1
Masterthesis (MT)
___________________________________________________________________________ Student
name : K.F. van der Zwaan
student number : 10002788
address : Regentesselaan 170C, 2562EG Den Haag
telephone : 06-22921905
e-mail : K.F.van_der_zwaan@lumc.nl
Supervisors
University of Amsterdam : Prof. Dr. B. Schmand
Onsite mentor : Prof. Dr. H.A.M. Middelkoop and drs. L. Schipper Research institution : Leiden University Medical Centre.
Date : 23-03-2017
Title: Evaluating the Neuropsychological Screening Process of Parkinson’s Disease Patients and Comparison to Structural Covariance Networks.
2 Goal of this research
Parkinson’s disease (PD) is a neurodegenerative disease accompanied by motor and non-motor symptoms. The non-motor symptoms are often treated by dopamine supplementation. The pharmacological effects of dopamine supplementation have a tendency to progressively fluctuate over time. When the disease progresses and the effects of the pharmacological intervention fluctuate, causing involuntary muscle movements in medication on-phase and rigidity in medication off-phase, deep brain stimulation (DBS) might be considered. To qualify for DBS, PD patients must meet certain conditions. Contraindications for DBS are e.g. postural imbalance, dementia or pre-existent (neuro)psychiatric problems. Cognitive and (neuro)psychiatric problems can be quantified by neuropsychological assessment.
The current study first sets out to describe patients from the PD population eligible for DBS in a comparison with those who were not. Especially looking at general cognitive performance, memory, executive functions and changes in mood. Because research shows that
neurodegeneration and cognitive decline may relate to neural network impairment, the second goal of this study was to correlate cognitive and emotional measures to subcortical volumes and a relatively novel analysis of MRI data: structural covariance networks.
3
1. Background
Parkinson’s disease (PD) is a neurodegenerative movement disorder. The disease has a prevalence of approximately 1.8 per 1000 males and 1.6 per 1000 women in Dutch general practice (Draijer, Eizenga, Sluiter, Opstelten, & Goudswaard, 2011). Due to an aging
population, the absolute number of patients suffering from PD is increasing. The degeneration results in motor symptoms and non-motor symptoms. Motor symptoms associated with PD are resting tremor, akinesia, hypokinesia, bradykinesia, rigidity and (postural) instability (Wolters, 2007). The non-motor symptoms are extensive as well and may include autonomic, sensory, (neuro)psychiatric symptoms and cognition (Jellinger, 2012). Cognitive impairment can be quantified through formal neuropsychological assessment, measuring cognitive abilities in five domains according to the clinical criteria of dementia, e.g. memory, executive functioning, praxis, gnosis, language (American Psychiatric Association, 2014). Cognitive domains in which deterioration is commonly seen in PD patients are memory and executive functions, e.g. attentional control (Aarsland, et al., 2010). Former research indicated that 26.7% of PD patients have mild cognitive impairment (MCI) at time of the PD diagnosis (Litvan, Aarsland, Adler, Goldman, Kulisevsky, Mollenhauer, et al., 2011).
PD is non-pharmacologically treated through physical, occupational, and speech therapy and pharmacologically treated by means of dopaminergic supplementation (Brichta, Greengard & Flajolet, 2013). When the disease progresses and the effects of the pharmacological
intervention fluctuate, causing involuntary muscle movements in medication on-phase and rigidity in medication off-phase, deep brain stimulation (DBS) of the subthalamic nucleus or globus pallidus pars interna might be considered. However, not all PD patients are eligible for this intervention, because some studies report serious adverse effects of DBS (Parsons, Rogers, Braaten, Woods & Tröster, 2006). Among the sequelae are cognitive impairment (e.g. memory deficits) and (neuro)psychiatric side effects (Okun, 2012; Rodriguez, Fernandez, Haq & Okun, 2007). As cognitive problems (e.g. memory and executive function deficits) and psychiatric pre-existent problems (e.g. depression) can be aggravated by the intervention, and
4 thus come to be contra-indications, patients are screened extensively to rule out any
neurological, neurophysiological, psychiatric or cognitive dysfunction. Bronstein et al. (2007) emphasized the importance of neuropsychological assessment of all PD patients screened for eligibility of DBS as it sets out to determine the cognitive, emotional and behavioral status of patients. To ascertain whether a PD patient is eligible for the neurosurgical DBS intervention all evidence collected during the neuropsychological assessment, physical examination & neurophysiological tests is discussed and weighted. The final decision of eligibility is a qualitative clinical judgment of functionality by (para)medical experts, based upon the evaluation and interpretation of all available clinimetrics such as observational data, anamnestic information, results of neuropsychological, physical and neurophysiological assessments, and MRI.
Alongside these assessments that determine the qualitative and functional neurological deterioration, research sets out to find biomarkers of early gray matter changes in structural quantitative methods like neuroimaging techniques. One of the methods that has been used in previous research in PD is voxel-based morphometry (VBM; Whitwell, Keith & Josephs, 2007). One property of this method is that VBM looks at specific cortical and subcortical regions separately (Coppen et al., 2016). However, prior studies suggests that
neurodegeneration might be the result of a change in neural networks instead of changes in separate regions (Seeley, Crawford, Zhou, Miller and Greicius, 2009). A technique that is suitable for answering these questions is structural covariance network (SCN) analysis. This technique quantifies the cortical and subcortical changes of grey matter and links them together. For example, we could interpret differences in volume as independent reactions of individual brains to different factors. However, if an individual has a relatively small volume of Broca’s area the same variation in volume is often observed in Wernicke’s area. This covariation of different cortical and subcortical structures is observed across the population. By clustering these regions trough component analysis networks can be formed (Alexander-Bloch 2013). This is in line with research that suggests neurodegeneration is the result of a changes in neural networks instead of changes in separate regions (Seeley, Crawford, Zhou,
5 Miller and Greicius, 2009; Rombouts et al., 2009). Certain clinical populations may show less coherent networks indicating changes in grey matter of the cortices and subcortical structures. Chou et al. (2015) compared the structural MRI data of PD patients to healthy controls. They found that in PD the pathological processes in the striatum affect the same networks that co-vary with the striatum in healthy controls. PD patients showed deviations compared to healthy controls in the following regions: caudate nucleus, parahippocampus, temporal cortices and cerebellum. Chou et al. (2015) recommend reinforcing their findings by further investigations using detailed neuropsychological examination.
2. Problem definition
This study sets out to determine the contribution of neuropsychological outcomes to the multidisciplinary consensus of patients’ (in)eligibility for DBS. To do so, we will first compare neuropsychological profiles of PD patients that eventually didn’t receive DBS to patients who did. Not looking at the whole neuropsychological qualitative assessment, but solely looking at the quantitative outcomes of individual tests and their contribution. We expect the group not receiving DBS to have a different performance on (neuro)psychological tests compared to patients who finally did receive DBS. This speaks for itself, since the clinical decision to undergo DBS is partially based upon the neuropsychological outcomes (incorporation bias; Worster & Carpenter, 2008). However, this hasn’t - to our knowledge - been a topic in previous research so far.
A secondary aim of the current study is to explore specific neuropsychological functions of SCN, as cognitive dysfunction in neurodegenerative disorders might not only be caused by focal damage but could be the result of impairment of brain networks (Hafkemeijer et al., 2016). Seeley et al. (2009) confirmed the network degeneration hypothesis in SCN in five patient groups, finding distinct connections between dysfunction of neural networks and neuropsychological measures in patients with Alzheimer’s disease, behavioral variant of frontotemporal dementia, semantic dementia, progressive nonfluent aphasia and corticobasal
6 syndrome. The current study will try to do the same for PD patients, correlating the
(neuro)psychological outcomes to SCNs.
2. Methods
2.1 Patients
This observational study uses data collected during the PROfiling PARKinson’s disease (PROPARK)-study conducted at the Leiden University Medical Centre and retrospective data collected during the DBS-screening. The PROPARK-study is a longitudinal study set out to profile Parkinson’s disease patients extensively on phenotype, disability and global outcomes of health, with assessment instruments that have been found to be valid and reliable in PD (Verbaan, Marinus, Visser, van Rooden, Stiggelbout, Middelkoop and van Hilten, 2007). All participants in the PROPARK-study are clinically characterized and underwent an MRI-scan of the brain.
Of the 173 participants in the PROPARK-study, 51 were screened for DBS eligibility. The current study initially included those 51 participants. The neuropsychological part of the DBS-screening is conducted by an experienced clinical neuropsychologist of the department of neurology in the Leiden University Medical Center. Patients are referred to the clinical neuropsychologist for this screening when the oscillations between medication off- and on-phase start to interfere with day to day life and adjusting the dosage or type of medication has little to no effects on these fluctuations.
2.2a Procedure
The neuropsychological and MRI data were collected at two separate moments. The ‘expiration date’ of MRI datum was set at half a year; if the neuropsychological data were collected more than half a year after or before the MRI data was collected participants were not included, 5 participants met this exclusion criterion. For the remaining 46 participants, the mean difference of days between MRI and neuropsychological assessment was 4.0 days (SD = 14.7, min. -8, max. 93). Furthermore, the T1-weigthed MRI-scan had to be of good enough
7 quality to perform voxel-based morphometry (VBM) in order to perform a structural
covariance network analysis; two participants were excluded based on this criterion. After being referred by the neurologist to the clinical neuropsychologist, patients were screened for dementia and mood problems. Patients are asked about cognitive and behavioral complaints, medical history and ensuing cognition and behavior is tested by means of a standardized neuropsychological assessment. All patients are tested by a neuropsychologist or neuropsychological intern supervised by a registered neuropsychologist. The
neuropsychological examination consists of the Cambridge Cognitive Examination – revised (CAMCOG-R), Wechsler Memory Scale (WMS), Rey Auditory Verbal Learning Test (RAVLT), Trail Making Test (TMT), Stroop Color Word Interference task (Stroop), Hospital Anxiety and Depression Scale (HADS) and Beck’s Depression Inventory (BDI).
Figure 1. Overview of the inclusion and exclusion criteria of the current study.
Participants in PROPARK study (n =173)
Participants excluded, because of missing neuropsychological screening for DBS
(n = 122) Participants screened for DBS and
MRI datum available (n = 51)
Participants excluded, because the neuropsychological datum was collected a half year before or after the MRI
(n = 5)
Participants screened and scanned within acceptable timeframe.
(n = 46)
Participants excluded, because the MRI datum could not be used for VBM
(n = 2)
Participants included in the current study.
8 2.2b Materials
Neuropsychological Assessment
In this section a brief overview is provided of the materials used in the current study. A more detailed description of psychometrics is provided in Appendix A. All participants had to follow a neuropsychological protocol, in which global cognitive abilities are measured by means of the Cambridge Cognitive Examination – revised (CAMCOG-R). This is a widely used cognitive battery to screen for dementia. Higher scores indicate a lower probability for dementia (Bouma, Mulder, Lindeboom, & Schmand, 2012). Memory was measured by means of the Wechsler memory scale (WMS) and the Dutch translation of the Rey Auditory Verbal Learning Test (RAVLT). The WMS is a global memory test with a special focus on verbal-, visual constructive- and working memory (Strauss, Sherman, & Spreen, 2006). The RAVLT is a verbal memory task with an immediate and delayed recall. In both the WMS and RAVLT higher scores indicate better memory performance. Executive functioning was measured with the Stroop Color-Word Interference task and the Trail Making Test A and B (TMT). The Stroop test consists of the Stroop Color Naming test (SCNT), the Stroop Word Reading test (SWRT) and Stroop Interference task (SIT). It measures processing speed and the ability to inhibit. The TMT is a measure of psychomotor processing speed. For both the TMT and the Stroop tasks seconds needed to complete the task measure executive control, with higher scores indicating less executive control (Bouma et al., 2012). Mood was measured by means of the Becks Depression Inventory II (BDI-II) and the Hospital Anxiety and Depression Scale (HADS), with higher scores indicating more mood problems.
MRI acquisition
Three-dimensional T1-weighted anatomical images were acquired on a 3 Tesla MRI scanner (Philips Achieva, Best, the Netherlands) using a standard 32-channel whole-head coil. Acquisition parameters were: TR = 9.8 ms, TE = 4.6 ms, flip angle = 8 °, FOV 220 x 174 x
9 156 mm, 130 slices with a slice thickness of 1.2 mm with no gap between slices, resulting in a voxel size of 1.15 mm x 1.15 mm x 1.20 mm.
MRI analyses and processing
All analyses were done using FMRIB’s software library (FSL; Woolrich et al., 2009; Smith et al., 2004; Jenkinson, Beckman, Behrens, Woolrich & Smith, 2012) after the scans were visually screened in order to rule out any major artifacts or significant brain injuries. To obtain only the brain tissue and disregard any other tissues (i.e. bone) the brain extraction tool (BET) was first used. The results were segmented for different brain tissues. Distinguishing between white matter, grey matter and cerebrospinal fluids. The segmentation does not define the voxels in a binary manner, but estimates a proportion of each tissue type. SIENAX (Smith, Zhang, Jenkinson, Chen, Matthews, Federico & De Stefano, 2002) was used to estimate the overall volumes of the grey matter, the white matter and cerebrospinal fluid. After this the grey matter was registered to the Montreal Neurological Institute 152 standard space image (MNI 152) (Jenkinson, Bannister, Brady & Smith, 2002), mapping the
(sub)cortical structures. This to estimate the volumes of the subcortical structures (Patenaude, Smith, Kennedy & Jenkinson, 2011).
After being registered to the MNI 152, the grey matter images were averaged to create a study specific grey matter template and re-registered to this study specific image. The resulting data is used to determine the network analysis.
The SCN were determined with an independent component analysis (ICA). A statistical analysis used to reduce several variables by clustering them into groups that have the largest statistical independence. This study used the networks defined by Hafkemeijer et al. (2014) as regions of interest (see figure 1). Every network consists of clusters of anatomical brain regions. As can be seen in figure 1 the first network, ‘a’, is mainly formed by the thalamus and consists of the nucleus accumbens, caudate nucleus, hippocampus, lingual gyrus, and cerebellum. Network ‘b’ is a cluster of the lateral occipital cortex, precuneus and
10 the paracingulate gyrus, subcallosal cortex, operculum cortex, and precuneus. Network ‘d’ consists of the anterior cingulate cortex, middle frontal gyrus, precentral gyrus, and frontal medial cortex. The fifth network ‘e’ consists of the temporal pole and temporal fusiform cortex cluster. Network ‘f’ consists mostly of the putamen, but also includes caudate nucleus (and insular cortex) and the superior parietal lobule. The ‘g’, ‘h’ and ‘i’ networks consist of the cerebellum. The MRI-data collected from the PD patients is used to calculate
characteristic values representing network integrity scores.
Figure 2. The structural covariance networks (SCN) (‘a’ to ‘i’) defined in healthy controls by Hafkemeijer. The figure shows which structures co-vary (red-yellow). The SCN were used as regions of interest in the current study. Figure was adapted from Hafkemeijer et al. 2014.
2.3 Data analysis
SPSS, version 23, is used to analyze all data. First the datum is checked for parametric assumptions. Correlations are studied between scores of different neuropsychological tests
network a Thalamus
network b Lateral occipital cortex
network c Posterior cingulate cortex
network d Anterior cingulate cortex
network e Temporal pole
network f Putamen
network g Cerebellum
network h Cerebellum
11 and integrity scores of the structural covariance networks. The Bonferroni method is used to correct for multiple comparisons. An effect size or correlation of r = ± 0.1 represents a small effect, r = ± 0.3 represents a medium effect and r = ± 0.5 a large effect.
The differences between the DBS-eligible and DBS-ineligible groups are analyzed by independent sample t-tests and Mann-Whitney tests for comparison of equality of means. Ideally all measures are entered in a logistic regression afterwards, however one important assumption of logistic regression is sufficient large sample size per group (n = 10 per group for every variable entered into the equation). Our sample size of the DBS-ineligible group is n = 9. Although the assumption is not met, the neuropsychological tests with the biggest effect size will be entered in the logistic regression. This might give an indication of the amount of variance explained by these two tests, however these results must be interpreted with caution. A p-value ≤ 0.05 indicates statistical significance, a p-value > 0.05 and < 0.10 indicates trend significance.
3. Results
The data of 44 participants were included in the analysis. These data did not contain any outliers that needed to be corrected. The mean age of participants was 61.3 (SD = 6.4, min. 47, max. 72). Between the two groups, DBS-eligible and DBS-ineligible, there was no significant difference between men and women; Fishers exact test (1) = 0.061, p = 0.586, age t(42) = 1.31, p = 0.20 and education t(42) = -0.40, p = 0.69. The cognitive and mood measures were compared for equality of means between the DBS-eligible and DBS-ineligible groups using a t-test and Mann-Whitney U. Overall the DBS-eligible participants performed better on all cognitive and behavioral tests. Although, not all of the between group differences were significant.
Patients did not score significantly different on the RAVLT total and delayed score, the SWRT, SCNT, HADS anxiety and depression and BDI. However, a trend towards significance can be observed for the RAVLT total score and the SCNT.
12 Patients did score significantly different on the CAMCOG – R, SIT, TMT A and B as shown in table 1.
Table 1. Cognitive performance and mood in patients eligible and ineligible for DBS.
DBS-eligible DBS-ineligible M SD n M SD n p d CAMCOG – R † 91.7 5.9 31 82.3 11.9 9 0.026** 1.00 WMS – R 59.3 9.7 31 51.3 12.4 9 0.046** 0.72 RAVLT total 37.7 9.6 33 31.2 5.9 9 0.063* 0.82
RAVLT delayed reproduction 7.6 3.6 33 6.0 2.3 8 0.229 0.53
SWRT 54.5 9.2 35 62.0 15.2 8 0.216 -0.60 SCNT 70.7 11.8 35 79.6 8.7 8 0.053* - 0.86 SIT † 117.5 31.2 35 165.4 80.0 7 0.031** - 0.36 TMT A † 51.3 26.0 35 68.2 29.0 9 0.043** - 0.61 TMT B † 114.8 46.9 35 234.4 121.1 7 0.001** - 1.30 HADS anxiety 6.1 3.8 29 7.3 3.9 9 0.402 - 0.31 HADS depression 5.4 2.7 29 6.4 3.8 9 0.352 - 0.30 BDI † 9.8 5.4 33 12.4 4.9 8 0.084* - 0.50
Note. CAMCOG-R = the number of correct answers on the Cambridge Cognitive Examination – revised. WMS - R = number of correct answers on the Weschler Memory Scale - Revised. RAVLT total = the number of words recalled in five direct recall moments in the Rey auditory verbal learning test. RAVLT delayed reproduction = the number of words recalled after twenty minutes of delay. SWRT = the amount of seconds participants needed to perform the Stroop word reading test. SCNT = the amount of seconds participants needed to perform the Stroop color naming test. SIT = the amount of seconds needed to perform the Stroop interference task. TMT A = the amount of seconds needed to perform the trail making test A. TMT B = the amount of seconds needed to perform the trail making test B. HADS anxiety = is the amount of anxiety problems reported on the Hospital anxiety and depression scale. HADS depression = is the amount of depressive problems reported on the Hospital anxiety and depression scale.
M = Mean. SD = Standard Deviation. N = sample size. Significant p-values are printed in bold.
* = p < 0.10; ** = p < 0.05; d = Cohen’s d for effect size. † = Mann-Whitney U test for equality of means.
A logistic regression analysis was performed to ascertain the effects of the TMT-B and the CAMCOG-R scores on the likelihood that patients were DBS eligible. The TMT-B and the CAMCOG-R were chosen based on their effect sizes. Of the 44 included patients, the data of 6 was missing due to 2 participants not completing the TMT-B and 4 patients not completing the CAMCOG-R. The logistic regression model was statistically significant, χ2(2) = 15.964, p
13 < .0005. The overall prediction of the model rose from 81.6% a priori to 89.5% a posteriori, with a sensitivity of 57.1% and a specificity of 98.6% as shown in table 2.
Table 2. Comparison of the predicted and observed numbers of (in)eligible patients.
predicted ineligible predicted eligible % cor
observed ineligible 4 3 57,1
observed eligible 1 30 96,8
Note. % cor = is the percentage of correctly classified patients.
The Wald criterion demonstrated that only the TMT B made a significant contribution to the prediction p = .04, OR = .98, 95% CI [.95, .99]. Because the assumption that suggests that there should be 10 cases for each independent variable (Agresti, 2007) was not met, the results of this test must be interpreted with caution. A discrimination analysis (using the leaving-one-out method) for cross validation of the logistic regression analysis showed similar results.
Secondly, we looked at correlations between SCNs and cognitive measures. The correlations not corrected for multiple comparison are shown in table 3. Contribution to the stability of the lateral occipital network ‘b’ was correlated with the depression scale of the HADS, r = -.47, p = <0.01. Less contribution to the stability of the anterior cingulate cortex network ‘c’ was significantly correlated to anxiety r = - .36 and depression r = - .42 (all ps < .05) as measured by the HADS. The time needed to complete TMT B showed a significant correlation with the thalamic network ‘a’, τ = -.24, p = < .05. This network also correlated significantly with mood problems measured by the BDI, τ = -.25, p < .05. There was a significant negative relationship between the performance on the Wechsler Memory Scale and patients’ contribution to the cerebellar network ‘h’, τ = -.40, p < .05.No correlations are significant when corrected for multiple comparison with the Bonferroni method; significant findings at p < 4.6 *10^-4. As shown in table 3 all tests had missing data, the tasks were not finished because there was either too little time to complete, force majeure or the neuropsychologist deciding the test was not necessary. Except for one case in which the participant was unable to perform the delayed reproduction of RAVLT and the TMT B.
14
Table 3. Correlations between cognitive tests and structural covariance networks. Global cognition and memory
CAMCOG-R WMS RAVLT total
RAVLT reproduction τ p r p r p r p network a .006 (.960) -.240 (.140) -.192 (.235) -.110 (.504) network b .058 (.614) .137 (.406) -.070 (.669) -.074 (.654) network c .224 (.052) .064 (.699) -.056 (.732) -.018 (.914) network d .073 (.528) -.078 (.636) -.136 (.403) .002 (.992) network e -.058 (.614) -.036 (.828) -.280 (.080) -.164 (.318) network f .070 (.545) .084 (.612) .044 (.787) .184 (.261) network g -.020 (.860) -.106 (.520) .101 (.536) .126 (.445) network h -.217 (.059) -.395 (.013) -.192 (.236) -.115 (.487) network i -.145 (.207) -.280 (.084) -.152 (.350) -.123 (.454) N 38 39 40 39 Missing 6 5 4 5 Executive functioning
SWRT SCNT SIT TMT A time TMT B time
r p r p τ p τ p τ p network a -.072 (.657) -.064 (.691) -.048 (.666) -.132 (.220) -.239 (.028) network b -.088 (.583) .035 (.826) .143 (.196) -.027 (.802) .109 (.317) network c .106 (.511) .164 (.304) .037 (.735) .008 (.939) -.021 (.848) network d .020 (.900) .115 (.472) .156 (.158) .121 (.264) .136 (.212) network e -.166 (.301) .232 (.144) .127 (.249) .032 (.770) .109 (.317) network f .141 (.380) .058 (.721) .030 (.789) .029 (.786) .121 (.266) network g .201 (.208) -.116 (.472) -.076 (.492) .053 (.625) -.104 (.339) network h .252 (.112) .200 (.210) .107 (.333) .161 (.137) .058 (.597) network i .217 (.173) .087 (.590) .084 (.449) .079 (.467) -.031 (.779) N 41 41 40 42 41 Missing 3 3 4 2 3
Mood and Anxiety
HADS anxiety HADS depression BDI r p r p τ p network a -.060 (.722) -.025 (.882) -.245 (.033) network b -.196 (.244) -.437 (.007) -.127 (.268) network c -.350 (.034) -.401 (.014) -.169 (.141) network d -.108 (.524) -.257 (.124) -.085 (.458) network e .147 (.384) .063 (.713) .074 (.519) network f -.067 (.693) -.237 (.157) -.147 (.201) network g -.114 (.502) .036 (.832) -.029 (.798) network h -.033 (.847) .013 (.938) .035 (.761) network i -.089 (.601) -.055 (.745) -.094 (.415) N 37 37 39 Missing 7 7 5
Note. Values in the body of the table represent correlation (p value). Kendall's τ and Pearson's r correlation coefficients were used in accordance to parametric and nonparametric.* = p <0.05. Significant correlations before correction for multiple comparison in bold.
15 4. Discussion
This study shows that ineligible patients performed worse on tests for general cognitive ability (CAMCOG-R), general memory (WMS), psychomotor speed (TMT-A), and cognitive flexibility (TMT-B and SIT). Whereas ineligible patients did not perform significantly different on verbal memory (RAVLT), processing speed (SWRT and SCNT), and depression and anxiety (HADS-anxiety, HADS-depression and BDI). When a logistic regression analysis was used to classify patients as (in)eligible using the CAMCOG-R and TMT-B, the initial prediction rose with 7,9 percent to 89,5 percent, indicating a contribution of
(neuro)psychology to the multidisciplinary consensus. After correction for multiple comparison none of the aforementioned (neuro)psychological outcomes correlated with SCNs.
Aarsland et al. (2010) labeled memory deficits, specifically verbal memory, as the most commonly reported problem in PD patients with MCI. Following this, we expected the DBS-ineligible group to have more degenerative symptoms, to be closer to or have MCI and thus more memory deficits. This however does not seem to be the case when looking at the verbal memory test, RAVLT. The WMS nonetheless does show a difference between the two groups indicating poorer performance by the DBS-ineligible group.
Researchers are trying to find (bio)markers in traditional methods, like neuropsychological assessment, and by developing novel methods, like structural covariance networks. Although SCNs seem to detect early grey matter changes (Coppen et al., 2016) and place them in a comprehensive network model, no relation has been found between neuropsychological data and SCN analysis in the current study. This is contrary to results reported by Hafkemeijer (2015), who found that cognitive dysfunction corresponded to impaired brain networks in normal aging individuals.
The aforementioned lack of relation between neuropsychological data and SCN analysis might be the result of the specific methods used in the current study. To correct for multiple comparisons the Bonferroni method was used, which is a conservative correction method.
16 Although this is an effective and straightforward approach, it tends to be strict when lots of tests are performed. The current study performs 108 correlations – between SCNs, cognitive and behavioral measures – rendering the chance of finding significant correlations almost nihil. Next to that the small sample size, especially in the DBS ineligible group, reduces the likelihood of detecting a true effect. Along with the rather small sample size, the datum was collected during a clinical screening of a very specific subgroup of PD-patients.
Consequently, results of the current study must be interpreted with caution when making statements about the general PD-population.
Taking this into consideration the current study does show some remarkable results that, to the knowledge of the author, have not yet been reported. As said, the group membership of patients is based upon (neuro)psychological outcomes – among other clinimetrics.
Consequently, there is an incorporation bias, which presupposes poorer performance on all tests by the DBS ineligible patients. However, there are test outcomes that did not differ between the two groups (suggesting no contribution to the multidisciplinary consensus). Furthermore, logistic regression analysis has shown an improvement of the likelihood of assigning patients to the correct group when just the TMT-B and the CAMCOG-R are predictors.
In light of the findings, one might propose an alteration of the screening battery for DBS. Logically, deletion of (neuro)psychological tests that do not seem to differ between the two groups should be more time and cost efficient. However, we have solely looked at the quantitative data in the current study. Neuropsychological assessment is most and foremost a clinical assessment which integrates this quantitative data, with observations and amnestic information. Ergo, the whole is greater than the sum of its parts. Nonetheless, the current study gives an interesting and novel insight in the contribution of neuropsychological tests to a multidisciplinary consensus. Future studies with bigger sample sizes should aim to replicate the current findings before decisions concerning the screening protocol can be made. An extension of the study looking at the (neuro)psychological outcomes after the DBS intervention, in the form of a longitudinal follow up study with inclusion of post-operative
17 data, could provide a better understanding of the difference in cognitive profiles. Moreover, such a study might give insight in patients who were found DBS eligible but in hindsight show complications like aggravated cognitive decline.
In conclusion, this study found no relation between (neuro)psychological performance and SCNs in PD patients who were candidates for DBS. Attentional, global memory and global cognition tasks seem to contribute to the multidisciplinary consensusof patients’
(in)eligibility for DBS. Whereas verbal memory tasks and questionnaires for depression and anxiety do not.
18 References
Aarsland, D., Bronnick, K., Williams-Grey, C., Weintraub, D., Marder, K., Kulisevsky, J., et al. (2010). Mild cognitive impairment in Parkinson disease A multicenter pooled analysis. Neurology, 75(12), 1062-1069.
Agresti, A., & Franklin, C. (2007). The art and science of learning from data. Upper Saddle River, NJ: Prentice Hall.
Alexander-Bloch, A., Giedd, J. N., & Bullmore, E. (2013). Imaging structural co-variance between human brain regions. Nature Reviews Neuroscience, 14(5), 322-336.
American Psychiatric Association (2014). Beknopt overzicht van de criteria (DSM-5). Nederlandse vertaling can de Desk Referenceto the Diagnostic Criteria from DSM-5. Amsterdam: Boom.
Beck, A.T., Steer, R.A., & Brown, G.K. (1996). Manual for the Beck Depression Inventory-II. San Antonio, TX: Psychological Corporation.
Bouma, A., Mulder, J., Lindeboom, J., & Schmand, B. (2012). Handboek
Neuropsychologische Diagnostiek.-2e herz. dr. Amsterdam: Pearson Assessment and Information B.V.
Brichta, L., Greengard, P., & Flajolet, M. (2013). Advances in the pharmacological treatment of Parkinson's disease: targeting neurotransmitter systems. Trends in neurosciences, 36(9), 543-554.
19 Bronstein, J. M., Tagliati, M., Alterman, R. L., Lozano, A. M., Volkmann, J., Alessandro
Stefani, et al. (2011). Deep brain stimulation for Parkinson disease: an expert consensus and review of key issues. Archives of neurology, 68(2), 165-170.
Chaytor, N., Schmitter-Edgecombe, M. & Burr, R. (2006). Improving the Ecological Validity of Executive Functioning assessment, Archives of Clinical Neuropsychology, 21, 217-227.
Chou, K. H., Lin, W. C., Lee, P. L., Tsai, N. W., Huang, Y. C., Chen, H. L., et al. (2015). Structural covariance networks of striatum subdivision in patients with Parkinson's disease. Human brain mapping, 36(4), 1567-1584.
Draijer, W., Eizenga, W., Sluiter, A., Opstelten, W., & Goudswaard, A. N. (2011). NHG-STANDAARD-Ziekte van Parkinson. Huisarts en Wetenschap, 54(7), 380.
Hafkemeijer, A., Altmann-Schneider, I., Craen, A. J., Slagboom, P. E., Grond, J., & Rombouts, S. A. (2014). Associations between age and grey matter volume in anatomical brain networks in middle-aged to older adults. Aging cell, 13(6), 1068-1074.
Hammes, J.G.W. (1971). De Stroop Kleur-Woord Test. Handleiding. Amsterdam: Pearson Assessment and Information B.V.
Jenkinson, M., Bannister, P., Brady, J. M. and Smith, S. M. (2002) Improved Optimisation for the Robust and Accurate Linear Registration and Motion Correction of Brain Images. NeuroImage, 17(2), 825-841.
20 Jenkinson, M., Beckmann, C.F., Behrens, T.E., Woolrich, W.M., & Smith S.M. (2012). FSL.
NeuroImage, 62, 782-790.
Jensen, A.R., & Rowher, W.D. (1966). The Stroop-Color Word Test: A Review. Acta Psychologica, 25, 36-93.
Jellinger, K. A. (2012). Neuropathology of sporadic Parkinson's disease: evaluation and changes of concepts. Movement Disorders, 27(1), 8-30.
Lindeboom, J., Horst, R., Hooyer, C., Dinkgreve, M. A. H. M., & Jonker, C. (1993). Some psychometric properties of the CAMCOG. Psychological medicine, 23(01), 213-219.
Lindeboom, R., Schmand, B., Holman, R., De Haan, R. J., & Vermeulen, M. (2004). Improved brief assessment of cognition in aging and dementia. Neurology, 63(3), 543-546.
Litvan, I., Aarsland, D., Adler, C. H., Goldman, J. G., Kulisevsky, J., Mollenhauer, B., et al. (2011). MDS task force on mild cognitive impairment in Parkinson's disease: Critical review of PD - MCI. Movement disorders, 26(10), 1814-1824.
Okun, M. S. (2012). Deep-brain stimulation for Parkinson's disease. New England Journal of Medicine, 367(16), 1529-1538.
Parsons, T. D., Rogers, S. A., Braaten, A. J., Woods, S. P., & Tröster, A. I. (2006). Cognitive sequelae of subthalamic nucleus deep brain stimulation in Parkinson's disease: a meta-analysis. The Lancet Neurology, 5(7), 578-588.
21 Patenaude, B., Smith, S.M., Kennedy, D., & Jenkinson M. A (2011). Bayesian Model of
Shape and Appearance for Subcortical Brain. NeuroImage, 56(3), 907-922.
Rodriguez, R. L., Fernandez, H. H., Haq, I., & Okun, M. S. (2007). Pearls in patient selection for deep brain stimulation. The neurologist, 13(5), 253-260.
Roth, M., Tym, E., Mountjoy, C.Q., Huppert, F.A., Hendrie, H., Verma, S., & Goddard, R. (1986). CAMDEX. A standardised instrument for the diagnosis of mental disorder in the elderly with special reference to the early detection of dementia. British Journal of Psychiatry, 149, 698-709.
Saan, R.J.,& Deelman, B.G. (1986). De 15-Woorden Tests A en B (Een voorlopige handleiding). Afdeling Neuropsychologie: Groningen.
Seeley, W. W., Crawford, R. K., Zhou, J., Miller, B. L., & Greicius, M. D. (2009).
Neurodegenerative diseases target large-scale human brain networks. Neuron, 62(1), 42-52.
Smith, S.M. et al. (2004). Advances in functional and structural MR image analysis and implementation as FSL. NeuroImage, 23(S1), 208-19
Smith, S. M., Zhang, Y., Jenkinson, M., Chen, J., Matthews, P. M., Federico, A., & De Stefano, N. (2002). Accurate, robust, and automated longitudinal and cross-sectional brain change analysis. Neuroimage, 17(1), 479-489.
Spinhoven, Ph., Ormel, J., Sloekers, P. P. A., Kempen, G. I. J. M., Speckens, A. E. M., & Hemert, A.M. van. (1997). A validation study of the Hospital Anxiety and Depression
22 Scale (HADS) in different groups of Dutch subjects. Psychological Medicine, 27, 363-370.
Strauss, E., Sherman, E. M., & Spreen, O. (2006). A compendium of neuropsychological tests: Administration, norms, and commentary. American Chemical Society.
Verbaan, D., Marinus, J., Visser, M., Van Rooden, S. M., Stiggelbout, A. M., & Van Hilten, J. J. (2007). Patient-reported autonomic symptoms in Parkinson disease. Neurology, 69(4), 333-341.
Verhey, F. R., Huppert, F. A., Korten, E. C., Houx, P., De Vugt, M., Van Lang, N., et al. (2003). Cross-national comparisons of the Cambridge Cognitive Examination-revised: the CAMCOG-R Results from the European Harmonization Project for Instruments in Dementia. Age and ageing, 32(5), 534-540.
Verhey, F. R., Houx, P., van Lang, N., Huppert, F., Stoppe, G., Saerens, J., Böhm, et al. (2004). Cross-national comparison and validation of the Alzheimer's Disease Assessment Scale: results from the European Harmonization Project for Instruments in Dementia (EURO‐HARPID). International journal of geriatric psychiatry, 19(1), 41-50.
Vingerhoets, G., Verleden, S., Santens, P., Miatton, M.. & De Reuck, J. (2003). Predictors of Cognitive Impairment in Advvanced Parkinson’s disease. Journal of Neurology, Neurosurgery and Psychiatry, 74, 793-796.
Wang, Y. P., & Gorenstein, C. (2013). Psychometric properties of the Beck Depression Inventory-II: a comprehensive review. Revista Brasileira de Psiquiatria, 35(4), 416-431.
23 Wechsler, D. (1984). Wechsler Memory Scale-Revised Manual. London, UK: The
Psychological Corporation.
Wolters, E. C. (2007). Parkinsonisme. In Het Geriatrie formularium. (pp. 450-458). Bohn Stafleu van Loghum.
Woolrich, M.W., Jbabdi,S. ,Patenaude, B., Chappell, M., Makni,S., Behrens, T., Beckmann, C., Jenkinson, M., Smith, S.M. (2009) Bayesian analysis of neuroimaging data in FSL. NeuroImage, 45, 173-186.
Worster, A., & Carpenter, C. (2008). Incorporation bias in studies of diagnostic tests: how to avoid being biased about bias. CJEM, 10(02), 174-175.
Zigmond, A. S. & Snaith, R. P. (1983). The Hospital Anxiety and Depression Scale. Acta Psychiatrica Scandinavia 67, 361–370.
24 Appendix A
All participants had to follow a neuropsychological protocol, in which global cognitive abilities are measured by means of the CAMCOG-R (Roth et al., 1986). This test is a general cognitive battery, screening for dementia in people over 65. The CAMCOG-R is reasonably validated for convergent and ecologic validity (Verhey et al, 2003; Verhey et al. 2004). It has a high internal consistency, Cronbach’s α = 0.93 (Lindeboom, Schmand, Holman, de Haan & Vermeulen, 2004). The test–retest reliability is also high (r = 0.97; Lindeboom, Horst, Hooyer, Dinkgreve & Jonker, 1993).
Memory was measured by means of the WMS - R (Wechsler, 1945) and the Dutch translation of the RAVLT (Saan & Deelman, 1986). There have not been many recent studies concerning the psychometric properties of the WMS, . Therefor there is not a lot of information on the validity of the WMS. The WMS tests verbal, visual and associative memory, it also has some (relatively) simple attentional tasks.
The Dutch translation of the RAVLT tests verbal memory by asking people to learn 15 words. After 20 minutes in the delayed recall phase the patient has to tell the examiner which words he still remembers. The Dutch version of the RAVLT has an high retest-reliability (the total good score r = 0,87 and delayed reproduction score r = 0,86; Bouma, Mulder,
Lindeboom & Schmand, 2012). The construct validity has been measured by performing a principal component analyses with other memory and executive function test, there was concluded that the RAVLT distinguishes as a verbal memory test (Vingerhoets, Verleden, Santen & de Reuck, 2003).
Executive functioning and attention was assessed by means of the Trail Making Test A and B (TMT A and B) and the Stroop Color-Word Interference Test (Stroop; Hammes, 1971). The TMT is widely used measurement for executive functioning. The predictive value of this test for executive disorders increases when it’s used as a predictor in combination with other executive functioning tests (Chaytor, Schmitter-Edgecombe & Burr, 2006). The test-retest
25 reliability is reasonable for TMT-A (r = 0,79) and high for TMT-B (r = 0,89) (Bouma et al., 2012).
The Stroop test (Hammes, 1971) has an high construct validity and is one of the most commonly used executive functioning tests. The task measures mental speed, response inhibition and executive attention (Bouma, Mulder, Lindeboom, Schmand, et al, 2012). Mood was assessed by means of the Hospital Anxiety and Depression Scale (HADS; Zigmond & Snaith, 1983) and Beck’s Depression Inventory (BDI-II; Beck, Steer & Brown, 1996). The HADS is regarded as a reliable (r = 0,80) but screenings questionnaire
(Spinhoven, Ormel, Sloekers, Kempen, Speckens & van Hemert, 1997). The BDI-II is considered to have a good reliability and a sufficient construct validity (Wang & Gorenstein, 2013).
