INTRODUCTION
Mild cognitive impairment (MCI) is believed to be a prodrome that eventually leads to dementia. Verbal and visual episodic memory deficits in MCI have been extensively researched. For example, a meta-analytic study conducted by Belleville, Fouquet, Hudon, Zomahoun and Croteau (Reference Belleville, Fouquet, Hudon, Zomahoun and Croteau2017) found that verbal memory, along with language tests, yielded high predictive accuracy of conversion from MCI to Alzheimer’s disease (AD) and were among the highest sensitivity and specificity values. Other studies suggest that visual episodic memory performance may be particularly sensitive to emergent illness (De Anna et al., Reference De Anna, Felician, Barbeau, Mancini, Didic and Ceccaldi2014; Didic et al., Reference Didic, Felician, Barbeau, Mancini, Latger-Florence, Tramoni and Ceccaldi2013; Okonkwo et al., Reference Okonkwo, Oh, Koscik, Jonaitis, Cleary, Dowling, Bendlin, LaRue, Hermann, Barnhart and Murali2014). However, there has been less research comparing verbal versus visual working memory (WM) performance in MCI. Episodic memory is also well known to be associated with hippocampal integrity (Burgess, Maguire, & O’Leefe, Reference Burgess, Maguire and O’Keefe2002; Eichenbaum, Cohen, & Squire, Reference Eichenbaum, Cohen and Squire2001). However, recent research has demonstrated hippocampal involvement in WM test performance (Leszczyński et al., Reference Leszczyński, Fell and Axmacher2015; Liang et al., Reference Liang, Pertzov, Nicholas, Henley, Crutch, Woodward, Leung, Fox and Husain2016). For example, Spellman et al. (Reference Spellman, Rigotti, Ahmari, Fusi, Gogos and Gordon2015) found evidence suggesting that hippocampal/prefrontal afferent pathways are particularly important during the cue-encoding phase of a spatial WM task. Chein, Moore and Conway (Reference Chein, Moore and Conway2011) studied a group of college-aged patients with WM tasks using fMRI and also found that successful WM test performance was highly associated with hippocampal activity.
A reason that visual as compared to verbal WM tests may be more sensitive to emergent illness may be due, in part, to the number and diversity of neurocognitive skills necessary to complete these tests. However, evidence for this idea has been mixed. For example, Kane et al. (Reference Kane, Hambrick, Tuholski, Wilhelm, Payne and Engle2004) assessed a large group of undergraduate volunteers using a protocol of tests measuring both verbal and visual WM and short-term memory. Confirmatory factor analyses and structural equation modeling tended to suggest that verbal and visual WM test performance could be explained on the basis of a single underlying cognitive construct. Park et al. (Reference Park, Lautenschlager, Hedden, Davidson, Smith and Smith2002) were interested in assessing relationships between visual WM and visuospatial operations across the life span. These researchers found that both visual WM and visuospatial operations were interrelated, again, suggesting convergence regarding the neurocognitive constructs underlying these operations.
By contrast, Liang et al. (Reference Liang, Pertzov, Nicholas, Henley, Crutch, Woodward, Leung, Fox and Husain2016) recently examined visual WM in asymptomatic and symptomatic patients diagnosed with familial AD (FAD). These researchers found that asymptomatic familial AD mutation carrier patients demonstrated intact memory for object identity and location; however, they frequently misplaced probed items to the location of one of the other items held in memory to suggest a problem with relational binding. Follow-up analyses showed a significant negative correlation between hippocampal volume and misbinded errors, but not for object identity or localization individually, suggesting a strong relationship between object-location binding and hippocampal function in mutation carriers. This research suggests that hippocampal functions may be an important anatomic region for visual WM tests, and that there are a number of discrete neurocognitive abilities that likely underlie performance on visual WM tests.
In the current research, memory clinic patients were administered well-known, co-normed clinical tests that assess verbal and visual WM. Hippocampal occupancy (HO), a ratio of hippocampal volume to adjacent lateral ventricular dilation was obtained from clinical T1-weighted images. On the basis of the research reviewed above, we hypothesized that patients will display greater difficulty on visual, compared to verbal, WM tests. Additionally, we hypothesize that greater hippocampal volume will be associated with better WM test performance.
METHODS
Participants
Patients (n = 103) were evaluated at the Rowan University, New Jersey Institute for Successful Aging, Memory Assessment Program (MAP). All patients sought an evaluation through the MAP program because of concerns regarding declining cognitive abilities by either the patient or their family. Rowan University institutional review board approved this investigation and the current research complied with the Declaration of Helsinki. The work-up for neurocognitive disorders included comprehensive neuropsychological assessment; evaluations provided by a social worker and a board-certified geriatric psychiatrist; a brain MRI/CT scan, and serum blood tests. An interdisciplinary team conference determined clinical diagnosis. All patients presented with subjective cognitive complaints; preservation of general functional abilities; and absence of dementia. Exclusion criteria included head injury, substance abuse, major/medical psychiatric disorders (e.g. major depression, cancer, epilepsy), B12, folate, or thyroid deficiency. A knowledgeable family member provided information regarding functional status. Demographic and clinical characteristics include age, education, Mini-Mental State Examination (MMSE) Performance (Folstein, Folstein, & McHugh, Reference Folstein, Folstein and McHugh1975), depression [Geriatric Depression Scale (GDS), Sheikh & Yesavage, Reference Sheikh and Yesavage1986], Wide Range Achievement Test-IV Reading subtest performance, and instrumental activities of daily living (IADL; Lawton & Brody, Reference Lawton and Brody1969).
Neuropsychological Assessment
The neuropsychological protocol used to classify patients into their respective groups is identical to procedures described by Emrani et al. (Reference Emrani, Libon, Lamar, Price, Jefferson, Gifford, Hohman, Nation, Delano-Wood, Jak and Bangen2018). Three domains of neuropsychological functioning were assessed: executive control, naming/lexical access, and verbal episodic memory. Nine neuropsychological parameters, three from each neurocognitive domain were obtained. Edmonds, Delano-Wood, Galasko, Salmon and Bondi (Reference Edmonds, Delano-Wood, Galasko, Salmon and Bondi2015) and Jak et al. (Reference Jak, Bondi, Delano-Wood, Wierenga, Corey-Bloom, Salmon and Delis2009) criteria were used to classify patients as presenting with non-MCI, subtle MCI, or MCI subtypes as described below. Tests were expressed as Z-scores, derived from normative data and are listed below.
Executive control
The Boston Revision of the Wechsler Memory Scale-Mental Control subtest-Accuracy Index (see Lamar, Price, Davis, Kaplan, & Libon, Reference Lamar, Price, Davis, Kaplan and Libon2002 for full details); letter verbal fluency (‘FAS’; Spreen & Struss, Reference Spreen and Strauss1990); and Trail Making Test-Part B (Reitan & Wolfson, Reference Reitan and Wolfson1985).
Lexical access/language
60-item Boston Naming Test (Kaplan, Goodglass, & Weintraub, Reference Kaplan, Goodglass and Weintraub1983); semantic (‘animals’) fluency (Carew, Lamar, Cloud, Grossman, & Libon, Reference Carew, Lamar, Cloud, Grossman and Libon1997); and Wechsler Adult Intelligence Scale-III Similarities subtest (Wechsler, Reference Wechsler1997).
Memory and learning
The 9-word California Verbal Learning Test-short form (Delis, Kramer, Kaplan, & Thompkins, Reference Delis, Kramer, Kaplan and Thompkins1987; Delis, Kramer, Kaplan, & Ober, Reference Delis, Kramer, Kaplan and Ober2000) including total immediate free recall, delayed free recall, and the delayed recognition discriminability index (Table 1).
Table 1. Neuropsychological test performance (Z-scores; means and standard deviations)
WMS-Mental Control = Boston Revision of the Wechsler Memory Scale Memory Control subtest; CVLT-Short Form = California Verbal Learning Test-Short Form; non-MCI = non-mild cognitive impairment; SCI = subtle cognitive impairment; aMCI = amnestic mild cognitive impairment; mixed/dys = mixed/dysexecutive mild cognitive impairment.
Determination of MCI Subtypes
Single and Mixed MCI: Jak et al. (Reference Jak, Bondi, Delano-Wood, Wierenga, Corey-Bloom, Salmon and Delis2009) comprehensive neuropsychological criteria were used to determine MCI subtype. Single domain MCI syndromes were diagnosed when scores fell >1.0 standard deviation (sd) below normative expectations (i.e. below the 16th percentile) on any two of the three measures within a single cognitive domain. For example, patients who performed below the 16th percentile on two measures solely within the memory domain would be classified as having an amnestic cognitive impairment. Mixed MCI syndromes were diagnosed when scores fell >1.0 sd below normative expectations on any two of the three measures across two or more cognitive domains. As such, patients whose scores fell >1.0 sd below the mean on two measures within the memory domain and two measures within the executive control domain would be classified as having a mixed MCI. Twenty-three patients met criteria for single domain amnestic MCI (aMCI), 8 patients for dysexecutive MCI (dys MCI), and 23 for mixed MCI. Because of the small number of dysMCI patients, and because prior research has shown that statistically defined mixed and single domain dysMCI patients produce similar patterns of performance on executive tests (Eppig et al. Reference Eppig, Wambach, Nieves, Price, Lamar, Delano-Wood and Lippa2012), a combined mixed/dysexecutive (mixed/dys) MCI subgroup (n = 31) was constructed.
Subtle MCI group
Eighteen of the 57 patients not meeting Jak et al. (Reference Jak, Bondi, Delano-Wood, Wierenga, Corey-Bloom, Salmon and Delis2009) criteria for MCI were classified as presenting with subtle MCI using a modification of Edmonds et al. (Reference Edmonds, Delano-Wood, Galasko, Salmon and Bondi2015) criteria. These patients scored >1 sd below the age-corrected normative mean on two of the nine neuropsychological measures in different cognitive domains (Edmonds et al., Reference Edmonds, Delano-Wood, Galasko, Salmon and Bondi2015). However, these criteria were constructed on the basis of a corpus of six neuropsychological tests. As described above, nine neuropsychological tests were used for classification. Therefore, a modification of these criteria was employed such that patients could be diagnosed with subtle cognitive impairment (SCI) if they obtained scores below criterion on 1 parameter across all three neuropsychological domains that were assessed. Of the 18 patients diagnosed with SCI, 12 patients scored below 1 sd on two neuropsychological parameters and 6 patients scored below 1 sd on three neuropsychological parameters.
Non-MCI group
Thirty-five patients did not meet criteria for either SCI (Edmonds et al., Reference Edmonds, Delano-Wood, Galasko, Salmon and Bondi2015) or MCI (Jak et al., Reference Jak, Bondi, Delano-Wood, Wierenga, Corey-Bloom, Salmon and Delis2009). Some of these individuals (n = 19) obtained scores where all nine neuropsychological parameters were above 1 sd. A second group of patients (n = 16) not meeting criteria for MCI or SCI presented with some, but very little cognitive impairment such that only one of the nine neuropsychological parameters was below the 1 sd cut-off. All of these patients were combined and classified as non-MCI.
No statistically significant differences were found for age, education, GDS, WRAT-IV, and IADLs (p > .05 for all). However, groups differed on MMSE test performance [F(3, 112) = 7.26, p < .001]. Bonferroni-corrected post hoc analyses revealed that non-MCI scored 1.82 points higher on the MMSE than aMCI patients (p = .001, d = .82, 95% CI: .58–3.06); and 1.65 points higher than mixed/dys MCI patients (p = .001, d = .83, 95% CI: .48–2.81).
Verbal and Visual WM Measures
Verbal and visual WM were assessed with the Wechsler Adult Intelligence Scale-IV (WAIS-IV) Digit Span Backwards and Wechsler Memory Scale-IV (WMS-IV) Symbol Span subtests administered and scored using standard administration procedures and expressed as Z-scores. The instructions provided to patients for the WAIS-IV Digit Span Backwards test asked patients to repeat digits in the reverse order. The string of digits ranged from two to nine, with two trials administered for each span for a total of 16 items. A patient received two points when both test trials within each span were correct and one point for a correct response when only one of the two trials was correct. The test is discontinued following the failure of two test items within the same span. In prior research with both MCI (Eppig et al., Reference Eppig, Wambach, Nieves, Price, Lamar, Delano-Wood and Lippa2012) and dementia patients, Lamar et al. (Reference Lamar, Price, Libon, Penney, Kaplan, Grossman and Heilman2007, Reference Lamar, Catani, Price, Heilman and Libon2008) found that Digit Span Backwards test performance was effective in differentiating patients with dysexecutive disorders from other groups.
The WMS-IV Symbol Span presents geometric symbols ranging in span from one to seven. Each test trial is presented for 5s. Patients are asked to remember visually presented information from left to right. After the target items are shown, patients are presented with a multiple choice array consisting of the original targets along with foils and are asked to identify the original test items as they are displayed from left to right. Two point scores are awarded for correct responses; one point is given for correctly identified symbols but in the incorrect order and zero point is given for incorrect symbols. The test is discontinued after four consecutive imperfect scores (i.e. score of one or zero). Similar to Liang et al. (Reference Liang, Pertzov, Nicholas, Henley, Crutch, Woodward, Leung, Fox and Husain2016), successful performance on this test assesses both memory for object and the location of objects in the stimulus array. The dependent variable obtained from both tests was the norm-referenced scaled score converted to Z-scores. Data from both technical manuals suggest adequate test–retest stability for both measures.
HO Measures
Only a portion of our sample had MRI hippocampal volumetric data available (n = 36). Patients were scanned using either 3.0T (n = 22) or 1.5T (n = 14) magnets compatible with the analysis software. Hippocampal volume was obtained from T1-weighted sequences. Acquisition protocol details are as follows: TR/TE = 2300/1.87/900, 192 × 192 matrix, 160 slices, voxel size = 1 × 1 × 1.2 mm. The scanners are detailed as follows: Siemens 3T Verio scanners with 16- and 32-channel head coils (Siemens Medical Systems, Erlangen, Germany), Siemens 3T Skyra scanners with a 32-channel head coil (Siemens Medical Systems, Erlangen, Germany), and Siemens 1.5T Aera scanners with a 16-channel head coil (Siemens Medical Systems, Erlangen, Germany). Following acquisition, images from the sagittal 3D T1 SPGR sequence underwent volumetric analysis using NeuroQuant® software (CorTechs Labs, Inc., San Diego, CA, USA), an FDA-approved software program used to obtain a bilateral HO score. HO is calculated as a ratio of hippocampal volume to adjacent inferior ventricular dilation using the formula described below.
Hippocampal occupancy = [left hippocampal volume (cm3)/left hippocampal volume (cm3) + left inferior lateral ventricular volume (cm3)] + [right hippocampal volume (cm3)/right hippocampal volume (cm3) + right inferior lateral ventricular volume (cm3)]
Left and right side ratios are summed and then normalized for intracranial volume (ICV), age, and gender using a database consisting of over two thousand healthy participants. Reduced hippocampal volume in conjunction with ex vacuo expansion of the inferior lateral ventricle suggests focused medial temporal lobe atrophy (McEvoy & Brewer, Reference McEvoy and Brewer2012), a finding often seen in patients with AD. This MRI measure was developed to differentiate patients presenting with congenitally small hippocampi due to other neurologic illness such as epilepsy (Farid et al., Reference Farid, Girard, Kemmotsu, Smith, Magda, Lim, Lee and McDonald2012). HO measures have been shown to differentiate between patients with intact cognitive abilities versus amnestic and non-amnestic MCI (Jak et al., Reference Jak, Panizzon, Spoon, Fennema-Notestine, Franz and Kremen2015) and predict eventual emergence of AD in conjunction with other biomarkers (Heister, Brewer, Magda, Blennow, & McEvoy, Reference Heister, Brewer, Magda, Blennow and McEvoy2011; Tanpitukpongse, Mazurowski, Ikhena, & Petrella, Reference Tanpitukpongse, Mazurowski, Ikhena and Petrella2017; Yau et al., Reference Yau, Tudorascu, McDade, Ikonomovic, James, Minhas, Mowrey, Sheu, Snitz, Weissfeld and Gianaros2015).
Statistical Analysis
Preliminary analyses
No statistically significant outliers were identified via visual inspection of the box plots for all variables analyzed. Significance tests were conducted to assess normality among all variables using a conventional but conservative alpha level of .01 based on recommendations for use with small to moderate sample sizes (Tabachnick & Fidell, Reference Tabachnick and Fidell2013). No statistically significant violations of skewness or kurtosis were found (p > .01). Visual inspection of frequency histograms with normal distribution overlays revealed no violations of normality.
Sex differences
Prior research has shown that healthy and MCI males perform better than females on visuospatial WM tasks (Elosúa, Ciudad & Contreras, Reference Elosúa, Ciudad and Contreras2017; Voyer, Voyer, & Saint-Aubin, Reference Voyer, Voyer and Saint-Aubin2017). Therefore, a between-group independent sample t test was performed to look at sex differences in both WAIS-IV Digit Span Backwards and WMS-IV Symbol Span performance.
Multivariate Analysis of Co-Variance
Between-group differences were assessed with a multivariate analysis of co-variance (MANCOVA) where WAIS-IV Digits Backward and WMS-IV Symbol Span Z-scores were dependent variables (Enders, Reference Enders2003; Huberty & Morris, Reference Huberty and Morris1989; Maxwell, Reference Maxwell and Thompson1992). Adjustment was made for MMSE test performance which was held constant as a covariate. The independent variable was diagnostic group (non-MCI, SCI, aMCI, and mixed/dys MCI). Analyses were conducted using the IBM SPSS MANOVA procedure. A subsequent analysis of variance (ANOVA) was employed on the multivariate composite from which the group centroids were derived and plotted to allow for a visual representation of group differences (Enders, Reference Enders2003; Huberty & Morris, Reference Huberty and Morris1989; Maxwell, Reference Maxwell and Thompson1992). The composite variable was computed utilizing the grand means from each dependent variable, the pooled standard deviations, and the standardized coefficients associated with each dependent variable. Structure coefficients obtained from the MANCOVA procedure were then utilized to label and describe the significant multivariate function.
Mixed design
Between-group differences were also assessed with a 4 (group) x 2 (WM tests) mixed design repeated measure ANOVA. This analysis was specifically included to assess for the presence of an interaction between patient group and WM tests.
Within-group analyses
Paired samples t tests were conducted to assess for differences between WAIS-IV Digit Span Backwards and WMS-IV Symbol Span Z-scores within each of the four diagnostic groups.
Hippocampal Occupancy
A hierarchical multiple linear regression analysis using block-wise entry of predictors was conducted to assess the relation between HO and WM test performance. In this regression model, MMSE was entered into block 1. In block 2, WAIS-IV Digit Span Backwards and WMS-IV Symbol Span Z-scores were entered. Results produced from block 2 were interpreted to assess the predictive utility of WM test performance on HO scores, after controlling for MMSE.
RESULTS
WM Analysis
MANCOVA test results
Using Wilk’s criterion, statistically significant differences were observed among the four diagnostic groups on the set of dependent variables only on Function 1 [λ1= .82, F(6, 194) = 3.48, p = .003, η 2 = .18] where approximately 18% of the variability among the two dependent variables can be attributed to the differences among the four diagnostic groups suggesting large effect size (Cohen, Reference Cohen1988). The multivariate function found that mixed/dys MCI patients scored lower than all other groups on both WM tests. Specifically, mixed/dys MCI patients scored 1.17 standard deviations lower than non-MCI patients (p < .001, 95% CI: −1.84 to −.50); 1.27 standard deviations lower than SCI patients (p < .001, 95% CI: −2.06 to −.47); and .78 standard deviations lower than aMCI patients (p = .04, 95% CI: −1.52 to −.04). Figure 1 provides a plot of the group centroids and a description of the multivariate function.
Fig. 1. Plot of composite variable centroids for MANCOVA analysis.
Gender and within-group analyses
No between-group differences were found for gender on either WAIS-IV Digit Span Backwards (p = .20) or WMS-IV Symbol Span (p = .68) test. Within-group test performance revealed no differences among non-MCI patients [t(34) = 1.82, p = .08, d = 0.31, 95% CI: −.033–.603]. By contrast, all other patient groups displayed worse performance on WMS-IV Symbol Span compared to WAIS-IV Digit Backward test performance: SCI [t(17) = 2.46, p < .026, d = 0.58, 95% CI: .08–.99], aMCI [t(22) = 2.54, p < .019, d = 0.53, 95% CI: .10–.97], and mixed/dys MCI groups [t(26) = 2.68, p < .012, d = 0.52, 95% CI: 1.02–2.69].
Repeated Measures Test Results:
A mixed design repeated measures ANOVA with group as the between-subjects variable and test as the within-subjects variable revealed a significant main effect for group [F(1, 99) = 22.80, p < .001, η 2 p = .19] and a significant main effect for test [F(1, 99) = 1.78, p < .001, η 2 p = .20]. There was no significant group × test interaction [F(3, 99) = .561, p = .642, η 2 p = .02].
Hippocampal Occupancy
A one-way ANOVA for HO was not significant (Table 2). Hierarchical multiple linear regression found that MMSE test performance explained 13.8% of the variance in HO scores [R = .371, r 2 = .138, F(1, 34) = 5.44, p = .026, b = .026, SE(b) = .011, 95% CI(b): .003–.049]. The addition of both visual and verbal tests in block 2 significantly incremented the amount of variance explained in HO scores by 15.3% [R = .539, ΔF(3, 32) = 3.45, p < .044, Δr 2 = .153]. However, in block 2, only WMS-IV Symbol Span predicted HO [b = .051, SE(b) = .021, p < .024, 95% CI(b): .007–.094] explaining an additional 12.5% of variance in HO (sr = .353). WAIS-IV Digit Span Backwards did not statistically predict HO [b= −.048, SE(b) = .027, p = .078, 95% CI(b): −.103–.006], explaining only 7.3% unique variance (sr = −.271) in HO.
Table 2. Demographic and clinical information: Means and standard deviations
SCI = subtle cognitive impairment; aMCI = amnestic mild cognitive impairment; mixed/dys MCI = multi-domain/dysexecutive mild cognitive impairment; IADL = instrumental activities of daily living; WRAT-IV = Wide Range Achievement Test-IV; ns = not significant.
DISCUSSION
The current research assessed the notion that MCI patients will obtain lower scores on visual compared to verbal WM tests. Implicit in this notion is that visual WM tests may be a particularly sensitive indicator of emergent illness compared to verbal WM tests. On the basis of prior research, we also tested whether successful WM test performance was associated with intact HO measures. Co-normed WAIS-IV Digit Span Backwards and WMS-IV Symbol Span tests were used to measure verbal and visual WM test performance, respectively. A protocol of neuropsychological tests was used to construct a grouping variable such that statistical criteria permitted the classification of patients into two MCI groups (aMCI & mixed/dys MCI; Jak, Reference Jak, Bondi, Delano-Wood, Wierenga, Corey-Bloom, Salmon and Delis2009): patients presenting with SCI (Edmonds et al., Reference Edmonds, Delano-Wood, Galasko, Salmon and Bondi2015) and patients presenting with little to no cognitive impairment (non-MCI).
Between-group analyses found that mixed/dys MCI patients scored lower on both WM tests compared to all other groups. However, the repeated measures ANOVA did not yield a statistical interaction showing worse visual WM test performance limited to SCI and MCI groups. A possible reason for this could be the modest sample size in our groups. Nonetheless, within-group analyses found that non-MCI patients did not perform significantly different whereas patients with SCI and MCI scored significantly lower on visual WM compared to verbal WM tests. However, the absence of a statistical interaction demonstrating worse visual WM performance confined to MCI groups could suggest that worse performance on visual versus verbal WM tests might be restricted to mixed/dys MCI groups rather than all patient groups. Alternatively, it might be the case that the WMS-IV Symbol Span subtest, as a more difficult task, provided lower scores in each group, and to a similar degree.
Despite some inconsistency regarding our between-group analyses, the data described above are in line with recent episodic serial list learning research suggesting that visual memory deficits may emerge before analogous verbal memory impairment (De Anna et al., Reference De Anna, Felician, Barbeau, Mancini, Didic and Ceccaldi2014; Didic et al., Reference Didic, Felician, Barbeau, Mancini, Latger-Florence, Tramoni and Ceccaldi2013; Okonkwo et al., Reference Okonkwo, Oh, Koscik, Jonaitis, Cleary, Dowling, Bendlin, LaRue, Hermann, Barnhart and Murali2014). Several reasons might explain why SCI and MCI patients scored lower on visual as compared to verbal WM tests. First, despite the fact that these tests are co-normed, it is possible that the WMS-IV Symbol Span subtest is more difficult than WMS-IV Digit Span Backwards subtest because the diverse neurocognitive skills necessary for successful Symbol Span test performance are less automatized. For example, the ability to repeat a string of numbers is frequently practiced in everyday life by the utilization of phone numbers and passwords, etc. Second, WMS-IV Symbol Span performance requires patients to concomitantly recall both geometric objects and their correct serial order location. Also, WMS-IV Symbol Span stimulus location varies between test trials with longer test trials (i.e. >3 span test stimuli) containing increasing numbers of response foils such that memory for object location (Liang et al., Reference Liang, Pertzov, Nicholas, Henley, Crutch, Woodward, Leung, Fox and Husain2016), in addition to memory for stimulus object, likely factors into successful test performance. Indeed, past research using visual episodic memory tests has dissociated memory for object versus object location (de Toledo-Morrell et al., Reference de Toledo-Morrell, Dickerson, Sullivan, Spanovic, Wilson and Bennett2000; Hampstead, Stringer, Stilla, Amaraneni, & Sathian, Reference Hampstead, Stringer, Stilla, Amaraneni and Sathian2011; Piekema, Kessels, Mars, Petersson, & Fernández, Reference Piekema, Kessels, Mars, Petersson and Fernández2006; Troyer et al., Reference Troyer, Murphy, Anderson, Hayman-Abello, Craik and Moscovitch2008). Third, it is very common for patients to assign verbal labels to non-representational visual stimuli suggesting the recruitment of verbally mediated cognitive operations. Thus, in addition to the non-automatized nature of the test paradigm, the degree to which patients’ call upon all even a portion of these neurocognitive operations suggests that a wider array of neurocognitive operations underlie successful visual WM test performance.
The non-automatized and diverse neurocognitive skills necessary for successful visual WM test performance is consistent with research showing an association between synaptic strength and integrity, and frequency of utilized neurocognitive networks. For example, post-synaptic density and integrity have been shown to be neuroprotective against dementia such as AD (Clare, King, Wirenfeldt, & Vinters, Reference Clare, King, Wirenfeldt and Vinters2010; Gillick and Zirpel, Reference Gillick and Zirpel2012; Markham & Greenough, Reference Markham and Greenough2004; Scheff & Price, Reference Scheff and Price2003; Terry et al., Reference Terry, Masliah, Salmon, Butters, DeTeresa, Hill, Hansen and Katzman1991). Also, past research has shown that older patients require greater neural resources, that is, a wider neurocognitive network, for successful performance on visual versus verbal WM tests than younger patients (Habeck, Rakitin, Steffener, & Stern, Reference Habeck, Rakitin, Steffener and Stern2012; Smith et al., Reference Smith, Geva, Jonides, Miller, Reuter-Lorenz and Koeppe2001). Neurodegenerative illness is often distributed across multiple brain regions including association cortex (Murray et al. Reference Murray, Graff-Radford, Ross, Petersen, Duara and Dickson2011). As such widely distributed and less routinized neurocognitive network(s) could very well be more susceptible to emergent pathology (Hultsch, Hertzog, Small, & Dixon, Reference Hultsch, Hertzog, Small and Dixon1999). All of these factors could explain why visual WM tests appear to be more sensitive to emergent pathology than verbal WM tests.
HO was associated with WMS-IV Symbol Span test performance, whereas the relationship between HO and WAIS-IV Digit Backwards test performance was not significant. This is consistent with literature suggesting that the hippocampus is particularly important for complex high-resolution bindings when tasks require the coordination of multiple neurocognitive activities (Leszczynski et al., Reference Leszczyński, Fell and Axmacher2015; Liang et al., Reference Liang, Pertzov, Nicholas, Henley, Crutch, Woodward, Leung, Fox and Husain2016; Moses & Ryan, Reference Moses and Ryan2006; Yonelinas, Reference Yonelinas2013). Indeed, there is lively debate in the literature regarding the exact nature of these putative binding mechanisms (Liang et al., Reference Liang, Pertzov, Nicholas, Henley, Crutch, Woodward, Leung, Fox and Husain2016; Moses & Ryan, Reference Moses and Ryan2006). Unfortunately, the WMS-IV Symbol Span test does not permit a level of analysis that would help clarify these issues. The association between HO and visual WM test performance described in the current research is also consistent with recently documented hippocampal/prefrontal afferent pathways necessary to encode spatial information (Spellman et al., Reference Spellman, Rigotti, Ahmari, Fusi, Gogos and Gordon2015).
The medial temporal lobes and bilateral hippocampi have been implicated in both verbal and visual episodic and WM (Bohbot et al., Reference Bohbot, Kalina, Stepankova, Spackova, Petrides and Nadel1998; Burgess et al., Reference Burgess, Maguire and O’Keefe2002; Crane & Milner, Reference Crane and Milner2005; Davachi & Wagner, Reference Davachi and Wagner2002; Ezzati et al., Reference Ezzati, Katz, Zammit, Lipton, Zimmerman, Sliwinski and Lipton2016; Olson, Moore, Stark, & Chatterjee, Reference Olson, Moore, Stark and Chatterjee2006; Piekema et al., Reference Piekema, Kessels, Mars, Petersson and Fernández2006; Squire, Reference Squire1992). The hippocampus has also been linked to spatial location with research showing deficient memory for object location in patients with damage to the right temporal lobe, including the hippocampus (Smith & Milner, Reference Smith and Milner1981, Reference Smith and Milner1989). Animal models have demonstrated that skills involving topographical location, navigation, orientation, and spatial mapping have been found in the medial temporal lobes and include place cells, grid cells, and head direction cells (Burgess et al., Reference Burgess, Maguire and O’Keefe2002; Fu et al., Reference Fu, Rodriguez, Herman, Emrani, Nahmani, Barrett, Figueroa, Goldberg, Hussaini and Duff2017; O’Keefe, Burgess, Donnett, Jeffery, & Maguire, Reference O’Keefe, Burgess, Donnett, Jeffery and Maguire1998).
Fu et al. (Reference Fu, Rodriguez, Herman, Emrani, Nahmani, Barrett, Figueroa, Goldberg, Hussaini and Duff2017) found deficiencies in grid cell function of a transgenic mouse model expressing tau in the entorhinal cortex, including destabilized grid fields, reduced firing rates, and altered network activity. Fu and colleagues (Reference Fu, Rodriguez, Herman, Emrani, Nahmani, Barrett, Figueroa, Goldberg, Hussaini and Duff2017) concluded that tau pathology was associated with deficits in spatial memory and may underlie deterioration of spatial neurocognitive skills seen in AD. A possible explanation for the association between HO and visual WM test performance may revolve around the role of the right hippocampus and encoding topographical and spatial location.
The current research is not without limitations. There was a relatively small number of patients where HO scores were available. In addition, there was limited areas of interests on the MRI scans. Another limitation is the absence of a normal control group and longitudinal assessment. All of these issues must be addressed before firm conclusions can be drawn whether visual WM is, in fact, more sensitive to emergent illness. Additionally, combining 1.5T and 3.0 MRI scans might have influenced our results (Chow et al., Reference Chow, Hwang, Hurtz, Green, Somme, Thompson, Elashoff, Jack, Weiner and Apostolova2015). Another limitation is our inability to examine visual WM in relation to performance on visual episodic memory and visual attention; and visuoperceptual abilities including object location, as performance on all of these neurocognitive domains may be moderated by other factors such as visual processing speed (Bublak et al., Reference Bublak, Redel, Sorg, Kurz, Förstl, Müller, Schneider and Finke2011; Ruiz-Rizzo et al., Reference Ruiz-Rizzo, Bublak, Redel, Grimmer, Müller, Sorg and Finke2017).
However, the current research has several strengths, including the fact that both WM tests were co-normed and the use of comprehensive, neuropsychological criteria to classify patients into their respective groups. In sum, the within-group analyses demonstrate worse WMS-IV Symbol Span compared to WAIS-IV Digit Span Backwards performance in SCI and MCI patents; and the association between WMS-IV Symbol Span test performance and greater HO ratios suggests that visual rather than verbal WM tests may be particularly sensitive to emergent neuropsychological decline in MCI.
ACKNOWLEDGEMENTS
No funding was received for this study.
CONFLICT OF INTEREST
The authors have nothing to disclose.
