Title: Are there Post-Stroke Memory Impairments and Which Lobe is Related to These Im-pairments?
Name: Sijtsma, H
Student number: 10178570
Research Masterthese Clinical Neuropsychology
Supervisor: dhr. prof. dr. E. H. F. de Haan & N. A. Lammers, MSc Date: 23-08-2016
Are there Post-Stroke Memory Impairments and Which
Lobe is Related to These Impairments?
Hester Sijtsma
Department of PsychologyUniversity of Amsterdam
Abstract
Post-stroke cognitive impairments may have a huge impact on someone’s life. The two cogni-tive domains execucogni-tive functions and cognicogni-tive processing speed are often impaired after having suffered a stroke. Less clarity exists about whether there are post-stroke memory impairments and if so, which components of memory are impaired. Traditionally, memory is linked to the medial temporal lobe. However, recent studies have also linked the parietal lobe and the frontal lobe to memory impairments. In the current study, post-stroke effects on working memory, short-term memory and long-term memory were investigated. Additionally, the test perfor-mances were related to the location of the lesion (frontal, temporal, parietal or occipital lobe). Seventeen patients were assessed on the Rey Auditory Verbal Learning Test and the Wechsler Adult Intelligence Scale-IV Digit Span Test and underwent MRI imaging. The patients did not perform aberrantly on the memory tests. One explanation for the absence of post-stroke memory impairments may be the patients’ relatively young age. In addition, no relationship between the memory test performances and the location of the lesion was found. In further research, memory should perhaps be framed in terms of a network that is distributed over different lobes instead of located within one lobe.
1
Introduction
According to the Dutch Ministry of Health, Welfare and Sport, 411,100 Dutch citizens have suffered a stroke in 2014 (male: 205,300, female: 205,800, total number of citizens in The Netherlands in 2014: 16,828,000), from which 9293 died (Centraal Bureau voor de Statistiek, 2016; Rijksinstituut voor Volksgezondheid en Milieu, 2016). Stroke refers to a collection of diseases in which the blood supply in the brain is disrupted. Stroke can occur in different arteries in the brain. According to a large study (N = 2213), 51% of the strokes were middle cerebral artery strokes, 7% were posterior cerebral artery strokes, 5% were anterior cerebral artery strokes, 13% were small vessel strokes, 11% were brainstem strokes, 4% were cerebellar strokes, and 9% of the patients suffered a stroke in multiple territories (Ng, Stein, Ning, & Black-Schaffer, 2007). In acute phase, stroke can often be recognized by a crooked mouth, paralysis of the arm or troubled speech (Hartstichting, n.d.). In comparison with healthy people, patients having suffered a stroke experience problems especially in physical functioning, social functioning and their perceived health status (Rijksinstituut voor Volksgezondheid en Milieu, 2016). In a follow-up study, it was found that during a 11-year period 20% of young patients (18-50 years) having suffered a stroke passed away (Rutten-Jacobs et al., 2013). The Dutch Ministry expects more citizens to suffer a stroke because of an increase of people that are overweight and an increase in people’s salt intake (Rijksinstituut voor Volksgezondheid en Milieu, 2016).
Apart from experiencing consequences in physical functioning and social functioning, patients also experience post-stroke cognitive impairments. Cognitive impairments may be more common after having suffered a stroke than is generally assumed, but prevalence rates vary due to the different criteria that are used to define cognitive deficits and due to sample differences (Rasquin et al., 2004). Estimates of the prevalence rate of post-stroke cognitive impairments vary from 10% to 82% (De Haan, Nys, & Van Zandvoort, 2006). In addition to the prevalence rates, the severity of post-stroke cognitive impairments varies. Patients may develop vascular dementia (prevalence rate one year post-stroke: 7.7%.), post-stroke mild cognitive impairment (prevalence rate one year post-stroke: 51.5%) or they may develop no cognitive symptoms at all (prevalence rate one year post-stroke: 20.6%) (Rasquin et al., 2004).
A literature search was conducted to find out which cognitive domains are most impaired after having suffered a stroke. This resulted in 21 articles; see Table 1 in Appendix 1. Analyzing the results of these articles, it is clear that information processing speed and executive functions are often impaired. This is in line with a review article by Cumming, Marshall, and Lazar (2013). They concluded that attention and executive functions are often most impaired after having suffered a stroke.
Cumming et al. (2013) argued that the pattern of post-stroke cognitive impairments mainly consists of executive disfunctions and attention impairments, while memory is less impaired or even remains unimpaired. However, the results from the 21 reviewed articles (Table 1) are some-what contradictory to the pattern mentioned by Cumming et al. (2013). In Table 1, the 21 articles were divided into three categories: (1) memory was impaired after having suffered a stroke (10 articles), (2) memory remained unimpaired or memory was one of the least impaired cognitive domains after having suffered a stroke (5 articles) and (3) some memory components were impaired after having suffered a stroke while other memory components remained unim-paired or were least imunim-paired after having suffered a stroke (6 articles). Based on these results, it looks like there is no clear conclusion possible about whether there are post-stroke memory impairments and if so, which kind of memory components are impaired.
1.1
Post-stroke memory impairments
To obtain a better understanding of the post-stroke memory impairments, an overview was made that shows per study which memory components were impaired and which memory components remained unimpaired; see Table 2 in Appendix 1. What should be realized is that the studies shown in Table 2 are a selection of the literature on post-stroke cognitive impairments, so Table 2 can provide an incomplete picture.
Table 2 shows that most researchers made a distinction between verbal memory and visual memory. It is remarkable that not everyone differentiated between immediate recall and delayed recall. In Table 2, the 21 articles were divided into six categories. The first category consisted of four articles that investigated the difference between post-stroke effects on immediate recall and on delayed recall for both verbal memory and visual memory, and the four articles reported results about all four test performances. The second category consisted of two articles that investigated verbal memory and visual memory, reported results about the difference in performance between the two kinds of memory but did not distinguish immediate recall and delayed recall. The two articles in category 3 investigated verbal memory and visual memory but did not report results with respect to a difference in performance and they did not distinguish immediate recall and delayed recall. In category 4, three articles only investigated verbal memory and within verbal memory, they distinguished immediate recall and delayed recall. Category 5 consisted of five articles that neither differentiated between visual memory and verbal memory nor immediate recall and delayed recall. The last category consisted of five articles that did not fit the other categories because of various reasons, such as that only working memory was investigated or only delayed recall of verbal memory was investigated. Furthermore, only eight articles investigated post-stroke effects on working memory (Ballard et al., 2003; Nys, Van Zandvoort, Van der Worp, Kappelle, & De Haan, 2006; Peng et al., 2016; Sachdev et al., 2004; Schaapsmeerders et al., 2013; Srikanth et al., 2003; Van Rooij et al., 2014; Viswanathan et al., 2015).
From the literature study, I conclude that whether memory is impaired or remains unimpaired is probably dependent on the way memory is operationalized. Although, some researchers have looked at post-stroke effects on working memory, immediate memory recall and on delayed memory recall seperately, it is still unclear which memory components are impaired and which memory components remain unimpaired after having suffered a stroke. Memory components should be investigated seperately because they can all be impaired, but it could also be the case that one component is affected while the other components remain unaffected. Therefore, in the current study, the post-stroke effects on working memory, immediate recall and delayed recall were investigated seperately. This will hopefully result in a clearer view on which memory components are impaired after a stroke and which components remain unimpaired.
1.2
Different memory components
In this study, post-stroke effects on working memory, immediate recall and delayed recall were investigated. In this paper, the term ‘long-term memory’ will be used to refer to delayed recall and is defined as the act of remembering different kinds of knowledge for a longer period of time and putting this knowledge to use (Radvansky, 2011). ‘Short-term memory’ refers to immediate recall and means keeping a limited amount of knowledge active for a very short time (less than a minute), while not manipulating the knowledge (Radvansky, 2011). The term ‘working memory’ refers to actively manipulating knowledge that is stored in short-term memory (Radvansky, 2011).
Already, in the Atkinson-Shiffrin human memory model, a distinction was made between short-term memory and long-term memory (Atkinson & Shiffrin, 1968; Radvansky, 2011). Their model consisted of three parts. First, information from the different senses is processed in the
sensory registers. Attention is being paid to information that is worth processing further in short-term memory, which is the second part of their model. In short-term memory, information is kept active for a short time, while it is not being manipulated. The capacity is limited and information decays rapidly. Through rehearsal, information ends up in long-term memory. This is the final part of their model and it has a limitless capacity.
The Atkinson-Shiffrin model has received criticism. One point of criticism was focused on the existence of different kinds of long-term memories instead of one general long-term memory. According to Tulving, a distinction can be made between episodic memory and semantic memory (Tulving, 1972). Episodic memory entails memories for events and experiences, which are bound to a specific time and place. Semantic memory consists of memories for general, factual knowledge (Tulving, 1972). This distiction is noticeable in Alzheimer’s disease (AD) and semantic dementia (SD), in which patients with AD are more impaired on episodic memory tests than patients with SD, while patients with SD are more impaired on semantic memory tests than patients with AD (La Joie et al., 2014). In the current study, post-stroke episodic memory effects were investigated. Baddeley’s criticized the Atkinson-Shiffrin model because, according to Baddeley, the short-term memory consists of several components rather than one unitary component (Baddeley, 2000). Therefore, Baddeley developed a theory concerning working memory. According to this theory, one working memory component processes, stores and manipulates verbal information and another component processes, stores and manipulates visual information. Both components are controlled by a component called the central executive. A fourth component unifies the different kinds of information and combines the unified information with information stored in long-term memory. According to Baddeley’s model, verbal information is processed, stored and rehearsed in a specialized component called the phonological loop (Baddeley, 2000; Radvansky, 2011). In the current study, post-stroke verbal working memory effects, which are processed in the phonological loop, were investigated.
1.3
The anatomy of memory
In the current study, post-stroke memory impairments were investigated and, in addition, the impairments were related to the anatomy of the brain. So, the relationship between anatomical structures and the performance on working memory tests, on short-term memory tests and on long-term memory tests was investigated.
Traditionally, the medial temporal lobe, with a special role dedicated to the hippocampus, has often been linked to episodic memory (Miyashita, 2004; Squire & Zola, 1998). A review study by Rugg and Vilberg (2013), concluded that episodic memory is related to a network that consists of the hippocampus in the middle, which has important connections with different cortical areas depending on the content of the memory. Also, patients with medial temporal lobe lesions performed worse when constructing stories about past and future events, which is considered to be a measure of episodic memory, than healthy controls (Race, Keane, & Verfaellie, 2011). One of the studies mentioned in Table 1 reported that stroke lesions in the temporal lobe were related to worse recovery of visual memory capabilities, which are considered to be part of episodic memory (Nys et al., 2005).
Suprisingly, recent studies have reported a link between the posterior parietal cortex (PCC) and episodic memory. Schoo et al. (2011) discussed a so-called paradox about the role of the PCC in memory. The paradox implied that many fMRI studies indicate PCC activity during episodic memory tasks but this finding was not supported by lesion studies. One explanation that Schoo et al. (2011) discuss concerning the PCC activity in fMRI studies, is that effort determines the PCC activity during retrieval processes, because when people encoded the stimuli in a deep manner, PCC activity increased compared to when people encoded stimuli in a shallow manner
(Shannon & Buckner, 2004, discussed in Schoo et al., 2011). Cabeza, Ciaramelli, Olson, and Moscovitch (2008) also discussed this paradox and suggested that parietal lobe activity reflects top-down and bottom-up attentional processes that guide memory processes, which are mediated by the medial temporal lobe. So, perhaps there is no direct causal relationship between parietal lobe lesions and post-stroke memory impairments after all.
There are two recent studies, both not dicussed in the review of Schoo et al. (2011) and in the review of Cabeza et al. (2008), in which it is shown that patients with ventral parietal lobe lesions and patients with posterior parietal lobe lesions experience episodic memory deficits (Ben-Zvi, Soroker, & Levy, 2015; Berryhill, Picasso, Arnold, Drowos, & Olson, 2010). In another study, it was argued that parietal lobe lesions do not cause memory impairments, but instead patients may feel less confident about their own memory skills (Simons, Peers, Mazuz, Berryhill, & Olson, 2009).
Parietal lobe lesions have also been related to impairments in retrieving information that is processed in working memory and to impairments in manipulating information (part of working memory) (Berryhill & Olson, 2008; Koenigs, Barbey, Postle, & Grafman, 2009). According to Baddeley, the storage part of the phonological loop, a component of working memory that was investigated in the current study, is linked to the inferior, lateral parietal cortex (Baddeley, 2000). This hypothesis was supported by Baldo and Dronkers (2006), who found that patients with inferior parietal lobe lesions were impaired with respect to tasks requiring phonological storage.
There are also studies in which episodic memory impairments are related to frontal lobe lesions. For example, patients with frontal lobe lesions showed impairments on free recall and on recognition of both verbal stimuli and visual stimuli (both are part of episodic memory) (MacPherson, Turner, Bozzali, Cipolotti, & Shallice, 2016). In addition, the frontal lobe is also known for its role in working memory. Barbey, Koenigs, and Grafman (2013) found that patients with a prefrontal cortex lesion were impaired with respect to manipulating verbal information and subsequently reproducing this information. They were not impaired when they only had to repeat information without having to manipulate it. So, frontal lobe lesions caused working memory impairments but no short-term memory impairments. According to Baddeley, the rehearsal part of the phonological loop, a component of working memory that was investigated in the current study, is linked to the inferior frontal cortex (Baddeley, 2000). This hypothesis was supported by Baldo and Dronkers (2006), who reported that patients with inferior frontal lobe lesions were impaired with respect to tasks requiring articulatory rehearsal.
To summarize, more than one lobe has been linked to both episodic memory and working memory. Traditionally the medial temporal lobe has been linked to memory but recent findings also indicate a relationship between memory and the parietal lobe and the frontal lobe. The aim of the current study was to clarify the performance of stroke patients on working memory tests, short-term memory tests and long-term memory tests, and to investigate which lobe is related to these test performances.
2
Method
2.1
Subjects and procedure
The current study is part of the project called ‘A functional Architecture of the Brain for Vision’ (FAB4V study). The FAB4V study takes place at four medical centres in The Netherlands. All centres use the same research protocol and materials. Data of all four centres were used for the current study. The FAB4V study was approved by the ethics committee.
Patients who suffered an ischemic stroke were informed about the study during their stay in the hospital. When they agreed to participate, patients signed the informed consent and they filled out a checklist to make sure there were no contra-indications for MRI scanning. The inclu-sion criteria were the following: aged between 18 and 80 years, Dutch speaking and a diagnosis of ischemic stroke. Exclusion criteria were a history of alcohol or drug abuse, psychiatric disorders, neurological history (except a history of strokes), any other non-neurological disorder influenc-ing cognitive functioninfluenc-ing, pre-existent cognitive decline, pre-existent dependency in activities in daily life and contra-indications for MRI imaging.
Patients were tested four to eight weeks after the incident of the stroke. Testing took place in the same medical centre as where they were hospitalized. The procedure of the FAB4V project consists of a neuropsychological assessment, tasks measuring visual acuity and MRI imaging. The current study used data on two neuropsychological tests and used one MRI image. Testing lasted approximately five hours. Travel expenses and lunch were compensated for.
2.2
Materials
MRI
All centres used a 3 Tesla MRI scanner with a SENSE 32-channel head coil. Three centres used a PHILIPS scanner and one centre used a SIEMENS scanner. MRI scanning took approximately 30-45 minutes. In the current study, a FLAIR image was used to investigate whether the lo-cation of the lesion was related to the test performances. The FLAIR image is the standard procedure in Dutch clinical care to detect stroke lesions. In addition, a review study showed that the FLAIR image is an accurate method to detect lesions caused by a stroke (Makkat et al., 2002). Parameter settings for the FLAIR image were the following:
Echo time: 295.38 ms, repetition time: 4800 ms, flip angle: 90, field of view: 250 mm (foot to head) x 250 mm (anterior-posterior) x 182.56 mm (right-left), reconstructed voxel size: 0.98 mm/0.98 mm/0.56 mm, inversion time: 1650 ms, SENSE (APxRL) = 3x2.
Figure 1 : Lesions within one lobe were seg-mented in the same colour. Red = frontal lobe, yellow = parietal lobe, blue = temporal lobe. To investigate whether the location of
the lesion was related to the test per-formances, the lesions were manually seg-mented. The manual segmentation was completed using the programme ITK-SNAP (Yushkevich & Gerig, 2014). Cortical le-sions caused by recent strokes and by old strokes were segmented. Signs of small ves-sel disease were not included. If a pa-tient had lesions in more than one lobe, le-sions within one lobe were all segmented in the same colour. In this way, le-sions within the same lobe were distin-guished from lesions located in the other lobes, see Figure 1. All segmented im-ages were checked by an experienced neu-roradiologist. Unfortunately, there was no second person to segment the images so inter-rater reliability was not calculated.
Neuropsychological assessment
In the current study, the pattern and the anatomy of post-stroke working memory impairments, short-term memory impairments and long-term memory impairments were investigated. Test results concerning visual memory were unavailable but test results concerning verbal memory were recorded, hence verbal memory was investigated. Working memory was operationalized by means of the backward phase of the Wechsler Adult Intelligence Scale-IV (WAIS-IV) Digit Span Test (Dutch version) (Wechsler, 2012). Short-term memory was operationalized by means of the WAIS-IV Digit Span Test forward phase (Dutch version) (Wechsler, 2012) and by means of the ‘immediate recall’ score on the Rey Auditory Verbal Learning Test (AVLT) (Dutch version) (Saan & Deelman, 1986). Long-term memory was operationalized by means of the ‘delayed recall given the immediate recall’ score on the AVLT (Dutch version) (Saan & Deelman, 1986). Next, descriptions for both tests are given.
Rey Auditory Verbal Learning Test
The AVLT aims to measure short-term verbal memory skills and long-term verbal memory skills by letting the patient remember a list of 15 unrelated words. The words were presented five times and after each time, the patient named the words he still remembered. The total num-ber of rememnum-bered words was transformed to a Z-score and used as a measure for short-term memory. After 20 minutes, the patient had to name the words for a sixth time. The number of words remembered at the sixth time was compared to the total number of words that were remembered after the first five rounds. This resulted into a relative score indicating the number of words remembered after 20 minutes divided by the number of words remembered after the first five rounds. This relative measure was transformed to a Z-score and indicated long-term memory skills. Earlier research shows that, compared to healthy controls, stroke patients ob-tained lower scores on the relative measure, which indicated long-term memory (Bouma, Mulder, Lindeboom, & Schmand, 2012). This means that the long-term memory capabilities of stroke patients deteriorated compared to healthy controls. The Commissie Testaangelegenheden Ned-erland (COTAN) qualified the reliability and the construct validity of the Dutch version of the test as adequate (Commissie Testaangelegenheden Nederland, 2000). The criterium validity was inadequate because of a lack of research and the norms were inadequate because they were too old. Therefore, the current study used norms from the Nederlands Instituut van Psychologen (2012). These norms are based on 847 healthy subjects between 14 and 87 years old and included a correction for sex, age and education.
Wechsler Adult Intelligence Scale-IV Digit Span test
The Wechsler Adult Intelligence Scale-IV is a well-known test battery that aims to measure different components of intelligence. The subtest Digit Span consists of a forward phase and a backward phase. In the forward phase, the patient had to repeat a sequence of numbers while the sequence increases in length. The forward phase aims to measure short-term mem-ory. In the backward phase, the patient had to repeat a sequence of numbers in the reverse order. This sequence also increases in length and aims to measure working memory. Both scores are transformed to a Z-score. Earlier research shows that stroke patients performed worse on the WAIS-III Digit Span Test than healthy controls (Nys et al., 2006). The COTAN quali-fied the norms, reliability and construct validity of the total WAIS-IV test battery as adequate (Commissie Testaangelegenheden Nederland, 2012). The criterium validity was inadequate be-cause no research has been done. Unfortunately, the COTAN did not evaluate the subtest Digit Span seperately. The norms that were used in the current study are described in the manual of the WAIS-IV. These norms are based on 1000 healthy subjects between 16 and 84 years old and
included a correction for age.
2.3
Data analysis
The aim of the current study was to clarify the performance of stroke patients on working memory tests, short-term memory tests and long-term memory tests, and to investigate which lobe is related to these test performances. Data of healthy controls was unavailable so a direct comparison of stroke patients with healthy controls was impossible. Instead, the raw test scores of the patients were compared to norm data of healthy controls and transformed to Z-scores. Mean Z-scores were calculated to find out if the patients performed aberrantly on the tests. Based on clinical practice, a deviant Z-score is Z 6 −2.
The ideal analysis to answer which impaired lobe (containing a frontal, temporal, parietal or occipital lesion) is related to the test performances, is a multivariate (multiple) regression analysis. In this kind of analysis, the relationship between several independent variables and several dependent variables is investigated simultaneously. If the relationships between indepen-dent variables, between depenindepen-dent variables, and between indepenindepen-dent and depenindepen-dent variables were investigated seperately by means of correlations, the relationships between the independent variables and the relationships between the dependent variables are ignored (see, for example, Jokinen et al., 2005). To perform a multivariate (multiple) regression analysis, assuming a lesion occured in only one lobe, patients would have to be divided into four groups, depending on the lobe in which the lesion was located. This results in three dummy variables. However, patients often have lesions in several lobes and belong to more than one group. For example, a patient can have a lesion in the frontal lobe and in the temporal lobe, in the frontal lobe and the parietal lobe or can have a lesion in all four lobes. Let score 1 denote presence of a lesion in a lobe and score 0 absence of a lesion in a lobe. Then each patient has a pattern with four zeros and ones. A pattern with four zeros means no lesions, and because all patients in this study have at least one lesion, the number of different lesion patterns is 24− 1 = 15. Fifteen possible lesion
patterns require 14 dummy variables. In a multivariate (multiple) regression, the number of dummy variables, together with the other numerical variables, determine the required sample size. In the regression model, there would be 14 dummyvariables, one covariate (the volume of the lesion) and one intercept. This results in 16 regression parameters, while there are only 17 observations. Estimating this regression model would be statistically infeasible, because it requires more observations.
Another reason why a large sample size is required to investigate the pattern of post-stroke memory impairments, is that variation in the sort and severity of post-stroke memory impair-ments is needed. With only 17 observations, the variation is limited. Also, to investigate which lobe is associated with post-stroke memory impairments, the location of the lesions should vary because a stroke can occur in many different parts of the brain. So, to answer these research questions, a big patient cohort with a variety of impairments and lesions is needed.
Estimating the multivariate (multiple) regression model is infeasible, so the patients were divided into groups in a way that required fewer independent variables, requiring another statis-tical test. Patients were divided based on the lobe in which the biggest lesion was located. This lobe was determined by calculating the total volume of the lesion per lobe. This resulted in four groups. To investigate if the groups perform differently on the memory tests, a MANOVA is the appropriate statistical test. The volume of the lesion was added as a covariate so a MANCOVA test was conducted. The covariate was required because if, for example, the temporal lesion group performed significantly worse on the memory tests than the other groups, this difference could be caused by either the location of the lesion or by the fact that coincidentally there are many big lesions in the temporal lobe. Earlier research also showed that the volume of stroke
lesions significantly predict cognitive impairments (Jokinen et al., 2005). If necessary, post-hoc tests were conducted to find out which groups differ in particular. Before the MANCOVA test was conducted, an ANOVA test was done to test whether the groups of patients differed on the demographic variables age and education. Education had seven ordered levels, hence an ANOVA was performed. A chi-square test was used to see whether the groups differed on the variable sex. Preferably, the groups do not differ on the three demographic variables because the variables influence the performance on both tests. The performance on the AVLT test is influenced by age, education and sex (Van Der Elst, Van Boxtel, Van Breukelen, & Jolles, 2005). Generally, younger people, higher educated people and females outperform older people, lower educated people and males, respectively. Concerning the effect of demographic variables on the Digit Span Test performance, Irwing (2012) did not find differences between the performance of males and females. Digit Span performance worsened when age increased (Miller, Myers, Prinzi, & Mittenberg, 2009) and higher education led to higher scores on the Digit Span Test (Monaco, Costa, Caltagirone, & Carlesimo, 2013). The calculations regarding the volume of the lesions per lobe were done with ITK-SNAP (Yushkevich & Gerig, 2014). The statistical analyses were performed with SPSS (Nie, Hull, & Bent, 1968).
3
Results
Based on information from the four medical centres, a total of 256 patients were excluded. The reasons were old age (> 80 years old, 124 patients), a history of addiction (28 patients), a psychiatric disorder (12 patients), a neurological disorder (28 patients), not Dutch speaking (24 patients), and inability to participate due to other post-stroke consequences (40 patients). After exclusion, 37 patients remained. For the current study, another 20 patients were excluded for the following reasons. There was no detectable lesion on eight FLAIR images, one patient turned out to have a bleeding instead of a stroke, one patient refrained from going into the MRI scanner, the FLAIR image from two patients were missing and there were technical problems with the FLAIR images of eight patients. Therefore, 17 patients were included in the current study (4 female, 13 male, age: M = 59.65 years, SD = 9.57). On average, they were tested 62 days (SD = 59) post-stroke.
Four out of a total of 68 Z-scores were missing and therefore imputed. SPSS (Nie et al., 1968) and the statistical software programme R (Ihaka & Gentleman, 1993) do not provide pooled MANCOVA test statistics based on multiple imputations. Therefore, the SPSS single imputation method was used to impute the missing data. This method used linear regression analysis for imputation.
To find out if the patients have post-stroke memory impairments, the mean Z-scores on the memory tests were analyzed. None of the mean Z-scores were 6 −2, see Figure 2. For the MANCOVA analysis, patients were divided into groups based on the lobe in which the biggest lesion was located. Group 1, the ‘frontal group’, consisted of five patients and their biggest lesion was located in the frontal lobe. Group 2, the ‘parietal group’, consisted of eight patients and their biggest lesion was located in the parietal lobe. None of the patients were eligible for group 3, the ‘temporal group’. Four patients were eligible for group 4, the ‘occipital group’. There were no group differences regarding age (F (2, 14) = 2.271, p = 0.140), education (F (2, 14) = 0.585, p = 0.570) and gender (χ2(2) = 5.468, p = 0.065). See Table 3 for specific information per patient.
Figure 2 : A bar graph showing the mean Z-scores on the four memory tests. A deviant Z-score is Z 6 −2. LM.AVLT refers to the long-term memory score on the AVLT test, SM.AVLT refers to the short-term memory score on the AVLT test, SM.Digit.Span refers to the forward phase of the Digit Span Test and WM.Digit.Span refers to the backward phase of the Digit Span Test. Gray bands refer to 95% confidence intervals.
Table 3: Detailed Information per Patient. The Patients are Divided in Three Groups: the Frontal (F) Group, the Parietal (P) Group and the Occipital (O) Group.
Patient Age Gender Impaired lobes Lobe most impaired Volume
1 45 F F, O F 16,570 2 53 F F, O F 2,154 3 69 M F F 95 4 41 F F F 357 5 60 M F, P F 541 6 63 M F, P, T P 2,467 7 74 F P, T, O P 816 8 45 M F, P, T P 13,410 9 68 M F, P, T, O P 21,870 10 63 M F, P, T, O P 11,660 11 56 M F, P, T P 15,400 12 52 M P P 51 13 60 M F, P, T, O P 24,720 14 64 M P, T, O O 29,820 15 68 M P, T, O O 21,650 16 65 M F, P, T, O O 25,750 17 68 M T, O O 5,560
Note: Age is in years; for Gender: F=female, M=male; for Impaired lobes: T=temporal; Volume indicates the volume of the biggest lesion in mm3, which corresponds to the lobe
The four assumptions required for a MANCOVA test were investigated (Field, 2009). First, the intention was to randomly sample patients from the population to satisfy the assumption of random sampling. Second, independence of observations was satisfied because patients were tested seperately so that the possibility that they influenced each other’s performance was ruled out. Third, since there is no direct way to test multivariate normality, Field (2009) recommends to check for univariate normality for each dependent variable seperately. The Shapiro-Wilk test was used to test univariate normality for each of the four dependent variables. The Shapiro-Wilk test is the most powerful normality test (Razali, Wah, et al., 2011). This test is also suited for small sample sizes (Ahad, Yin, Othman, & Yaacob, 2011). Because the sample size of the current study was small (N = 17), the Shapiro-Wilk test was appropriate to check univariate normality. Normality within each group was checked instead of normality of the overall distribution of each dependent variable because in the MANCOVA analysis, the sample was divided into groups that were compared to each other. The Z-scores on the short-term memory score of the Digit Span Test, the Z-scores on the working memory score of the Digit Span Test, the Z-scores on the short-term memory score of the AVLT test and the Z-scores on the long-term memory score of the AVLT test were all normally distributed; see Table 4 for the results of the Shapiro-Wilk test. Although it is assumed that the dependent variables are normally distributed within all groups, Field (2009) warns that univariate normality does not imply multivariate normality. Fortunately, Pillai’s trace is robust to violations of multivariate normality. Fourth, The Box’s test was nonsignificant, indicating that the assumption of homogeneity of covariances matrices is also satisfied (F (10, 328) = 0.550, p = 0.854).
Using Pillai’s trace, there was no significant effect of the location of the lesion on the perfor-mances on the memory tests after controlling for the volume of the lesions in the lobe of interest (F (8, 22) = 1.223, p = 0.332). The covariate, which is the volume of the lesions in the lobe of interest, was not related to the memory test performances (F (4, 10) = 0.739, p = 0.586).
In an exploratory analysis, the effect of the number of lobes in which a lesion was present on the memory test performances was investigated. Investigating the number of lobes allows testing whether the combination of lesions across different lobes causes cognitive impairments. In this perspective, memory is framed in terms of a network that is distributed over different lobes instead of within one lobe. A MANCOVA analysis was conducted with the four memory tests as dependent variables, the total volume of the lesions across all lobes as a covariate and the number of lobes in which a lesion was present as the independent variable. Using Pillai’s trace, there was no significant effect of the number of lobes in which a lesion was present on the performances on the memory tests after controlling for the total volume of the lesions across all lobes (F (12, 33) = 0.572, p = 0.848). The covariate, which is the total volume of the lesions across all lobes, was not related to the memory test performances (F (4, 9) = 0.487, p = 0.745).
Table 4: Shapiro-Wilk Test Results
Shapiro-Wilk
Test Group Statistic Df P-value
SM Digit Span F 0.997 5 0.997 P 0.868 8 0.144 O 0.900 4 0.429 WM Digit Span F 0.947 5 0.713 P 0.914 8 0.386 O 0.937 4 0.635 SM AVLT F 0.836 5 0.155 P 0.920 8 0.431 O 0.921 4 0.544 LM AVLT F 0.894 5 0.378 P 0.900 8 0.288 O 0.971 4 0.846
Note: SM Digit Span: Z-scores short-term memory Digit Span, WM Digit Span: Z-scores working memory Digit Span, SM AVLT: Z-scores short-term memory AVLT, LM AVLT: Z-scores long-term memory AVLT, F: Frontal group, P: Parietal group, O: Occipital group. The four dependent variables were normally distributed within all three groups.
4
Discussion
The purpose of this study was to investigate the pattern of post-stroke memory impairments and which lobe containing lesions due to stroke is related to post-stroke memory test performances. Compared to norm data, the results show that the patients, on average, do not perform aberrantly on the memory tests. This result suggests that the current sample does not have post-stroke working memory impairments, no stroke short-term memory impairments and no post-stroke long-term memory impairments. In addition, the groups did not differ significantly on their test performances. So, a relationship between the memory test performances and the location of the lesion was not found. Neither a relationship between the number of lobes and the memory test performances was found. Memory could not be linked to the gross anatomy of the brain.
So, on average, the patients did not experience any post-stroke memory impairments. What is remarkable is that only five out of the total of 68 Z-scores were deviant (that is, 7%). An explanation why some studies shown in Table 1 found post-stroke memory impairments and why the patients of the current study did not experience impairments, may have to do with the mean age of the patients. The mean age of the patients that participated in the 16 studies (see category 1 and category 3 in Table 1) that found post-stroke memory impairments was 65.06 years (SD = 8.79). The mean age of the patients that participated in the current study was younger, 59.65 (SD = 9.572). So, the patients that participated in the current study were younger than those shown in Table 1. Generally, memory skills decline as someone becomes older (Grady & Craik, 2000; Radvansky, 2011; Van Der Elst et al., 2005). This means that
the memory impairments that the patients from the studies mentioned in Table 1 experienced, may have been caused by their age instead of the stroke. The impairments also could have been caused by the combination of suffering a stroke at an older age. Perhaps a stroke has fewer negative consequences for cognitive capabilities at a younger age than at an older age (Khedr et al., 2009; Renjen, Gauba, & Chaudhari, 2015). The young age may be an explanation for the absence of post-stroke memory impairments and the small number of deviant Z-scores found in the current study.
The fact that the current sample does not have post-stroke memory impairments may also be due to an unintentional bias in the sample. Patients were asked to participate in the study during their stay in the hospital. Patients with severe post-stroke consequences may not want to participate in research because they worry too much about their current situation and what their future would be like. It could also be that patients are shocked that they have suffered a stroke and, therefore, do not feel like participating in research. This may happen in any kind of clinical research but perhaps it may have occured more frequent in the current study because testing lasted five hours. This can be quite long for some patients so perhaps patients were more inclined not to participiate in the current study. Therefore, the current sample may have, unintentionally, existed of patients that are relatively cognitively intact and/or enthusiastic to participate in research.
A limitation of the current study is the small sample size. A small sample size causes the study to have a low statistical power, which means there is a small chance of finding a true effect. Low power can also be the reason why neither post-stroke memory impairments nor a relationship between the location of the lesion and the test performances were found. Another consequence of the small sample size is that the relationship between the test performances and the brain anatomy had to be investigated in terms of broad categories. If, for example, the parietal group experienced significantly more memory impairments than the other groups, it still would have been unclear which part of the parietal lobe is important for memory. It would have been more interesting to examine a specific part of the parietal lobe, for example, the posterior parietal cortex, because earlier research has already linked this part of the parietal lobe to post-stroke memory impairments (Schoo et al., 2011). Unfortunately, this was impossible due to the small sample size.
As suggested in the exploratory analysis, memory should perhaps be framed in terms of a network that is distributed over different lobes instead of within one lobe. In the current study, no significant effect was found of the number of lobes in which a lesion was present on the memory test performances. However, only the number of lobes was investigated in this analysis while other aspects like the amount and the kind of structural connections between certain brain areas may also be important factors underlying memory. So, the network hypothesis approach, which involves investigating the combination of different brain areas in which a lesion is present instead of only considering the lobe or the brain area in which the biggest lesion was located, could still be reasonable to investigate the anatomy of post-stroke memory impairments. An increasing number of studies found evidence that the brain and its cognitive functions are organized by means of a widely-distributed network (Beason-Held, Hohman, Venkatraman, An, & Resnick, 2016; Burianova, McIntosh, & Grady, 2010; Van den Heuvel & Pol, 2010). See Beason-Held et al. (2016) and Burianova et al. (2010) for research that has been done on brain networks in relation to memory. Investigating the network hypothesis would have been possible with a multivariate (multiple) regression analysis in which patients are eligible for several groups (for example, for both the frontal and parietal group) because then patterns of lesions are investigated and not merely the number of impaired brain areas. Unfortunately, the sample size of the current study was too small for a multivariate (multiple) regression analysis. Future research can investigate the network hypothesis with a multivariate (multiple) regression or with other statistical techniques,
for example, the correlation analysis explained in Burianova et al. (2010). Perhaps working memory is mediated by a co-operation between the frontal lobe and the parietal lobe. Long-term memory could be established by a network between the frontal lobe, the parietal lobe and the temporal lobe.
A benefit of the current study is that it is a multi-centre study. In this way, the sample size was increased in a relatively simple way by collecting data at several centres. With a bigger sample size, the statistical power of the study increases. Conducting more multi-centre studies would be a positive development among neuroscientific research since the statistical power among this kind of studies is very low. An estimation of the statistical power of 48 neuroscientific studies resulted in a disappointing 21%, which means that in 79% of the times that a study is conducted, a true effect will be missed (Button et al., 2013). So, more multi-centre studies may help to decrease the low statistical power in this field of research.
A limitation that is related to the benefit explained above, is that the current multi-centre study made use of conventional MRI images. There is an ongoing discussion on whether con-ventional MRI images can simply be pooled, which is typically done in multi-centre studies. Technical difficulties may arise when pooling the images because of differences between scanners and differences between the scanning procedures (Focke et al., 2011; Tofts & Collins, 2014). For example, research demonstrates that two T1-weighted images, acquired at two different scanners, show differences in depicting the volume of the same anatomical structures (Focke et al., 2011). Focke et al. (2011) argued that these differences could bias the statistical analysis, so MRI images from different centres cannot simply be pooled. On the other hand, there are studies that show that MRI images indeed depict differences due to different scanners, but that there are statistical methods to correct for these differences. For example, including ‘site’ as an independent variable in the statistical analysis reduces between-site variance (Pardoe, Pell, Abbott, Berg, & Jackson, 2008).
The ongoing discussion applies to the current study because patients were divided into groups according to the volume of the lesion per lobe. This means that the volumes of brain areas were important measures. Since Focke et al. (2011) showed that there is variation between MRI images depicting the volume of the same anatomical structure, there may also be structural differences between the FLAIR images in depicting the volume of the lesions. Quantitative MRI (qMRI) is a relatively new MRI technique that allows images to be pooled without any problems. An advantage of the current study is that it is a multi-centre study, but it seems it would have been better if the current study used qMRI images instead of FLAIR images.
In short, the reason why qMRI can be pooled is that qMRI measures the signal intensity of a voxel at several points while conventional MRI records only one measurement point. Therefore, when using conventional MRI, physical properties of voxels are estimated and this results in signal intensity values of voxels that are relative to each other. Relative signal intensity values do not represent the exact signal intensity but represent which voxel has a high signal intensity compared to the other voxels and which voxel has a low signal intensity compared to the other voxels. This means that signal intensities of voxels are analyzed with respect to each other. In contrast, qMRI records several measurement points and therefore, physical properties of the voxels can be calculated precisely. So with qMRI, physical properties are measured in a direct manner resulting in an absolute signal intensity value of a voxel. Absolute signal intensity values represent the exact value, which is not measured with respect to other voxels. Absolute signal intensity values can be directly compared with each other and therefore, qMRI images can be pooled. For a more elaborate explanation of the difference between conventional MRI and qMRI and for more benefits of using qMRI instead of conventional MRI, see Appendix 2.
To conclude, the aim of the current study was to clarify the performance of stroke patients on working memory tests, short-term memory tests and long-term memory tests, and to investigate
which lobe is related to these test performances. The patients did not experience any post-stroke memory impairments. No relationship between the location of the lesion and the test performances was found, and no relationship between the number of lobes in which a lesion was present and the test performances was found. Further research must pay attention to the network hypothesis approach and to the possibility to use qMRI instead of conventional MRI.
References
Ahad, N. A., Yin, T. S., Othman, A. R., & Yaacob, C. R. (2011). Sensitivity of normality tests to non-normal data. Sains Malaysiana, 40 (6), 637–641.
Atkinson, R. C., & Shiffrin, R. M. (1968). Human memory: A proposed system and its control processes. Psychology of Learning and Motivation, 2 , 89–195.
Babulal, G. M., Huskey, T. N., Roe, C. M., Goette, S. A., & Connor, L. T. (2015). Cognitive impairments and mood disruptions negatively impact instrumental activities of daily living performance in the first three months after a first stroke. Topics in Stroke Rehabilitation, 22 (2), 144–151.
Baddeley, A. (2000). The episodic buffer: A new component of working memory? Trends in Cognitive Sciences, 4 (11), 417–423.
Baldo, J. V., & Dronkers, N. F. (2006). The role of inferior parietal and inferior frontal cortex in working memory. Neuropsychology, 20 (5), 529–538.
Ballard, C., Stephens, S., Kenny, R., Kalaria, R., Tovee, M., & OBrien, J. (2003). Profile of neuropsychological deficits in older stroke survivors without dementia. Dementia and Geriatric Cognitive Disorders, 16 (1), 52–56.
Barbey, A. K., Koenigs, M., & Grafman, J. (2013). Dorsolateral prefrontal contributions to human working memory. Cortex , 49 (5), 1195–1205.
Beason-Held, L. L., Hohman, T. J., Venkatraman, V., An, Y., & Resnick, S. M. (2016). Brain network changes and memory decline in aging. Brain Imaging and Behavior , 1–15. Ben-Zvi, S., Soroker, N., & Levy, D. A. (2015). Parietal lesion effects on cued recall following
pair associate learning. Neuropsychologia, 73 , 176–194.
Berryhill, M. E., & Olson, I. R. (2008). Is the posterior parietal lobe involved in working memory retrieval? evidence from patients with bilateral parietal lobe damage. Neuropsychologia, 46 (7), 1775–1786.
Berryhill, M. E., Picasso, L., Arnold, R., Drowos, D., & Olson, I. R. (2010). Similarities and differences between parietal and frontal patients in autobiographical and constructed experience tasks. Neuropsychologia, 48 (5), 1385–1393.
Blystad, I., Warntjes, J. B. M., Smedby, O., Landtblom, A.-M., Lundberg, P., & Larsson, E.-M. (2012). Synthetic MRI of the brain in a clinical setting. Acta Radiologica, 53 (10), 1158–1163.
Bouma, A., Mulder, J., Lindeboom, J., & Schmand, B. (2012). Handboek neuropsychologische diagnostiek.-2e herz. dr. (Handbook of neuropsychological diagnostics-2nd revised version). Amsterdam: Pearson.
Burianova, H., McIntosh, A. R., & Grady, C. L. (2010). A common functional brain network for autobiographical, episodic, and semantic memory retrieval. Neuroimage, 49 (1), 865–874. Button, K. S., Ioannidis, J. P., Mokrysz, C., Nosek, B. A., Flint, J., Robinson, E. S., &
Mu-naf`o, M. R. (2013). Power failure: Why small sample size undermines the reliability of neuroscience. Nature Reviews Neuroscience, 14 (5), 365–376.
Cabeza, R., Ciaramelli, E., Olson, I. R., & Moscovitch, M. (2008). The parietal cortex and episodic memory: An attentional account. Nature Reviews Neuroscience, 9 (8), 613–625.
Centraal Bureau voor de Statistiek. (2016). Bevolkingspiramide. Retrieved from https://www.cbs.nl/nl-nl/visualisaties/bevolkingspiramide
Chen, Y., Wang, A., Tang, J., Wei, D., Li, P., Chen, K., . . . Zhang, Z. (2015). Association of white matter integrity and cognitive functions in patients with subcortical silent lacunar infarcts. Stroke, 46 (4), 1123–1126.
Commissie Testaangelegenheden Nederland. (2000). Nieuwe vijf-tien woordentest A en B, 15 wt-A en 15 wt-B. Retrieved from http://www.cotandocumentatie.nl/test details.php?id=352
Commissie Testaangelegenheden Nederland. (2012). Wechsler adult intelligence scale, fourth edition, nederlandstalige bewerking, WAIS-IV NL. Retrieved from http://www.cotandocumentatie.nl/test details.php?id=810
Cumming, T. B., Marshall, R. S., & Lazar, R. M. (2013). Stroke, cognitive deficits, and rehabilitation: Still an incomplete picture. International Journal of Stroke, 8 (1), 38–45. De Haan, E. H. F., Nys, G. M. S., & Van Zandvoort, M. J. E. (2006). Cognitive function
following stroke and vascular cognitive impairment. Current Opinion in Neurology , 19 (6), 559–564.
Deoni, S. C., Williams, S. C., Jezzard, P., Suckling, J., Murphy, D. G., & Jones, D. K. (2008). Standardized structural magnetic resonance imaging in multicentre studies using quanti-tative T1 and T2 imaging at 1.5T. Neuroimage, 40 (2), 662–671.
Engstr¨om, M., Warntjes, J. B., Tisell, A., Landtblom, A.-M., & Lundberg, P. (2014). Multi-parametric representation of voxel-based quantitative magnetic resonance imaging. PloS ONE , 9 (11), 1-10.
Field, A. (2009). Discovering statistics using SPSS - third edition. Thousand Oaks, CA: Sage publications.
Focke, N. K., Helms, G., Kaspar, S., Diederich, C., T´oth, V., Dechent, P., . . . Paulus, W. (2011). Multi-site voxel-based morphometry - not quite there yet. Neuroimage, 56 (3), 1164–1170. Grady, C. L., & Craik, F. I. (2000). Changes in memory processing with age. Current Opinion
in Neurobiology , 10 (2), 224–231.
Hartstichting. (n.d.). Herken de signalen van een beroerte. Retrieved from https://www.volksgezondheidenzorg.info/onderwerp/beroerte
Hattingen, E., Jurcoane, A., Nelles, M., M¨uller, A., N¨oth, U., M¨adler, B., . . . Schild, H. H. (2015). Quantitative MR imaging of brain tissue and brain pathologies. Clinical Neuroradiology , 25 (2), 219–224.
Huettel, S. A., Song, A. W., & McCarthy, G. (2004). Functional magnetic resonance imaging -Second edition. Sunderland, MA: Sinauer Associates.
Huijben-Schoenmakers, M., Rademaker, A., & Scherder, E. (2016). Cognition in relation to independency in older, comorbid stroke patients in a stroke unit. International Journal of Geriatric Psychiatry, 1–8.
Ihaka, R., & Gentleman, R. (1993). R - programming. Retrieved from https://www.r-project.org/
Irwing, P. (2012). Sex differences in g: An analysis of the US standardization sample of the WAIS-III. Personality and Individual Differences, 53 (2), 126–131.
Jokinen, H., Kalska, H., M¨antyl¨a, R., Ylikoski, R., Hietanen, M., Pohjasvaara, T., . . . Erk-injuntti, T. (2005). White matter hyperintensities as a predictor of neuropsychological deficits post-stroke. Journal of Neurology, Neurosurgery & Psychiatry , 76 (9), 1229–1233. Jokinen, H., Melkas, S., Ylikoski, R., Pohjasvaara, T., Kaste, M., Erkinjuntti, T., & Hietanen, M.
(2015). Post-stroke cognitive impairment is common even after successful clinical recovery. European Journal of Neurology, 22 (9), 1288–1294.
Khedr, E. M., Hamed, S. A., El-Shereef, H. K., Shawky, O. A., Mohamed, K. A., Awad, E. M., . . . Eltahtawy, M. A. (2009). Cognitive impairment after cerebrovascular stroke: Relationship to vascular risk factors. Neuropsychiatric Disease and Treatment , 5 (2), 103–116.
Koenigs, M., Barbey, A. K., Postle, B. R., & Grafman, J. (2009). Superior parietal cortex is crit-ical for the manipulation of information in working memory. The Journal of Neuroscience, 29 (47), 14980–14986.
La Joie, R., Landeau, B., Perrotin, A., Bejanin, A., Egret, S., P´elerin, A., . . . others (2014). Intrinsic connectivity identifies the hippocampus as a main crossroad between alzheimers and semantic dementia-targeted networks. Neuron, 81 (6), 1417–1428.
MacPherson, S. E., Turner, M. S., Bozzali, M., Cipolotti, L., & Shallice, T. (2016). The doors and people test: The effect of frontal lobe lesions on recall and recognition memory performance. Neuropsychology, 30 (3), 332–337.
Madureira, S., Guerreiro, M., & Ferro, J. M. (2001). Dementia and cognitive impairment three months after stroke. European Journal of Neurology, 8 (6), 621–627.
Makkat, S., Vandevenne, J. E., Verswijvel, G., Ijsewijn, T., Grieten, M., Palmers, Y., . . . Parizel, P. M. (2002). Signs of acute stroke seen on fluid-attenuated inversion recovery MR imaging. American Journal of Roentgenology, 179 (1), 237–243.
Mandzia, J. L., Smith, E. E., Horton, M., Hanly, P., Barber, P. A., Godzwon, C., . . . Coutts, S. B. (2016). Imaging and baseline predictors of cognitive performance in minor ischemic stroke and patients with transient ischemic attack at 90 days. Stroke, 47 (3), 726–731. Margaret Cheng, H.-L., Stikov, N., Ghugre, N. R., & Wright, G. A. (2012). Practical medical
applications of quantitative MR relaxometry. Journal of Magnetic Resonance Imaging , 36 (4), 805–824.
Miller, L. J., Myers, A., Prinzi, L., & Mittenberg, W. (2009). Changes in intellectual functioning associated with normal aging. Archives of Clinical Neuropsychology, 24 (7), 681–688. Miyashita, Y. (2004). Cognitive memory: Cellular and network machineries and their top-down
control. Science, 306 (5695), 435–440.
Monaco, M., Costa, A., Caltagirone, C., & Carlesimo, G. A. (2013). Forward and backward span for verbal and visuo-spatial data: Standardization and normative data from an italian adult population. Neurological Sciences, 34 (5), 749–754.
Ng, Y. S., Stein, J., Ning, M., & Black-Schaffer, R. M. (2007). Comparison of clinical character-istics and functional outcomes of ischemic stroke in different vascular territories. Stroke, 38 (8), 2309–2314.
Nie, N. H., Hull, C. H., & Bent, D. H. (1968). IBM SPSS statistics 24. Retrieved from http://www.ibm.com
Nys, G. M. S., Van Zandvoort, M. J. E., De Kort, P. L. M., Jansen, B. P. W., Van der Worp, H. B., Kappelle, L. J., & De Haan, E. H. F. (2005). Domain-specific cognitive recov-ery after first-ever stroke: A follow-up study of 111 cases. Journal of the International Neuropsychological Society, 11 (07), 795–806.
Nys, G. M. S., Van Zandvoort, M. J. E., Van der Worp, H. B., Kappelle, L. J., & De Haan, E. H. F. (2006). Neuropsychological and neuroanatomical correlates of perseverative responses in subacute stroke. Brain, 129 (8), 2148–2157.
Pardoe, H., Pell, G. S., Abbott, D. F., Berg, A. T., & Jackson, G. D. (2008). Multi-site voxel-based morphometry: Methods and a feasibility demonstration with childhood absence epilepsy. Neuroimage, 42 (2), 611–616.
Park, J. H., Kim, B. J., Bae, H.-J., Lee, J., Lee, J., Han, M.-K., . . . others (2013). Impact of post-stroke cognitive impairment with no dementia on health-related quality of life. Journal of stroke, 15 (1), 49–56.
Regional coherence alterations revealed by resting-state fmri in post-stroke patients with cognitive dysfunction. PloS ONE , 11 (7), 1–13.
Race, E., Keane, M. M., & Verfaellie, M. (2011). Medial temporal lobe damage causes deficits in episodic memory and episodic future thinking not attributable to deficits in narrative construction. The Journal of Neuroscience, 31 (28), 10262–10269.
Radvansky, G. A. (2011). Human memory - Second edition. Boston, MA: Pearson education. Rasquin, S. M. C., Lodder, J., Ponds, R. W. H. M., Winkens, I., Jolles, J., & Verhey, F. R. J.
(2004). Cognitive functioning after stroke: A one-year follow-up study. Dementia and Geriatric Cognitive Disorders, 18 (2), 138–144.
Razali, N. M., Wah, Y. B., et al. (2011). Power comparisons of shapiro-wilk, kolmogorov-smirnov, lilliefors and anderson-darling tests. Journal of Statistical Modeling and Analytics, 2 (1), 21–33.
Renjen, P. N., Gauba, C., & Chaudhari, D. (2015). Cognitive impairment after stroke. Cureus, 7 (9), 1–9.
Riba-Llena, I., Koek, M., Verhaaren, B. F., Vrooman, H. A., Lugt, A., Hofman, A., . . . Vernooij, M. W. (2015). Small cortical infarcts: Prevalence, determinants, and cognitive correlates in the general population. International Journal of Stroke, 10 (A100), 18–24.
Rijksinstituut voor Volksgezondheid en Milieu. (2016). Beroerte. Retrieved from https://www.volksgezondheidenzorg.info/onderwerp/beroerte
Roussel, M., Martinaud, O., H´enon, H., Vercelletto, M., Bindschadler, C., Joseph, P.-A., . . . others (2016). The behavioral and cognitive executive disorders of stroke: The GREFEX study. PloS ONE , 11 (1), 1–13.
Rugg, M. D., & Vilberg, K. L. (2013). Brain networks underlying episodic memory retrieval. Current Opinion in Neurobiology, 23 (2), 255–260.
Rutten-Jacobs, L. C., Arntz, R. M., Maaijwee, N. A., Schoonderwaldt, H. C., Dorresteijn, L. D., van Dijk, E. J., & de Leeuw, F.-E. (2013). Long-term mortality after stroke among adults aged 18 to 50 years. Jama, 309 (11), 1136–1144.
Saan, R. J., & Deelman, B. G. (1986). De 15-woordentests A en B. een voorlopige handleiding. Groningen, afdeling Neuropsychologie UMCG.
Sachdev, P. S., Brodaty, H., Valenzuela, M. J., Lorentz, L., Looi, J. C. L., Wen, W., & Zagami, A. S. (2004). The neuropsychological profile of vascular cognitive impairment in stroke and TIA patients. Neurology, 62 (6), 912–919.
Schaapsmeerders, P., Maaijwee, N. A., van Dijk, E. J., Rutten-Jacobs, L. C., Arntz, R. M., Schoonderwaldt, H. C., . . . de Leeuw, F.-E. (2013). Long-term cognitive impairment after first-ever ischemic stroke in young adults. Stroke, 44 (6), 1621–1628.
Schoo, L. A., van Zandvoort, M. J. E., Biessels, G. J., Kappelle, L. J., Postma, A., & De Haan, E. H. F. (2011). The posterior parietal paradox: Why do functional magnetic resonance imaging and lesion studies on episodic memory produce conflicting results? Journal of Neuropsychology, 5 (1), 15–38.
Shannon, B. J., & Buckner, R. L. (2004). Functional-anatomic correlates of memory retrieval that suggest nontraditional processing roles for multiple distinct regions within posterior parietal cortex. The Journal of Neuroscience, 24 (45), 10084–10092.
Simons, J. S., Peers, P. V., Mazuz, Y. S., Berryhill, M. E., & Olson, I. R. (2009). Dissocia-tion between memory accuracy and memory confidence following bilateral parietal lesions. Cerebral Cortex , 20 , 479–485.
Squire, L. R., & Zola, S. M. (1998). Episodic memory, semantic memory, and amnesia. Hip-pocampus, 8 (3), 205–211.
Srikanth, V. K., Thrift, A. G., Saling, M. M., Anderson, J. F., Dewey, H. M., Macdonell, R. A., & Donnan, G. A. (2003). Increased risk of cognitive impairment 3 months after mild
to moderate first-ever stroke a community-based prospective study of nonaphasic english-speaking survivors. Stroke, 34 (5), 1136–1143.
Tatemichi, T. K., Desmond, D. W., Stern, Y., Paik, M., Sano, M., & Bagiella, E. (1994). Cognitive impairment after stroke: Frequency, patterns, and relationship to functional abilities. Journal of Neurology, Neurosurgery Psychiatry, 57 (2), 202–207.
Thong, J. Y. J., Hilal, S., Wang, Y., Soon, H. W., Dong, Y., Collinson, S. L., . . . others (2013). Association of silent lacunar infarct with brain atrophy and cognitive impairment. Journal of Neurology, Neurosurgery & Psychiatry, 1–7.
Tofts, P. (2005). Quantitative MRI of the brain: Measuring changes caused by disease. John Wiley & Sons, Chichester.
Tofts, P., & Collins, D. (2014). Multicentre imaging measurements for oncology and in the brain. The British Journal of Radiology, 213-226.
Tulving, E. (1972). Episodic and semantic memory. Organization of Memory. London: Academic, 381 (4), 382–404.
Van den Heuvel, M. P., & Pol, H. E. H. (2010). Exploring the brain network: A review on resting-state fMRI functional connectivity. European Neuropsychopharmacology, 20 (8), 519–534.
Van Der Elst, W. I. M., Van Boxtel, M. P. J., Van Breukelen, G. J. P., & Jolles, J. (2005). Rey’s verbal learning test: Normative data for 1855 healthy participants aged 24–81 years and the influence of age, sex, education, and mode of presentation. Journal of the International Neuropsychological Society, 11 (03), 290–302.
Van Rooij, F. G., Schaapsmeerders, P., Maaijwee, N. A., van Duijnhoven, D. A., de Leeuw, F.-E., Kessels, R. P., & van Dijk, E. J. (2014). Persistent cognitive impairment after transient ischemic attack. Stroke, 45 (8), 2270–2274.
Viswanathan, A., Macklin, E. A., Betensky, R., Hyman, B., Smith, E. E., & Blacker, D. (2015). The influence of vascular risk factors and stroke on cognition in late life: Analysis of the nacc cohort. Alzheimer Disease & Associated Disorders, 29 (4), 287–293.
Wechsler, D. (2012). Wechsler adult intelligence scale-fourth edition (WAIS-IV), nederland-stalige bewerking, afname- en scoringhandleiding. Pearson Assessment and Information: Amsterdam.
Yushkevich, P., & Gerig, G. (2014, october). ITK-SNAP version 3.2.0. Retrieved from http://www.itksnap.org/pmwiki/pmwiki.php
5
Appendix 1
Table 1: An Overview of the Articles Specifying the Impaired and Unimpaired Cognitive Do-mains. The Articles are Divided into Three Categories: The First Category Specified Memory as Impaired, the Second Category Specified Memory as Unimpaired or as Least Impaired and the Third Category Specified Memory as Impaired as well as Unimpaired or as Least Impaired.
Study Impaired domains Unimpaired or least impaired domains
Babulal, Huskey, Roe,
Goette, and Connor
(2015)
EF, memory, orientation
-Chen et al. (2015) EF, attention, memory, language Visuospatial ability
Jokinen et al. (2015) EF, memory, visuoconstructive skills i.a. Language, abstract reasoning Madureira, Guerreiro,
and Ferro (2001)
i.a. IPS, memory, orientation, language, learning
i.a. Calculation, visuoconstructive skills
Nys et al. (2006) EF, memory, motor perseveration
-Park et al. (2013) EF, memory, visuoconstructive skills, lan-guage
-Roussel et al. (2016) EF, memory Visuoconstructive skills, language
Schaapsmeerders et al. (2013)
EF, IPS, attention, memory Visuoconstructive skills
Tatemichi et al. (1994) i.a. Attention, memory, orientation Visuospatial ability, abstract reasoning Thong et al. (2013) EF, IPS, attention, memory, language Visuoconstructive skills
Mandzia et al. (2016) - EF, IPS, memory, visuoconstructive skills,
language
Nys et al. (2005) i.a. EF, Visual perception/construction Memory, language
Rasquin et al. (2004) IPS, calculation i.a. Memory, orientation, praxis
Srikanth et al. (2003) EF, attention, language, spatial ability Memory, orientation
Viswanathan et al.
(2015)
EF, IPS, language IPS, memory
Ballard et al. (2003) EF, IPS, memory Memory
Huijben-Schoenmakers,
Rademaker, and
Scherder (2016)
EF, memory Memory
Peng et al. (2016) EF, IPS, attention, memory Memory
Riba-Llena et al. (2015) EF, IPS, memory Memory
Sachdev et al. (2004) i.a. EF, IPS, memory, visuocontruction Memory, language, praxis-gnosis function
Van Rooij et al. (2014) EF, IPS, attention, memory Memory
Table 2: An Overview of the Memory Components that were Impaired and Memory Components that were Unimpaired or Least Impaired. The Articles Were Divided into Six Categories (see Introduction).
Study Specification on memory problems
Mandzia et al. (2016) vrmemory (IR&DR): UI, vsmemory(IR&DR): UI
Peng et al. (2016) vrmemory (IR): UI, vrmemory (DR): I, vsmemory (IR&DR):
I, WM: I
Sachdev et al. (2004) vrmemory (IR&DR): UI, vsmemory (IR&DR): I, WM: UI Srikanth et al. (2003) vrmemory (IR&DR): UI, vsmemory (IR&DR): UI, WM: UI
Tatemichi et al. (1994) vrmemory: I, vsmemory: I
Nys et al. (2005) vrmemory: UI, vsmemory: UI
Thong et al. (2013) memory: I
Jokinen et al. (2015) memory: I
Riba-Llena et al. (2015) vrmemory (IR): UI, vrmemory (DR): I
Van Rooij et al. (2014) vrmemory (IR&DR): UI, WM: I
Viswanathan et al. (2015) vrmemory (IR&DR): UI, WM: UI
Babulal et al. (2015) memory: I
Ballard et al. (2003) episodic memory: UI, WM: I
Chen et al. (2015) memory: I
Rasquin et al. (2004) episodic memory: UI
Roussel et al. (2016) episodic memory: I
Huijben-Schoenmakers et al. (2016) vrmemory (IR&DR): UI, vsmemory: I
Madureira et al. (2001) vrmemory: I,
Nys et al. (2006) spatial WM*: I, verbal WM: I
Park et al. (2013) vrmemory (DR): I
Schaapsmeerders et al. (2013) IR: I, DR: I, WM: I
Note: I: impaired, UI: unimpaired or it was one of the least impaired cognitive domains, vrmemory: verbal memory, vsmemory: visual memory, IR: immediate recall, DR: delayed recall, WM: working memory. *: only patients with attentional deficits or with neglect performed worse on spatial WM tasks.
6
Appendix 2
This appendix provides an explanation on the difference between qMRI and conventional MRI. This will be done in light of the T1 relaxation time, a physical property of voxels.
T1 relaxation time
The T1 relaxation time is the time that is takes for the aggregated direction of the protons within a voxel (called the vector) to return to 63% of the direction of the main magnetic field. When someone is put into a scanner, the vector’s direction aligns along the main magnetic field (see Figure 3A). When a magnetic pulse is applied, it will change its pointing direction, for example with 90 degrees (see Figure 3B). When the pulse is turned off, the vector’s pointing di-rection recovers to align along the magnetic field again (see Figure 3C and Figure 3D). The time that this return takes is depicted by means of an exponential curve, called a T1-recovery curve. Based on this recovery curve, the T1 relaxation time of the voxel is calculated. Every kind of tis-sue, and thus every voxel, has an identifiable relaxation time (Huettel, Song, & McCarthy, 2004).
(a) (b) (c) (d)
Figure 3 : A: vector (red arrow) is aligned along magnetic field (black vertical arrow), B: the vector turns 90 degrees to the right after a pulse is turned on, C: the vector returns back to align along the magnetic field again, D: the vector is aligned along the magnetic field again.
Conventional MRI
When the vector is returning (Figure 3C), the signal intensity of a voxel is measured at only one time point. Without a second measurement point to indicate the slope of the curve, it is impossible to estimate the recovery curve. Therefore, the exact relaxation time of the voxels cannot be calculated. This is not necessarily problematic because the exact absolute relaxation times are not required to create MRI images.
To create MRI images, only relative relaxation times are required. These are estimated based on the relative signal intensities of voxels which are measured at the chosen measurement point. Relative relaxation times show which voxel has a high signal intensity (and thus has a shorter T1 relaxation time) compared to other voxels and which voxel has a low signal intensity (and thus a longer T1 relaxation time) compared to other voxels. This means that signal intensities and relaxation times of voxels are analyzed with respect to each other.
So, conventional MRI does not measure the physical properties of voxels in a direct manner but it measures the properties relative to one another. A limitation of relative measures is that they are influenced by surrounding factors like characteristics of the specific scanner, such as the amount of Tesla, the kind of radiofrequency coil, magnetic field inhomogeneities but also by the temperature of the room. So, the measured MRI signal intensity is not only affected by physi-cal properties of the voxel, but also by unintentionally measured factors from the surroundings (Deoni et al., 2008). These factors differ across time and across sites so it is difficult to compare MRI images over time and across sites (Blystad et al., 2012; Margaret Cheng, Stikov, Ghugre,
& Wright, 2012). These difficulties cause technical problems when pooling the MRI images. Quantitative MRI
The crucial difference between MRI and qMRI is that qMRI measures the signal intensity of a voxel at several points. Subsequently, the T1 recovery curve of the voxel can be accurately calculated. This can be done for two images having a different flip angle but with the same measurement points. Based on the ratio between the two recovery curves, the exact T1 relax-ation time of the voxel can be calculated. Doing this for every voxel, will create a qMRI image specifying the absolute values of the physical properties of all voxels. The absolute values, which represent the physical properties of the voxels, are not influenced by factors like the temperature of the room or magnetic field inhomogeneities.
Measuring the physical properties of the voxels in a direct manner, has several advantages. First, the statistical power of a study can be increased because by pooling qMRI images from different sites, the sample size of the study can be increased in a simple way (Hattingen et al., 2015; Tofts, 2005). Second, absolute values make it possible to compare qMRI images over time. In this way, changes in lesioned tissue can be depicted accurately (Hattingen et al., 2015). Third, the quantitative measurement of physical properties of voxels enable a more precise and objective display of the kinds of tissue than a conventional MRI image. qMRI can therefore be consid-ered a more reliable method than conventional MRI (Engstr¨om, Warntjes, Tisell, Landtblom, & Lundberg, 2014; Margaret Cheng et al., 2012).
Conclusion
To summarize, qMRI records more measurement points and it is based on more images (for ex-ample, two images with a different flip angle) than conventional MRI. Therefore, a qMRI image can be created that depicts absolute values of the physical properties of all voxels. Conventional MRI measures the physical properties of voxels not in a direct manner but relative to each other. As a consequence, qMRI has some advantages over conventional MRI.
