From: Functional neuroimaging in subjective cognitive decline: current status and a research path forward
Authors | Year | Citation | Mod | Task/Rest | Methodology | Operationalization | Participants | Results | Pattern |
---|---|---|---|---|---|---|---|---|---|
Alexander et al. | 2006 | [25] | EEG | Eyes open and closed resting. Visual working memory task | Average power spectra over 4 s periods during resting state. Measure of phase synchrony during task. | Participants with SCD recruited from general population with complaint of memory problems increasing over time and informant confirmation. Controls age and gender matched. | 79 SCD, 79 control | Greater α band power, increased frontal θ power, and increased spatiotemporal wave activity during working memory in SCD. Greater α power and wave activity related to lower verbal memory performance and reaction time, and greater reverse digit span performance. | Increase |
Rodda et al. | 2009 | [26] | fMRI | Verbal episodic memory task | Voxelwise task contrasts. Group comparison with ANOVA. Cluster correction. | Participants with SCD had persistent memory concerns and were recruited from a memory clinic. Controls were recruited from a prior study and were similar to SCD in age and education. | 10 SCD, 10 control | Greater left prefrontal cortex activation in SCD during encoding. No differences in recognition rate. | Increase |
Maestu et al. | 2011 | [27] | MEG | Sternberg’s letter-probe task | MEG source reconstruction. Nonparametric permutation testing to identify spatio-temporal clusters that distinguish the groups. | Participants with SCD recruited from memory clinic with memory deterioration confirmed by informant. Participants scored below 9 on the Geriatric Depression Scale and greater than 13 on the Memory Failures of Everyday test. Healthy elderly control participants recruited from university educational courses. | 12 SCD, 18 control | Greater task-related activation within 200–900 ms after stimulus onset in SCD in parietal, temporal, occipital, motor/premotor, and dorsal prefrontal regions. | Increase |
Rodda et al. | 2011 | [28] | fMRI | Divided attention task with visual letters and auditory numbers | Voxelwise task contrasts. Group comparison with ANOVA. Cluster correction. | Participants with SCD had persistent perceived decline from prior memory performance and were recruited from a memory clinic. Controls were recruited as part of a prior study. | 11 SCD, 10 control | Greater thalamus, caudate, posterior cingulate, hippocampus, and parahippocampus task-related activation in SCD. No differences on performance. | Increase |
Dumas et al. | 2013 | [29] | fMRI | N-back working memory task | Voxelwise analysis. Group comparisons with random effects ANOVA (group by working memory load). Cluster-level correction. | Healthy, older, post-menopausal women recruited from general population. Participants identified as SCD if they endorsed > 20% of the items of the Cognitive Complaint Index and as control otherwise. | 12 SCD, 11 control | Greater working memory task-related activity in precuneus, midfrontal, and cingulate gyri in SCD. | Increase |
Hafkemeijer et al. | 2013 | [30] | fMRI | Eyes closed resting state | Independent components analysis and dual regression. | Participants with SCD recruited from memory outpatient clinic. | 25 SCD, 29 control | Greater DMN and medial temporal connectivity in SCD. | Increase |
Dillen et al. | 2016 | [31] | fMRI | Eyes open resting state | Functional seed-based correlation analysis. | Participants with SCD recruited from outpatient memory clinic. Healthy controls recruited from general population through advertisements. Participants with SCD had memory complaint scores > 24 and normal cognitive test results. | 27 SCD, 25 control | Greater functional connectivity between retrosplenial cortex and ventromedial prefrontal cortex in SCD. | Increase |
Sun et al. | 2016 | [32] | fMRI | Eyes closed resting state | Voxelwise amplitude of low-frequency fluctuation analysis. Group differences evaluated with voxelwise general linear model analysis. | Participants with SCD recruited from a memory clinic. Participants with SCD reported persistent decline in memory corroborated by an informant. Healthy controls recruited from the general population and did not harbor cognitive concerns. | 25 SCD, 61 control | Greater low-frequency signal amplitude in superior temporal, cerebellar, occipital, and inferior parietal cortex in SCD. Greater low-frequency amplitude associated with poor verbal learning within the SCD group. | Increase |
Cespon et al. | 2018 | [33] | EEG | Simon task | Evaluation of P300 latency and amplitude during executive control processes. | Participants recruited from general population and split into low-SCD and high-SCD groups based on scores of a memory complaints questionnaire. | 18 low-SCD, 16 high-SCD | Greater medial prefrontal negativity during task associated with greater degree of subjective cognitive decline. | Increase |
Lazarou et al. | 2018 | [34] | EEG | Administration of pictures of facial affect during EEG | N170 event-related potential evaluation. | Participants recruited from a memory clinic. SCD status based on SCD international working group suggestions. | 14 SCD, 12 control | Greater N170 (negative) amplitude in response to faces expressing fear. | Increase |
Verfaillie et al. | 2018 | [35] | fMRI | Resting state | Parcellation-based functional connectivity analysis. Linear regression evaluated strength of association between SCD and connectivity strength. | Healthy older adults recruited from general population if they had a family history of Alzheimer’s disease. Participants subsequently classified as SCD if they responded “Yes” to the question “Do you think your memory is becoming worse?” | 68 SCD, 56 control | Greater connectivity between posterior DMN and medial temporal regions, in SCD. | Increase |
Kawagoe et al. | 2019 | [36] | fMRI | Eyes closed resting state | Principle component multivariate pattern analysis for functional connectivity. | Participants recruited from the general population. Subjective memory score measured as a continuous variable. | 155 participants with SCD as a continuous variable | Greater connectivity between lingual gyrus and cuneus, lingual gyrus and precuneus, superior parietal lobe and postcentral gyrus, and between cuneus and many occipital and association areas in SCD. | Increase |
Bajo et al. | 2012 | [37] | MEG | Sternberg’s letter-probe task. | Functional connectivity patterns, via synchronization likelihood, evaluated for task hits. | Participants with SCD recruited from memory clinic with memory deterioration confirmed by informant. Participants scored below 9 on the Geriatric Depression Scale and greater than 13 on the Memory Failures of Everyday test. Healthy elderly control participants recruited from university educational courses. | 12 SCD, 25 control | Diffuse lower synchronization between electrode pairs in α and β bands in SCD. | Decrease |
Wang et al. | 2013 | [38] | fMRI | Eyes closed resting state | Independent components analysis and dual regression. | Participants recruited through advertisements and through referrals from medical centers. Group classification then based on results of neuropsychological assessment, self-, and informant-report. All participants classified by clinical consensus panel. | 23 SCD, 16 control | Lower connectivity between right hippocampus and the DMN in SCD. | Decrease |
Yasuno et al. | 2015 | [18] | fMRI | Eyes closed resting state | Region-of-interest based Pearson’s correlation functional connectivity analysis. | Participants with and without SCD recruited from hospital psychiatry unit. SCD classification based on Reisberg criteria. | 23 SCD, 30 control | Lower functional connectivity between retrosplenial cortex and dorsomedial prefrontal cortex, and between retrosplenial and anterior cingulate cortex in SCD. No differences between groups in amyloidopathy. | Decrease |
Lopez-Sanz et al. | 2016 | [39] | MEG | Eyes closed resting state | MEG source reconstruction and spectral analysis. Groups differences assessed with ANCOVA. | Participants recruited from a hospital neurology department, a center for the prevention of cognitive impairment, and a senior center. Cognitive concerns self-reported and SCD status determined by multidisciplinary panel. | 39 SCD, 41 control | Diffuse lower relative α power in SCD. No differences in α peak frequency slowing. | Decrease |
Contreras et al. | 2017 | [40] | fMRI | Eyes closed resting state | Independent components analysis resting-state network analysis. | Participants with SCD had cognitive concerns but tested normally on cognitive tests. Controls had no significant cognitive concerns. | 16 SCD, 13 control | Greater functional connectivity within all resting-state networks negatively associated with the cognitive complaint index. | Decrease |
Dillen et al. | 2017 | [41] | fMRI | Resting state | Region-of-interest based temporal network modeling. | Participants with SCD recruited from outpatient memory clinic. Healthy controls recruited from general population through advertisements. Participants with SCD had memory complaint scores > 24 and normal cognitive test results. | 28 SCD, 25 control | Decreased functional connectivity between hippocampus and posterior DMN in SCD. Retrosplenial cortex mediated connectivity between hippocampus and DMN in controls but not SCD. | Decrease |
Hayes et al. | 2017 | [42] | fMRI | Event-related visual memory encoding task | Voxelwise contrasts based on subsequently remembered items. Higher level analysis carried out with FMRIB’s Local Analysis of Mixed Effects. | Participants with and without SCD recruited from general population. However, participants with SCD saw a medical professional regarding their complaints prior to participation. | 23 SCD, 41 control | Weaker task-related activity related to successful encoding in occipital, superior parietal, and cingulate cortex in SCD. No differences in retrieval performance. | Decrease |
Hu et al. | 2017 | [43] | fMRI | Intertemporal decision task with an episodic imagination task embedded within | Voxelwise task contrasts for choice process and subjective valuation components of task. Group level factorial designs. | Participants with SCD recruited from memory clinic and reported decline in memory with onset in the past 5 years. Control participants recruited from the general population and reported no cognitive concerns. | 20 SCD, 24 control | Poor temporal reward decision-making related to reduced hippocampal engagement in SCD. Lower insulae activation during task-switching in SCD. | Decrease |
Mazzon et al. | 2018 | [44] | EEG | Eyes closed resting state and Rey’s Auditory Verbal Learning task | Kruskal-Wallis and Wilcoxon rank sum tests compared groups on regional relative band power. | Participants recruited from a memory center. Inclusion criteria for SCD included persistent memory complaints within the past 5 years, normal cognitive performance, and no psychiatric disease. | 8 SCD, 7 control | Lower parietal β and γ band power in SCD during a memorization task. | Decrease |
Yang et al. | 2018 | [45] | fMRI | Eyes closed resting state | Amplitude of low-frequency fluctuation evaluation. Support vector machine evaluation of ALFF for group discrimination. | Participants with SCD recruited from a memory clinic. Healthy controls recruited from general population. SCD determination based on the SCD international working group definition and made by experienced neurologists. | 44 SCD, 57 control | Lower amplitude and fraction amplitude of low-frequency fluctuations in precuneus, anterior cingulum, and cerebellum in SCD. | Decrease |
Lopez-Sanz et al. | 2019 | [46] | MEG | Eyes closed resting state | Source power spectra estimation and group classification based on source power analysis with regularized logistic regression with the Least Absolute Shrinkage and Selection Operator. | Participants recruited from a hospital neurology department, a center for the prevention of cognitive impairment, and a senior center. Cognitive concerns self-reported and SCD status determined by multidisciplinary panel. | 91 SCD, 70 control | Lower relative α band power, predominantly in frontal regions, associated with SCD. | Decrease |
Viviano et al. | 2019 | [47] | fMRI | Eyes closed resting state | Region-of-interest based functional connectivity analysis. | Participants with and without SCD recruited from general population. However, participants with SCD saw a medical professional regarding their complaints prior to participation. | 35 SCD, 48 control | Lower functional connectivity between retrosplenial cortex and precuneus in SCD. | Decrease |
Wang et al. | 2019 | [48] | fMRI | Eyes closed resting state | Vertex-based functional connectivity analysis and graph-theoretic approaches to the functional connectome. | Participants with SCD recruited from memory clinic and status determined by two psychiatrists according to the international working group standard proposed by Jessen et al. Participants without SCD recruited from general population. | 32 SCD, 40 control | Lower subgraph centrality in occipital and paracentral regions in SCD. | Decrease |
Babiloni et al. | 2010 | [49] | EEG | Eyes closed resting. | Spectral and cortical source analysis of EEG data. | Participants with SCD recruited from outpatient clinics with memory centers. Participants with SCD had z-scores > − 1.5 on cognitive battery. Controls also recruited from the clinics, but without and neurologic or psychiatric disease, and without social limitations. | 53 SCD, 79 control | Greater frontal δ band amplitude, lower parietal and occipital θ and α1 (8–10.5 Hz) amplitude, and greater occipital α2 (10.5–13 Hz) amplitude in SCD. | Complex |
Erk et al. | 2011 | [50] | fMRI | Visual memory encoding, recall, and recognition tasks. N-Back working memory task | Voxelwise task contrasts. Group comparison with ANOVA (full-factorial design). Family-wise error correction across whole brain at p < .05, or across a-prior hippocampal and dorsolateral prefrontal regions of interest. | Participants with SCD recruited from memory clinic and had informant corroboration. Healthy controls without complain recruited from the general population. All participants scored within 1.5 standard deviations on all subtests of the Consortium to Establish a Registry for Alzheimer’s Disease. | 19 SCD, 20 control | Lower posterior hippocampal activation and greater right dorsolateral prefrontal cortex activation during recall in SCD. No differences in task performance or task-related activity during memory encoding. No group differences in N-back task. | Complex |
Vega et al. | 2016 | [51] | fMRI | Resting state | Voxelwise seed-based functional connectivity. Second-level regression analysis associated cognitive complaint index with voxelwise connectivity. | Healthy, older, post-menopausal women recruited from general population. Participants identified as SCD if they endorsed > 20% of the items of the Cognitive Complaint Index and as control otherwise. Cognitive complaints evaluated as a continuous variable. | 31 SCD (continuous) | Greater within executive control network and within middle temporal gyrus connectivity, and lower frontal cortex functional connectivity in SCD. | Complex |
Lopez-Sanz et al. | 2017 | [52] | MEG | Eyes closed resting state | Source reconstruction and phase synchronization functional connectivity analysis. | Participants recruited from a hospital neurology department, a center for the prevention of cognitive impairment, and a senior center. Cognitive concerns self-reported and SCD status determined by multidisciplinary panel. | 41 SCD, 39 control | Increased α-band connectivity between anterior regions and decreased α-band connectivity between posterior regions in SCD. | Complex |
Jiang et al. | 2018 | [53] | fMRI | Eyes closed resting state | Vertex-based functional connectivity analysis. | Participants with SCD recruited from memory clinic and status determined by two psychiatrists according to the international working group standard proposed by Jessen et al., [1]. Participants without SCD recruited from general population. | 42 SCD, 54 control | Lower vertex-wise index of functional criticality (a measure of signal standard deviation, within cluster correlation, and correlation with components outside a cluster) in the inferior temporal gyri and frontal poles, and greater criticality in precuneus, cingulate, parietal, and ventromedial prefrontal cortices in SCD. | Complex |
Li et al. | 2018 | [54] | fMRI | Eyes open resting state | Whole-brain Pearson’s correlations and graph-theoretic methods to evaluate functional connectivity differences between groups. | Participants with SCD and healthy controls pulled from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database. | 44 SCD, 40 controls | Participants with SCD exhibited greater degree centrality in hippocampus, and fusiform gyrus, and lower inferior parietal degree centrality. No differences between groups in eigen centrality. No differences between groups in amyloidopathy or tauopathy. | Complex |
Xie et al. | 2019 | [55] | fMRI | Eyes closed resting state | Static and dynamic functional connectivity based on connectivity matrices derived from the Automated Anatomical Labeling atlas. | Participants with SCD recruited from a memory clinic. Healthy controls recruited from general population. SCD determination based on the SCD international working group definition and made by experienced neurologists. | 40 SCD, 53 control | Centrality frequency (proportion of time a degree centrality hub region appeared in a dynamic functional connectivity analysis) differed between SCD and controls. Lower gyrus rectus and cingulum, and greater hippocampus, calcarine, lingual, and occipital centrality frequency in SCD. | Complex |
Yan et al. | 2019 | [56] | fMRI | Eyes closed resting state | Multimodal (functional connectome and structural connectome via diffusion imaging) classifier training. | Participants with SCD recruited from a memory clinic. Healthy controls recruited from general population. SCD determination based on the SCD international working group definition and made by experienced neurologists. | 39 SCD, 45 control | Classification accuracy for distinguishing SCD from controls ranged from 72 to 77% over a set of classifiers trained on connectome data. Important regions for classification included frontal, parietal, temporal, and hippocampal regions. | Complex |
Dierks et al. | 1997 | [57] | EEG | Eyes closed resting. | EEG segmented into microstates based on the locations of centroids of electrical activity during spikes in spatial variance. | Participants with SCD recruited from memory clinic, did not exhibit difficulties in everyday living, and performed less than one standard deviation below age reference average on cognitive tasks. Control subjects did not harbor memory complaints and came from the general population. | 31 SCD, 21 control | Duration of microstates, distinct spatial patterns of electrical activity, reduced in dementia. No difference between controls and SCD. | No difference |
Lopez-Sanz et al. | 2017 | [58] | MEG | Eyes closed resting state | Source reconstruction and region-of-interest based functional connectivity analysis. Graph theoretic approaches utilized. | Participants recruited from a hospital neurology department, a center for the prevention of cognitive impairment, and a senior center. Cognitive concerns self-reported and SCD status determined by multidisciplinary panel. | 55 SCD, 63 control | Decreased small-world characteristics of mild cognitive impairment brains compared to healthy controls in θ and β bands. Individuals with SCD exhibited a similar pattern, though not significantly different from controls. | No difference |
Teipel et al. | 2018 | [59] | fMRI | Resting state | Pearson’s correlation based functional connectivity. Amplitude of low-frequency fluctuation evaluation. Evaluation of functional connectivity to distinguish SCD via cross-validated discrimination accuracy based on penalized logistic regression. | Participants recruited from the general population and grouped as SCD if they answered “Yes” to the questions “Are you complaining about your memory?” and “Is it a regular complaint that lasts more than 6 months?” | 90 SCD, 80 controls | Functional connectivity did not reach significant discrimination accuracy for distinguishing SCD from controls. | No difference |
Contreras et al. | 2019 | [60] | fMRI | Eyes closed resting state | Within- and between-network functional connectivity based on whole-brain parcellation. | Participants drawn from two cohorts. Participants with SCD had cognitive concerns but tested normally on cognitive tests. Controls had no significant cognitive concerns. | 27 SCD, 31 control | No significant difference between controls and SCD. | No difference |
Scarapicchia et al. | 2019 | [61] | fMRI | Eyes open resting state | Association between resting-state voxelwise BOLD variability and white matter hyperintensities older adults with and without SCD. | Participants drawn from ADNI database. SCD status determined by ADNI investigators. Participants self-reported cognitive concerns and scored > 15 on the first 12 items of the Cognitive Change Index. Control participants had no significant cognitive concerns. | 19 SCD, 19 control | No significant difference between controls and SCD in blood-oxygen-level dependent variability. Higher white matter hyperintensity burden associated with greater variability in temporal regions for controls only. | No difference |