symptoms would be expected. The follow-up time could have been too brief to see variation in the motor symptoms. Provided that this study

continued, we might eventually see an association. As the cognitive training effect would then no longer be present and the motor skills will decline more rapidly. This association can probably be demonstrated when the follow-up after two years is included in the data analysis. Furthermore, there was a connection in the latent class solutions between the tasks. This was seen between the Letter fluency and the RAVLT. The relatively healthier and younger

participants that made more progress were in the same class in both the Letter fluency and the RAVLT. This indicates the fact that these participants made global progress on multiple domains. The correlation was not found for the SCWT-I in comparison to the other tasks, because of the uneven distribution.

There were a few limitations in this study. The most important one was not including the data after two years. This would have given us more information about both the motor and

cognitive skills. For example, after one year we saw less big improvement on the Letter fluency, this could have been seen because of the intervention effect that seems to diminish after a year. It would be interesting to see what the effect of CT was on the Letter fluency

(21)

after two years. We have to take into account that the enjoyment of the tasks and training could also play a role in performance. Besides, the tests were taken at the current state of the participant at that time which can be influenced by environmental factors.

In conclusion, we found an effect of time in three cognitive tasks and we saw different subgroups of cognitive changes over time with significant differences in their characteristics. Cognitive function did improve, the relatively young and healthy test subjects scored higher on the cognitive tasks and showed more progression over time. We also hypothesized that people who deteriorate in motor symptoms may show other cognitive improvement. But there was no difference found in motor functions over time. What we did found was that people with greater disease progression benefit less from cognitive training. For follow-up research, it is important that the effect of CT after two years is also taken into account to see the result of the intervention effect after a longer time. This way, CT can be more involved in relieving the cognitive complaints and improving the overall wellbeing of PD patients.

(22)

Appendix A: Cognitive training

Table 10. The several games included in the cognitive training with their duration and which

cognitive domain the game targets.

Game Duration Cognitive domain

Drum rhythm 3 mistakes Working memory and

attention

Flanker task 80 seconds Cognitive flexibility

Correct sequence order 180 seconds Visuospatial function and

attention N-back task with bottles of

different shapes and colors

180 seconds Working memory

Totem pole task 2 mistakes Visuospatial function and

mental rotation

Target following 4 mistakes Focused and divided

attention Accept or decline stimuli by

switching rules and speed

90 seconds Cognitive flexibility and processing speed

Memory task 120 seconds Working memory and

attention

Click on stimuli at the right time 180 seconds Divided attention,

psychomotor and processing speed

Search task with distractors 300 seconds Visuospatial function, processing speed Stack blocks on top of each

other

180 seconds Planning

Remembering characteristics of a penguin

180 seconds Working memory and

processing speed

Puzzle task 240 seconds Visuospatial function and

processing speed Adapted with permission from van Balkom.

(23)

Appendix B: Neuropsychological assessment (T0-T2-T3)

Montreal Cognitive Assessment (MOCA) ((Nasreddine et al., 2005)

Brief test with several (drawing) questions which measures cognitive function in attention, memory, language, abstraction, orientation and visuospatial/executive functions.

Self-administered Gerocognitive Examination (Scharre et al., 2010)

Brief self-administered memory test to screen MCI and dementia. This test is aimed at cognitive, memory or thinking impairments.

Tower of London (Shallice, 1982)

Computer task where the participants have to move objects from a starting position to a goal position to task planning skills.

Stroop Color Word Test (Hammes, 1971)

There are three forms of the Stroop Color Word Test: word reading, naming the color and naming the color of the word. The first form is aimed at attention and processing speed. The other forms are used to test executive function.

Visual Form Discrimination Test (Benton et al., 1994)

A figure is shown on the top page and on the bottom page there are four similar figures but only one is the same as the figure on the top page. To test if the participant can still match the identical object with each other. Visuospatial function is targeted with this test.

Rey Auditory Verbal Learning Test (Saan & Deelman, 1986)

A memory test containing fifteen words which are repeated five times. Repetition and

intrusions are noted and there will be a recall after approximately fifteen minutes. Thereafter, a list of words that were and were not present in the RAVLT will be read. The participant may indicate whether he or she recognizes the word or not.

Rey Complex Figure Test (Meyers & Meyers, 1995)

The participant has to draw a complex figure. This test is used to determine if the drawing ability is still intact but also to test planning and organization.

(24)

Category fluency (Luteijn & Barelds, 2005)

Participants have to name as many words within one category in one minute. This way, we can demonstrate the language domain.

Letter fluency (Schmand et al., 2008)

Participants have to name as many words with the same initial in one minute. We ask the participants to not include names of people or places. Also, they may not say the same word with a different ending multiple times. This way, we can test the participants’ executive functions.

Location Learning Test (Kessels et al., 2004)

The Location Learning Test is a memory task where the participant has to remember different objects in different places. This test is repeated several times and the score is based on the mistakes being made. So, a lower score is better on this task. The episodic spatial memory for object locations is targeted.

Boston Naming Test (Kaplan et al., 2001)

Participants get to see different object drawn and they have to name those. For example, a bed or a house. Thus, we can test their ability of visual confrontation naming.

Wechsler Adult Intelligence Scale-III digit span (Wechsler, 2000)

In this test, the participants must repeat the numbers that are read. First in a forward order and then backwards.

Pentagon Copy from the Mini-Mental State Examination (Folstein et al., 1975)

The patient has to draw the same pentagon as seen on paper to see if the drawing ability of patients is still intact.

(25)

Appendix C: SCWT-I

Table 11. The characteristics of the 3-class model of the SCWT-I.

Class 1 Class 2 Class 3 p-value

N 5 115 13 Age (years) 65.60 (5.90) 62.61 (7.43) 63.62 (8.55) .9100 Sex Male Female 4 1 70 45 8 5 .6901 Disease duration (years) 8.80 (8.41) 6.37 (4.96) 7.38 (5.17) .6120 Verhage Educational Level 6 (5-7) 6 (3-7) 6 (4-7) .2803 Education (years) 15 (4.12) 16.13 (3.97) 16.46 (3.73) .5430 MOCA total score 26.4 (2.30) 26.22 (2.08) 25.31 (2.53) .3156 UPDRS total score 24 (5.43) 20.40 (9.01) 24.62 (8.2) .1477

UPDRS Hoehn & Yahr

2.5 (2-3) 2 (1-3) 2.5 (1.5-3) .2287

BDI total score 8.60 (1.95)

8.10 (4.19) 7.15 (4.26) .6766

PAS total score 8.20 (3.49) 9.81 (6.50) 13.62 (8.73) .2962 PDCFRS total score 7.20 (2.17) 8.90 (4.42) 9.50 (5.44) .7935 Classification (N) Normal cognition SD-MCI MD-MCI PDD 0 2 2 1 27 13 59 16 1 1 5 6 .05563

We separated class 2 by performing another LCGA on this group only. The 3-class model with a BIC value of 2250.22 and an AIC value of 2213.15 was chosen. Class 1 contained 12.5% of the participants, 82.03% of the participants were in class 2 and the other 5.47% were in class 3. For each latent class, time was a significant predictor (Table 12). There was

progress seen for class 1 and class 3 and decline for class 2. But there were no significant differences found between the three latent classes at the covariates (Table 13) due to the fact that there was a highly uneven distribution again (Figure 6).

(26)

Figure 6. The LCGA results of the SCWT-I, the largest group of the first LGCA was separated in a three-class

model.

Table 12. Fixed effects in the longitudinal model.

Time effect at T2 Class 1 -1.52 0.56 .0065 Class 2 12.38 1.93 <.0001 Class 3 -7.70 1.73 <.0001 Time effect at T3 Class 1 -1.88 0.61 .0019 Class 2 5.85 1.86 .0016 Class 3 -10.48 1.87 <.0001

(27)

Table 13. The characteristics of the SCWT-I class 2 separated in three latent classes.

Class 1 Class 2 Class 3

N 15 98 2 Age (years) 63.60 (6.92) 62.49 (7.59) 61 (4.24) Sex Male Female 9 6 59 39 2 0

Disease duration (years) 5.20 (3.53) 6.58 (5.18) 4.50 (0.71)

Verhage Educational Level

5 (4-7) 6 (3-7) 6 (5-7)

Education (years) 15.7 (5.02) 16.22 (3.83) 15 (2.83)

MOCA total score 26.20 (2.46) 26.24 (2.04) 25 (1.41)

UPDRS total score 23.07 (10.08) 19.93 (8.90) 23.5 (0.71)

UPDRS Hoehn & Yahr 2.5 (1-3) 2 (1-3) 2.25 (2-2.5)

BDI total score 10.60 (5.23) 7.69 (3.94) 9 (1.41)

PAS total score 11 (6.25) 9.71 (6.59) 6 (2.83)

PD-CFRS total score 8.88 (4.78) 8.80 (4.37) 14 (2.83) Classification (N) Normal cognition SD-MCI MD-MCI PDD 2 2 8 3 25 11 50 12 0 0 1 1

(28)

Literature

Aarsland, D., Andersen, K., Larsen, J. P., & Lolk, A. (2003). Prevalence and characteristics of dementia in Parkinson disease: an 8-year prospective study. Archives of neurology, 60(3), 387-392.

Andreasen, N. C. (1984). Scale for the assessment of positive symptoms. Group, 17(2), 173-180.

Ba, F., Obaid, M., Wieler, M., Camicioli, R., & Martin, W. W. (2016). Parkinson disease: the relationship between non-motor symptoms and motor phenotype. Canadian Journal of

Neurological Sciences, 43(2), 261-267.

Beck, A. T., Ward, C. H., Mendelson, M., Mock, J., & Erbaugh, J. (1961). An inventory for measuring depression. Archives of general psychiatry, 4(6), 561-571.

Benton, A.L., Sivan, A.B., deS Hamsher, K., Varney, N.R. (1994). Contributions to neuropsychological assessment: A clinical manual: Oxford University Press, USA.

Bosboom, J. L. W., Stoffers, D., & Wolters, E. C. (2004). Cognitive dysfunction and dementia in Parkinson’s disease. Journal of neural transmission, 111(10-11), 1303-1315.

Broadbent, D.E., Cooper, P.F., FitzGerald, P., Parkes, K.R. (1982). The Cognitive Failures Questionnaire (CFQ) and its correlates. Br J Clin Psychol, 21(Pt 1),1-16.

Brown, R.L., Rounds, L.A. (1995). Conjoint screening questionnaires for alcohol and other drug abuse: criterion validity in a primary care practice. Wis Med J, 94(3),135-140.

Cole, V. T., Apud, J. A., Weinberger, D. R., & Dickinson, D. (2012). Using latent class growth analysis to form trajectories of premorbid adjustment in schizophrenia. Journal of abnormal psychology, 121(2), 388.

Connolly, B. S., & Lang, A. E. (2014). Pharmacological treatment of Parkinson disease: a review. Jama, 311(16), 1670-1683.

(29)

Devilly, G.J., Borkovec, T.D. (2000) Psychometric properties of the credibility/expectancy questionnaire. J Behav Ther Exp Psychiatry 31(2):73-86.

Díez‐ Cirarda, M., Ojeda, N., Peña, J., Cabrera‐ Zubizarreta, A., Lucas‐ Jiménez, O., Gómez‐ Esteban, J. C., ... & Ibarretxe‐ Bilbao, N. (2018). Long‐ term effects of cognitive rehabilitation on brain, functional outcome and cognition in Parkinson's disease. European

journal of neurology, 25(1), 5-12.

Emre, M. (2003). Dementia associated with Parkinson's disease. The Lancet Neurology, 2(4), 229-237.

Emre, M., Aarsland, D., Brown, R., Burn, D. J., Duyckaerts, C., Mizuno, Y., ... & Goldman, J. (2007). Clinical diagnostic criteria for dementia associated with Parkinson's

disease. Movement disorders: official journal of the Movement Disorder Society, 22(12), 1689-1707.

Ewing, J.A. (1984). Detecting alcoholism. The CAGE questionnaire. JAMA, 252(14), 1905- 1907.

Fahn, S., Elton, R.L., members of the UPDRS Development Committee (1987). The Unified Parkinson's Disease Rating Scale. In: Fahn S, Marsden CD, Calne DB, Goldstein M, editors. Recent developments in Parkinson's disease. Vol. 2. Florham Park, NJ: MacMillan Healthcare Information; p. 153–163, 293–304.

Folstein, M.F., Folstein, S.E., McHugh, P.R. (1975). "Mini-mental state". A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res, 12(3):189-198.

Goldman, J. G., & Litvan, I. (2011). Mild cognitive impairment in Parkinson's disease. Minerva medica, 102(6), 441–459.

Hammes, J.G.W. (1981). De Stroop Kleur-Woord Test. Handleiding. Amsterdam: Pearson Assessment and Information B.V.

(30)

Hindle, J. V., Petrelli, A., Clare, L., & Kalbe, E. (2013). Nonpharmacological enhancement of cognitive function in Parkinson's disease: a systematic review. Movement Disorders, 28(8), 1034-1049.

Hoehn, M. M., & Yahr, M. D. (1967). Parkinsonism: onset, progression. and mortality.

Neurology, 50(2), 318-318.

Cooper, J. A., Sagar, H. J., Jordan, N., Harvey, N. S., & Sullivan, E. V. (1991). Cognitive impairment in early, untreated Parkinson's disease and its relationship to motor

disability. Brain, 114(5), 2095-2122.

Kaplan, E., Goodglass, H., Weintraub, S. (2001). Boston naming test: Pro-ed.

Kessels, R.P., Nys, G.M., Brands, A.M., van Zandvoort, M.J. (2004). [The Location Learning Test as a measure of spatial memory: applicability of a modified administration procedure and normative data]. Tijdschr Gerontol Geriatr, 35(4),147-152.

Klepac, N., Trkulja, V., Relja, M., & Babić, T. (2008). Is quality of life in non‐ demented Parkinson’s disease patients related to cognitive performance? A clinic‐ based cross‐ sectional study. European Journal of Neurology, 15(2), 128-133.

Kulisevsky, J., Fernandez de Bobadilla, R., Pagonabarraga, J., Martinez-Horta, S.,

Campolongo, A., Garcia-Sanchez, C., Pascual-Sedano, B., Ribosa-Nogue, R., Villa-Bonomo, C. (2013). Measuring functional impact of cognitive impairment: validation of the Parkinson's disease cognitive functional rating scale. Parkinsonism Relat Disord, 19(9), 812-817.

Kwok, T., Wong, A., Chan, G., Shiu, Y. Y., Lam, K. C., Young, D., ... & Ho, F. (2013). Effectiveness of cognitive training for Chinese elderly in Hong Kong. Clinical Interventions

in Aging, 8, 213.

Leentjens, A.F., Dujardin, K., Pontone, G.M., Starkstein, S.E., Weintraub, D., Martinez-Martin, P. (2014). The Parkinson Anxiety Scale (PAS): development and validation of a new anxiety scale. Mov Disord, 29(8),1035-1043.

(31)

Leung, I. H., Walton, C. C., Hallock, H., Lewis, S. J., Valenzuela, M., & Lampit, A. (2015). Cognitive training in Parkinson disease: a systematic review and

meta-analysis. Neurology, 85(21), 1843-1851.

Litvan, I., Aarsland, D., Adler, C. H., Goldman, J. G., Kulisevsky, J., Mollenhauer, B., ... & Weintraub, D. (2011). MDS Task Force on mild cognitive impairment in Parkinson's disease: critical review of PD‐ MCI. Movement disorders, 26(10), 1814-1824.

Litvan, I., Goldman, J. G., Tröster, A. I., Schmand, B. A., Weintraub, D., Petersen, R. C., ... & Emre, M. (2012). Diagnostic criteria for mild cognitive impairment in Parkinson's disease: Movement Disorder Society Task Force guidelines. Movement disorders, 27(3), 349-356.

Luteijn F, Barelds D (2005). Herziening van de GIT. Handleiding bij de GIT-2. In.: Harcourt Publishers, Amsterdam.

Mayeux, R., Denaro, J., Hemenegildo, N., Marder, K., Tang, M. X., Cote, L. J., & Stern, Y. (1992). A population-based investigation of Parkinson's disease with and without dementia: relationship to age and gender. Archives of neurology, 49(5), 492-497.

McLean, G., Tobias, M. (2004). The New Zealand physical activity questionnaires: Report on the validation and use of the NZPAQ-LF and NZPAQ-SF self-report physical activity survey instruments. New Zealand: SPARC.

Meyers, J.E., Meyers, K.R. (1995). Rey Complex Figure Test and recognition trial professional manual: Psychological Assessment Resources.

Naismith, S. L., Mowszowski, L., Diamond, K., & Lewis, S. J. (2013). Improving memory in Parkinson's disease: a healthy brain ageing cognitive training program. Movement

Disorders, 28(8), 1097-1103.

Nasreddine, Z.S., Phillips, N.A., Bedirian, V., Charbonneau, S., Whitehead, V., Collin, I., Cummings, J.L., Chertkow, H. (2005). The Montreal Cognitive Assessment, MoCA: a brief screening tool for mild cognitive impairment. J Am Geriatr Soc, 53(4), 695-699.

(32)

Park, D. C., & Bischof, G. N. (2013). The aging mind: neuroplasticity in response to cognitive training. Dialogues in clinical neuroscience, 15(1), 109.

Petrelli, A., Kaesberg, S., Barbe, M. T., Timmermann, L., Rosen, J. B., Fink, G. R., ... & Kalbe, E. (2015). Cognitive training in Parkinson's disease reduces cognitive decline in the long term. European journal of neurology, 22(4), 640-647.

Politis, M., Wu, K., Molloy, S., G. Bain, P., Chaudhuri, K. R., & Piccini, P. (2010).

Parkinson's disease symptoms: the patient's perspective. Movement Disorders, 25(11), 1646-1651.

Rahe, J., Liesk, J., Rosen, J. B., Petrelli, A., Kaesberg, S., Onur, O. A., ... & Kalbe, E. (2015). Sex differences in cognitive training effects of patients with amnestic mild cognitive

impairment. Aging, Neuropsychology, and Cognition, 22(5), 620-638.

Reuter, I., Mehnert, S., Sammer, G., Oechsner, M., & Engelhardt, M. (2012). Efficacy of a multimodal cognitive rehabilitation including psychomotor and endurance training in Parkinson's disease. Journal of aging research.

Saan, R.J., Deelman, B.G. (1986). De 15-woordentest A en B (een voorlopige handleiding). Groningen: Afdeling Neuropsychologie, AZG.

Schapira, A. H., Chaudhuri, K. R., & Jenner, P. (2017). Non-motor features of Parkinson disease. Nature Reviews Neuroscience, 18(7), 435.

Scharre, D.W., Chang, S.I., Murden, R.A., Lamb, J., Beversdorf, D.Q., Kataki, M., Nagaraja, H.N., Bornstein, R.A. (2010). Self-administered Gerocognitive Examination (SAGE): a brief cognitive assessment Instrument for mild cognitive impairment (MCI) and early dementia. Alzheimer Dis Assoc Disord, 24(1), 64-71.

Shallice, T. (1982). Specific impairments of planning. Philos Trans R Soc Lond B Biol Sci, 298(1089), 199-209.

(33)

Schmand, B., Groenink, S., Van den Dungen, M. (2008). Letterfluency: psychometrische eigenschappen en Nederlandse normen. Tijdschr Gerontol Geriatr, 39(2), 64-74.

Starkstein, S.E., Mayberg, H.S., Preziosi, T.J., Andrezejewski, P., Leiguarda, R., Robinson, R.G. (1992). Reliability, validity, and clinical correlates of apathy in Parkinson's disease. J Neuropsychiatry Clin Neurosci, 4(2), 134-139.

van Balkom, T. D., Berendse, H. W., van der Werf, Y. D., Twisk, J. W., Zijlstra, I., Hagen, R. H., ... & van den Heuvel, O. A. (2019). COGTIPS: a double-blind randomized active

controlled trial protocol to study the effect of home-based, online cognitive training on cognition and brain networks in Parkinson’s disease. BMC neurology, 19(1), 179.

Walton, C. C., Naismith, S. L., Lampit, A., Mowszowski, L., & Lewis, S. J. (2017). Cognitive training in Parkinson’s disease: a theoretical perspective. Neurorehabilitation and neural

repair, 31(3), 207-216.

Wechsler, D. (2000). Wechsler adult intelligence scale - third edition. Nederlandstalige bewerking. Amsterdam: Pearson Assessment and Information B.V.

Weintraub, D., Mamikonyan, E., Papay, K., Shea Ja Xie, S.X., Siderowf, A. (2012).

Questionnaire for impulsive-compulsive disorders in Parkinson's Disease-Rating Scale. Mov Disord, 27, 242-247.

Xia, R., & Mao, Z. H. (2012). Progression of motor symptoms in Parkinson’s disease. Neuroscience bulletin, 28(1), 39-48.