Authors: Seyda Cankaya, Betul Yalciner, Melek Kandemir Yilmaz, Burak Yulug
Categories: Research Articles, MCI, MRI, VMPT, depression, Research Article
Source: Alzheimer's & Dementia : Translational Research & Clinical Interventions
Doi: 10.1002/trc2.12448
Authors: Seyda Cankaya, Betul Yalciner, Melek Kandemir Yilmaz, Burak Yulug
The present study aims to assess the differences between major depressive disorder (MDD) and mild cognitive impairment (MCI) in terms of verbal learning profile together with structural changes in the brain on magnetic resonance imaging (MRI) and to reveal predictive factors for MCI.
Fifty‐six patients with MDD and 31 MCI subjects were assessed using the Turkish Verbal Memory Processes Test (VMPT). Brain MRI was used to evaluate sulcal atrophy (SA), ventricular atrophy, periventricular white matter hyperintensity (WMH), subcortical WMH, basal ganglia infarct, medial temporal lobe atrophy, and infratentorial infarct scores based on the Modified Visual MRI Rating Scale (MVMRS). The symptoms of depression were evaluated with the Beck Depression Inventory in both groups. Demographic factors, VMPT scores, and MVMRS scores between MDD and MCI groups were compared. Also, potential predictors of MCI were analyzed by binary logistic regression analyses.
The total scores of VMPT and the scores of VMPT subgroups, including immediate memory, highest learning, total learning, and delayed recall, were significantly higher in the MDD groups compared to MCI patients (Mann‐Whitney U, Student's t‐test, p < 0.05), indicating that higher scores were associated with better memory. The total MVMRS score and a subgroup of MVMRS, the SA score, were significantly higher in MCI patients compared to the MDD group, suggesting more atrophic changes and a higher burden of infarction in MCI patients. In our statistical analyses, impaired immediate memory (p < 0.001; OR = 6.002; 95% CI: 1.996−18.042), increased SA (p = 0.008; OR = 1.522; 95% CI: 1.118−2.073), and education (p = 0.028; OR = 0.84; 95% CI: 0.719−0.981) were significant predictive values obtained through backward Wald elimination in the binary logistic regression model for detecting MCI.
Our findings suggest that VMPT may potentially represent a novel neuropsychiatric test that might be combined with MRI‐based morphometric evaluation methods, such as MVMRS.
Although depressive symptoms occur in a significant proportion of individuals with mild cognitive impairment (MCI), the mechanisms underlying the association between depression and cognitive decline have not yet been elucidated. Among the various mechanisms proposed for the common pathophysiology between depression and MCI, ^1^ a special role has been attributed to structural and functional hippocampal integrity, a strong indicator of immediate memory performance in both healthy and depressive patients. ^2^
Based on the evidence of these structural and functional data, it is not unreasonable to assume that the presence of depression has been increasingly identified as a risk factor for dementia. ^3^ However, subsyndromal depressive symptoms can also be observed during the prodromal phases of dementia, making not only a clear‐cut diagnosis in this intermediate zone challenging, ^4^ but also appearing to increase the risk of progression of dementia, which prompted us to evaluate the impact of both factors on MCI.
The situation is also complicated by the fact that even some cerebrovascular risk factors, as well as cerebrovascular disease itself, play a critical role in contributing to the development of MCI, dementia, and depression late in life. ^5^ This accords with a recent meta‐analysis suggesting that the incidence of not only MCI but also major depressive disorder (MDD) increases significantly with age. ^6^ However, as briefly mentioned above, there are challenges in differentiating between depression and MCI, given the existing body of data that supports the presence of a subgroup with depression within MCI. This includes considerable meta‐analyses indicating the prevalence of depression within MCI is 32%, suggesting that it is essential to consider that depressed symptoms might also serve as early indicators or markers of dementia. ^7^ Another main reason contributing to the diagnostic difficulties between MCI and MDD is that cognitive impairment in MCI is often less pronounced than in dementia. Additionally, there are some overlapping critical brain regions responsible for emotions and cognition which present a strong rationale for the increased risk of dementia in the MDD population.
Especially worthy of mention here is that MDD patients with cognitive impairment are also at a greater risk of developing dementia compared to those without MDD. ^8^ This, in turn, indicates the importance of accurately identifying these patients by recommending regular cognitive follow‐up examinations.
However, such a clear‐cut identification of these specific groups of patients can be challenging, and a more comprehensive and specific assessment is urgently required. In addition to well‐known conventional screening tools such as the Mini‐Mental State Examination (MMSE), we also need a more specific approach for these particular groups of neurological and psychiatric diseases.
The Turkish Verbal Memory Process Test (VMPT) ^9^ is one of the most frequently used memory tests for this purpose due to its ability to elicit verbal information in several ways, focusing especially on normal and interfered prefrontal executive functions. ^10^ , ^11^ The VMPT, an adapted and validated form of the Rey Auditory Verbal Learning Test (RAVLT) for the Turkish population, has been extensively used in various cognition studies and it has been shown that the VMPT is a very useful test for assessing neurodegenerative and non‐degenerative cognitive disorders, and MCI. ^10^
^−^
^12^ Although the VMPT is a valuable tool for detecting executive functions in patients with cognitive impairment, there is still insufficient data available regarding the radiological correlates of these test scores. To fill this gap, we used a combination of the VMPT and the Modified Visual MRI (magnetic resonance imaging) Rating Scale (MVMRS) in this present study. The MVMRS has been developed by Yalciner et al. as a scale that evaluates the atrophy, white matter hyperintensities (WMHs), basal ganglia infarct (BGI), and infratentorial infarct (ITI) together for proposing a practical and standardized MRI for the clinicians to be used in daily practice. ^13^ In this study, we used the MVMRS for detecting levels of WMHs, ventricular atrophy (VA), and parahippocampal atrophy, which were applied to differentiate MDD from MCI in the study population.
In this context, we have collected real‐world data from patients who applied for a non‐specific memory clinic, making our findings valuable and relevant for the general neurology population.
Our main aim was to show the utility of the MVMRS and cognitive tests in distinguishing MDD from MCI and to evaluate the predictive role of the existence of MDD and/or depressive symptoms in MCI based on our previous work that indicated the value of the MVMRS for patients with forgetfulness. Our hypothesis was suggested by Yalcin et al.’s observation that the MVMRS is a useful scale for discriminating patients with Alzheimer's dementia and depressive patients. ^13^
Participants diagnosed with MDD and MCI were recruited from the general neurology clinic at the Bayindir Hospital Icerenkoy. Individuals who completed evaluations of their neuropsychological and depressive complaints were included in this analysis (n = 87). Fifty‐six pure MDD patients and 31 with pure MCI constituted the study group.
Depression and MCI were diagnosed based on all available clinical and neuropsychological information by consensus of a panel of neurologists, neuropsychologists, and psychiatrists in line with international guidelines. All participants underwent a comprehensive clinical assessment, including the Structured Clinical Interview for DSM Disorders (SCID) for the Diagnostic and Statistical Manual of Mental Disorders, 5th Edition (DSM‐V) ^14^ to establish a history of MDD. Also, MCI subjects were identified according to the criteria for MCI, ^15^ which included (1) a memory complaint, preferably confirmed by an informant; (2) objective memory impairment, adjusted for age and education; (3) normal or near‐normal performance on general cognitive functioning and no or minimum impairment of daily life activities; and (4) not meeting the criteria for dementia according to the DSM‐V.
Using the Beck Depression Inventory (BDI; categorized as 0 to minimal depression, 10 to mild depression, 19 to moderate depression, 30 to severe depression) and the DSM‐V, we included patients who were either MCI or MDD, while excluding patients who fell into the overlapping category between these two diseases, as well as those with acute stages of depression and severe dementia that might potentially bias the cognitive results.
Following neuropsychological assessments, the results were discussed in order to exclude other potential causes, such as acute neurological diseases (eg, stroke, Parkinson's disease, normal pressure hydrocephalus), severe or unstable medical diseases (eg, cardiac disease, hypertension, current infection, neoplasia, and clinically significant hepatic, renal, pulmonary, metabolic, or endocrine disturbances), or the presence of comorbid diseases, including delirium, alcohol or drug dependency, and schizophrenia. The study was conducted in a retrospective and double‐blinded manner. Within that context, we have excluded potential causes that can mimic a degenerative cognitive status.
This retrospective study of clinical data was approved by the Bayindir Hospital Icerenkoy Ethics Committee (2010/826). Due to the retrospective nature of the study, no informed consent was required or obtained.
The MVMRS was created by Yalciner et al. based on the combined measurement of atrophy, WMH, BGI, and ITI, ^13^ with a high ability to discriminate MDD from MCI and high interrater reliability in the Turkish population, making this multi‐modal tool a strong candidate to provide a practical and standardized MRI for clinicians in daily practice. ^13^ It is also worth noting that evaluating the interrater reliability of magnetic resonance visual assessment using the MVMRS between two neurologists and a radiologist has resulted in results ranging from good to excellent. ^16^ The details of the MVMRS are shown in Table 1.
In this scale, SA, VA, medial temporal lobe atrophy (MTA), periventricular WMH (PWMH), subcortical WMH (SCWMH), BGI, and ITI are scored separately on the scale.
SA was graded from 0 to 9 based on the widths of the central sulci, interhemispheric fissure, and other cortical sulci on the axial T1 (repetition time [TR] = 673 ms, echo time [TE] = 12 ms, 3 or 5 mm slice thickness, 0.6 mm gap) image corresponding to the slice level where the central sulci are best viewed (most similar to the shape of an inverted omega). VA was graded on a 0 to 9 range on the axial T1 (TR = 673 ms, TE = 12 ms, 3 or 5 mm slice thickness, 0.6 mm gap) slice showing the frontal and occipital horns of the lateral ventricles together with the third ventricle.
Periventricular WMH was graded on a 0 to 4 range based on the WMH identified as continuous, confluent areas of high signal intensity adjacent to anterior or posterior horns of the lateral ventricles and along the ventricular system on fluid‐attenuated inversion recovery (FLAIR; TR = 8000 ms, TE = 80 ms, 3 or 5 mm slice thickness, 0.6 mm gap) images.
Subcortical WMH was graded on a 0 to 4 range based on the WMH identified as lesions located in the white matter but not touching the periventricular area on FLAIR (TR = 8000 ms, TE = 80 ms, 3 or 5 mm slice thickness, 0.6 mm gap) images.
BGIs were graded on a 0 to 2 range from FLAIR (TR = 8000 ms, TE = 80 ms, 3 or 5 mm slice thickness, 0.6 mm gap) images, based on the hyperintensities in the caudate nucleus, putamen, globus pallidus, and thalamus regions.
ITIs were graded in a 0 to 2 range from FLAIR (TR = 8000 ms, TE = 80 ms, 3 or 5 mm slice thickness, 0.6 mm gap) images, based on the hyperintensities in the brainstem and cerebellum.
The VMPT is a word‐list learning test developed, adapted, and validated by Öktem based on the RAVLT with high test‐retest and interrater reliability. ^9^ , ^17^
The VMPT differs from the RAVLT in terms of the cognitive processes targeted for assessment and the application processes of the verbal material. ^18^ The VMPT permits the evaluation of the processes of working memory, learning or acquiring knowledge, retention of information, and recalling. It differs in some respects from the traditional RAVLT, which prioritizes the measurement of normal and interfered prefrontal executive functions. The Turkish version of RAVLT, like the original version, consists of several word lists (Lists A and B) measuring learning, free recall, free recall under proactive interference, recall after a very long time, and recognition over recall (RR), as follows. The immediate memory score (number of responses given by the subject in the initial pronunciation of the words), complete learning points (number of attempts ensuring complete learning = “access to criteria” score), total learning score (total number of words recalled in each trial), the highest learning point (the maximum number of words the subject was capable of remembering in trials), and long‐term recall scores are determined in a two‐step process. In the first step, 15 words are read out 10 times. After each reading, the patient is asked to repeat what he can remember (total learning). In the second step, the patient is asked to repeat the same 15 words after 45 minutes (delayed free recall). ^9^ , ^18^
Immediate memory is categorized as 0 = normal (5 to 10 points), or 1 = impaired (0 to 4 points); highest learning is categorized as 0 = normal (14 or 15 points), 1 = moderate (10 to 13 points), or 2 = severe (10> points); total learning is categorized as 0 = normal (100 to 140 points), 1 = moderate (80 to 99 points), or 2 = severe (80> points); delayed recall is categorized as 0 = normal (11 to 15 points), 1 = moderate (7 to 10 points), or 2 = severe (7> points); the VMPT total score is categorized as 0 = normal (14 to 15 points), 1 = moderate (10 to 13 points), or 2 = severe (10> points) in the VMPT. ^17^
Patient characteristics of MDD and MCI were compared using the respective two‐sample tests (t‐test, Mann‐Whitney U test, or chi‐square test for normally distributed variables, non‐normally distributed metric variables, and nominal variables, respectively).
To identify predictors of MCI, in a binary logistic regression model, all potential predictors were considered singly (univariate analysis) in the first step. Variables included sociodemographics (age, education), verbal process (VMPT), and radiologic findings (MVMRS). The potential predictors were analyzed in the second step with multivariable logistic regression analysis (enter method). Finally, a binary logistic regression model was built using a backward Wald approach in the third step, with variables retained at a p‐value of <0.05. In addition, odds ratios (ORs) were determined to measure effect sizes.
Completed data on the MRI and neurocognitive tests were available for 87 individuals, and these were included in the study. There was no significant difference in terms of age, sex, and years of education between groups (p > 0.05) (Table 2). Also, the severity of the BDI score (χ ^2^ = 0.035, p = 0.983) and the mean of BDI total score (Student's t‐test, p = 0.946) showed no difference between depressive and MCI patients (Table 2).
The median VMPT total score (Mann‐Whitney U test, p = 0.034) and selected subscores of the VMPT (immediate memory, highest learning, total learning, delayed recall) were significantly higher in depressive patients than in the MCI group (higher scores indicating greater memory) (p < 0.05) (Table 2).
We detected that not only the MVMRS total score (Mann‐Whitney U test, p = 0.047), but also the SA score (Mann‐Whitney U test, p = 0.013), which is a subgroup of the MVMRS scale, were higher in the MCI group, indicating a high burden of infarct and larger atrophy (Table 2).
In univariate logistic regression analyses, we observed that impaired immediate memory, low highest learning score, low total learning score, increased SA, delayed recall, and aging were significant predictors for MCI. Multivariable logistic regression analysis adjusted for age and education showed that impaired immediate memory and atrophic findings (SA and VA) remained as significant predictive values. In our statistical analyses, impaired immediate memory (p = 0.001; OR = 6.002; 95% confidence interval [CI]: 1.996−18.042), increased SA (p = 0.008; OR = 1.522; 95% CI: 1.118−2.073), and education (p = 0.028; OR = 0.84; 95% CI: 0.719−0.981) were predictive values obtained through backward Wald elimination in the binary logistic regression model for detecting MCI (Table 3). Our findings of the logistic regression analyses for the identification of predictors of MCI are shown in Table 3.
The results of this study revealed significant differences in VMPT subscores and morphometric scores of the MVMRS assessment between the MCI and depression groups. Furthermore, impaired immediate memory scores and increased SA were associated with a sixfold and a one and one half‐fold greater diagnosis rate with MCI, respectively. The risk of dementia decreased by 0.84 times as the years of education increased.
VMPT sub‐scores also changed, suggesting impaired executive functions in the MCI group compared to the MDD group.
Although there is sizable literature on the association of late‐life depressive symptoms with incident MCI and dementia, ^3^ we revealed that neither the levels of depression nor its existence might significantly contribute to the development of MCI, suggesting that parameters such as altered morphology and impaired cognitive scores might be an independent risk factor for MCI. A good example is seen in the study by Rosenberg et al.,^8^ indicating that neuropsychiatric symptoms observed in the early phases of preclinical Alzheimer's disease (AD) in individuals with MCI are strongly linked to a higher likelihood of developing dementia and AD.
Herein, our binary logistic regression analysis showed no significant impact of BDI severity on being MCI while backward Wald regression showed that only SA and impaired immediate memory were significant in predicting MCI along with education (Table 3).
Several studies have shown that the RAVLT is an important test for differentiating dementia from pseudo‐dementia, as well as cognitively normal individuals from those with MCI. ^19^ Additionally, Estévez‐González et al. found that the RAVLT is a powerful tool for discriminating individuals with subjective memory complaints from those with MCI. ^20^
Powell et al. found that particularly seven RAVLT recall trials and the total of Trials I‐V were more sensitive in discriminating a group of normal individuals from a mixed cognitive impairment group. ^21^
Furthermore, performance on the RAVLT is affected by a variety of neurological conditions, ^22^ including hydrocephalus, ^23^ vertebrobasilar insufficiency, ^24^ and early Alzheimer‐type dementia. ^25^
However given the undeniable overlapping of depressive states and degenerative dementia, ^8^ it is worth mentioning that not only depressive states but even some subtle depressive symptoms might be prominent signs of MCI. For instance, Smith et al. have recently highlighted this issue by showing that MCI patients might be initially presented with some subtle depressive neuropsychiatric symptoms related to a concomitant degeneration of serotonergic and dopaminergic neurons before the cognitive degenerative stage is apparent. ^26^ This is clinically evident in a recent work by Ebrahim et al., showing that affective dysregulation and apolipoprotein (APOE) ε4 carrier status in the context of mild behavioral impairment is associated with a considerable risk of developing dementia and amnestic MCI, ^27^ fitting well with a novel study by Smith et al. indicating a possible link between serotonergic degeneration and amyloid beta accumulation in the pathogenesis of late‐life depression. ^28^ In a reversed pattern, the role of serotonergic and dopaminergic polymorphism in behavioral symptoms of dementia has already been observed, ^29^ which has been suggested by several studies indicating that depressive patients might show some cognitive dysfunctional patterns resembling a cognitive profile in MCI, making it difficult to make a clear‐cut discrimination between these two diseased conditions. ^8^ , ^30^ , ^31^ Considering all these diagnostic difficulties, we evaluated only pure MCI patients and patients with the diagnosis of only MDD to prevent any bias related to overlapped symptoms and diseased states reported in both conditions.
In the backward Wald binary logistic regression analysis, we observed that immediate memory was the only variable as a predictive factor of MCI that survived among other VMPT parameters. A recent study from Turkey using the adapted Turkish‐language version of the RAVLT, in other words, the VMPT, showed that the subtests of the verbal memory processes, including short‐term memory recall (total learning) and long‐term memory recall (delayed free recall), were capable of differentiating MCI from mild dementia. ^32^ Our findings indirectly suggest recent functional data indicating that impaired connectivity between the right amygdala, occipital and parietal lobe, and left hippocampus is accompanied by a greater decrease in the RAVLT and immediate recall in MCI than in MDD‐MCI and controls. ^2^
To summarize, these findings suggest that the VMPT may potentially represent a novel neuropsychometric test that sufficiently differentiates MCI from MDD. Further studies with additional imaging tools, particularly those considering the relevant brain regions, should be used to differentiate patients with MDD from those in the early stages of dementia, such as MCI.
Despite some minor limitations, such as the small sample size and lack of a control group, this study is clinically significant in showing that practical and powerful cognitive tests are essential in identifying subtle cognitive impairments in both MDD and MCI patients. Such a purely clinical approach can be reinforced with practical morphological assessment tools, of which the MVMRS total score assessment seems to be a particularly suitable candidate.
The authors declare no conflicts of interest. Author disclosures are available in the supporting information.
The ethics committee at Bayindir Hospital Icerenkoy, Istanbul, Turkey, approved the study protocol. Written informed consent was obtained from all participants or, in the case of cognitively impaired persons, from a proxy (usually a guardian or a family member).