Neuropsychological,Metabolic,and Connectivity Underpinnings of Semantic Interference Deficits Using the LASSI-L

Abstract

Background:

LASSI-L is a novel neuropsychological test specifically designed for the early diagnosis of Alzheimer’s disease (AD) based on semantic interference.

Objective:

To examine the cognitive and neural underpinnings of the failure to recover from proactive semantic and retroactive semantic interference.

Methods:

One hundred and fifty-five patients consulting for memory loss were included. Patients underwent neuropsychological assessment, including the LASSI-L, and FDG-PET imaging. They were categorized as subjective memory complaints (SMC) (n=32), pre-mild cognitive impairment (MCI) due to AD (Pre-MCI) (n=39), MCI due to AD (MCI-AD) (n=71), and MCI without evidence of neurodegeneration (MCI-NN) (n=13). Voxel-based brain mapping and metabolic network connectivity analyses were conducted.

Results:

A significant group effect was found for all the LASSI-L scores. LASSI-L scores measuring failure to recover from proactive semantic interference and retroactive semantic interference were predicted by other neuropsychological tests with a precision of 64.1 and 44.8%. The LASSI-L scores were associated with brain metabolism in the bilateral precuneus, superior, middle and inferior temporal gyri, fusiform, angular, superior and inferior parietal lobule, superior, middle and inferior occipital gyri, lingual gyrus, and posterior cingulate. Connectivity analysis revealed a decrease of node degree and centrality in posterior cingulate in patients showing frPSI.

Conclusion:

Episodic memory dysfunction and the involvement of the medial temporal lobe, precuneus and posterior cingulate constitute the basis of the failure to recover from proactive semantic interference and retroactive semantic interference. These findings support the role of the LASSI-L in the detection, monitoring and outcome prediction during the early stages of AD.

Keywords

Alzheimer’s disease connectivity FDG-PET interference memory mild cognitive impairment neuropsychological assessment

INTRODUCTION

The early diagnosis of Alzheimer’s disease (AD) is one of the main research priorities in the field of neurodegenerative brain disorders. Despite the advances in the development and validation of neuroimaging techniques and other biomarkers, neuropsychological testing plays a central role in the evaluation of patients with suspected AD. In the last few years, several novel paradigms have been developed for the detection of memory impairment in the early stages of AD, including the preclinical and prodromal phases [1]. In this regard, Loewenstein et al. (2003) found that patients with AD were especially vulnerable to the effects of semantic interference [2]. Accordingly, the Loewenstein-Acevedo Scales of Semantic Interference (LASSI-L) was developed [3]. In this paradigm, two lists of 15 items belonging to three semantic categories are presented to evaluate key processes of episodic memory, such as encoding, storage and learning, free and cued recall, proactive and retroactive semantic interference, delayed recall, and uniquely, the recovery from proactive semantic interference. Importantly, this novel test has been validated for the diagnosis of mild cognitive impairment (MCI) and mild AD in several independent cohorts and countries [4–7] and has shown adequate cross-cultural properties [8, 9]. The assessment of recovery from proactive semantic interference using a second administration of the second list of items has been shown as one of the main advantages of the LASSI-L over other traditional memory tests. Furthermore, the interest in studying the effects of interference in episodic memory is even more important in the current society due to the increasing amount of information and data sources that are encountered in daily living.

Proactive interference occurs when previously stored memories jeopardize the learning or remembering of new information. Mechanisms underlying proactive interference and its recovery are not well understood. Executive functioning and, specifically inhibition, could play a role in interference by suppressing previously acquired information in the short-term memory [10]. A post-retrieval selection process that resolves the competition between relevant and non-relevant information has been involved, in combination with post-retrieval monitoring [11]. Both processes are associated with the detection and resolution of proactive interference [12] and have been linked to the ventrolateral and dorsolateral prefrontal cortex, respectively. According to the frontal aging hypothesis and the expected role of the frontal lobe in proactive interference, previous research found an increase in the susceptibility to proactive interference with aging in working memory, episodic memory, and associative memory [12, 13]. However, recent investigations using diffusion decision models have found that inhibitory control and proactive interference remained unimpaired with aging [14]. Similarly, studies including fMRI have shown no differences between older and young adults in their efficacy in resolving proactive interference, although middle aged asymptomatic offspring of late-onset AD patients have produced semantic intrusion errors in recovering from proactive semantic interference, which is related to cortico-limbic dysconnectivity on MRI [7]. Another interesting but less studied aspect of proactive interference is the ability to release or recover from proactive interference effects [15]. Both encoding and retrieval processes participate in the recovery from proactive interference [16]. On the one hand, encoding is challenged in the context of competing information during learning. On the other hand, at the retrieval stage, previously encoded information induces a problem of discrimination. Neuroimaging studies have associated release from proactive interference to the left ventrolateral prefrontal cortex, but also to the medial prefrontal cortex and hippocampus due to the participation of encoding and retrieval [16]. Early studies of some of these paradigms in the context of AD also suggested that failure to release from proactive interference might be a consequence of combined impairment of episodic memory and executive function emanating from frontal lobe impairment [17, 18].

Retroactive semantic interference (RSI) occurs when newer memories make difficult the retrieval of or damage previous acquired memories. Patients with mild dementia due to AD and MCI showed greater susceptibility to RSI effects [2, 19]. In contrast to proactive semantic interference, RSI did not discriminate mild AD and MCI [3]. In healthy young participants, posterior (but not anterior) hippocampal connectivity and glutamate concentrations were associated with RI [20]. In patients with MCI and non-demented community-dwelling subjects, RSI showed moderate correlations with amyloid load on ¹⁸F-Florbetapir in anterior cingulate, and to a lesser extent in the precuneus, posterior cingulate and frontal lobe [4]. Correlations with medial temporal brain volumes were inconclusive [21] or non-significant [22, 23].

Several studies have evaluated the correlation between the LASSI-L performance and main neuroimaging biomarkers of AD (Table 1). Specifically, the test has been associated with the amyloid load [1 , 24] and cortical thickness in several brain regions, including the hippocampus, precuneus, inferior temporal lobes, superior parietal lobules, and temporal poles [22]. Interestingly, CRB2 (Cued Recall from the List 2), the LASSI-L score measuring the failure to recover from proactive semantic interference (frPSI), was more closely associated with brain volumes and amyloid load than other LASSI-L scores associated with encoding or storage. In addition, these findings have been reproduced in several cohorts covering the different stages of the early stages of AD, including the offspring of patients with late-onset AD, pre-MCI, and MCI [23–26]. In the pioneer study of the LASSI-L, Loewenstein et al. calculated several indices by subtracting the raw scores from the two lists [6]. Using these indices, the authors found statistically significant differences between MCI and controls, especially for those indices associated with proactive interference and recovery from proactive interference. These indices were also calculated in the clinical validation study in MCI and AD, replicating the findings [5]. The comparison between scores is one of the strengths of the test, as the performance may be referenced against the individual’s own performance and memory capacity [22]. However, these indices are less sensitive in group analysis due to the different stages of the memory impairment in the AD course and the non-normal distribution of the indices. In this regard, the CRB2 was considered a better score for capturing the frPSI according to classification analyses between groups and for neuroimaging correlation [5 , 25].

Table 1

Main studies evaluating the neuroimaging correlates of the LASSI-L

	Number of participants	Diagnosis	Neuroimaging technique	Main results
Loewenstein et al. 2016 [4]	31	7 controls, 11 subjective memory complaints, 3 pre-MCI, and 10 MCI	PET (18F-Florbetapir; region of interest analysis)	Correlation with amyloid load, especially CRB2.
Loewenstein et al. 2017 [22]	67	38 controls, 29 MCI	MRI (T1-weigthed imaging) (cortical thickness; region of interest analysis)	Correlation with hippocampal, precuneus, inferior temporal lobe, superior parietal lobe, temporal pole, and middle frontal volumes (especially CRB2).
Loewenstein et al. 2017 [23]	45	45 MCI	MRI (T1-weigthed imaging) (cortical thickness)	Correlation with hippocampus, precuneus, temporal pole, superior parietal lobule, and inferior temporal lobe.
Sanchez et al. 2017 [7]	41	21 offspring of late-onset AD and 20 controls	Functional MRI (resting-state)	CRB2 intrusion errors inversely correlated with connectivity between antero-dorsal thalamus and contralateral posterior cingulate in the offspring of patients with AD.
Crocco et al. 2018 [25]	73	53 controls, 23 Pre-MCI	MRI (T1-weigthed imaging) (cortical thickness; region of interest analysis)	Correlation of CRB2 with hippocampus, precuneus, superior parietal lobule, posterior cingulate, superior frontal, supramarginal gyrus, rostral middle frontal
Abulafia et al. 2019 [26]	45	27 offspring of patients with late-onset AD and 18 controls	MRI (cortical thickness; region of interest analysis)	Correlation of intrusions in CRB2 with left medial posterior parietal cortex, left temporo-occipital cortex and right superior frontal gyrus in offspring of AD patients.
Present study	155	32 subjective memory complaints, 39 Pre-MCI, 71 MCI-AD, 13 MCI-NN	FDG-PET (voxel-based analysis, connectivity)	Correlation with brain metabolism of precuneus, fusiform gyrus, angular gyrus, lingual gyrus, supramarginal gyrus, inferior and superior parietal lobule, inferior and middle temporal gyri, superior, middle and inferior occipital, and posterior cingulate. Association with connectivity of the posterior cingulate.

The previous results from the LASSI-L support the hypothesis that the recovery from proactive semantic interference is a valid cognitive hallmark that may be more sensitive than traditional memory paradigms in evaluating the early stages of AD [6]. However, the neuropsychological mechanisms and biological substrates of both recovery from proactive semantic interference and RSI in the context of the early stages of AD are still unclear [25].

Positron emission tomography with the ¹⁸F-fluorodeoxyglucose radiotracer (FDG-PET) is a reliable biomarker of regional brain metabolism, synaptic failure, and neurodegeneration [27]. In AD, hypometabolism is highly correlated with tau deposition, and the extension of hypometabolism is associated with the stage of the disease [28]. Thus, FDG-PET is a useful technique to evaluate the neural underpinnings of cognitive function [29]. In addition, in recent years, FDG-PET has also been proposed to evaluate brain connectivity (i.e., metabolic connectivity). This method has been suggested as a reliable marker of normal and pathological cognitive function [30, 31].

In this study, we aimed to examine the cognitive and neural underpinnings of frPSI and RSI. We hypothesized that frPSI and RSI are more associated with episodic memory function than other neuropsychological domains (e.g., executive functioning), and thus, they are associated with brain regions impaired in the early stages of AD. Considering that the LASSI-L has been suggested as a sensitive test in the early stages of AD, and frPSI one of the hallmarks in the preclinical and prodromal stages of the disease, episodic memory should have a greater role than executive function because memory deficits are among the most initial and predominant symptoms since the earliest stages of the disease [32, 33]. However, executive function could also have a role, although we hypothesized that its contribution should be lower because structures associated with executive function are less impaired in the first stages of the disease [34, 35].

To this end, we examined a large cohort of patients evaluated due to memory loss that underwent a comprehensive neuropsychological examination, including the LASSI-L and FDG-PET imaging. Patients were categorized according to the presence of AD or not, as well as the clinical stage, and we evaluated the best scores of the LASSI-L to distinguish between groups, the neuropsychological predictors of the frPSI and RSI, and their metabolic correlates.

METHODS

In this section, we report how we determined our sample size, all data exclusions, all inclusion/exclusion criteria, whether inclusion/exclusion criteria were established prior to data analysis, all manipulations, and all measures in the study.

Participants

One hundred and fifty-five patients consulting for memory loss with no functional impairment in daily living activities were included. Mean age was 71.75 ´ 7.60, 78 (50.3% were women, and mean years of education was 11.15 ´ 4.88). All patients were Spaniards and were natives Spanish-speakers. All patients were examined with a comprehensive neuropsychological protocol and underwent FDG-PET imaging. Time between neuropsychological assessments and FDG-PET was less than two months. The LASSI-L was administered to all participants but not used for diagnostic purposes.

These sample sizes were determined to be sufficient to detect likely overall group effect sizes, based on empirical observations in other studies involving these aspects and this disorder, and supported by formal power calculations.

The neuropsychological protocol used for diagnosis and group classification included the following tests: digit span forward and backward, Corsi’s test forward and backward, Trail Making Test (TMT) (parts A and B), Symbol Digit Modalities Test (SDMT), Stroop Color-Word Interference test (word reading, color naming, and interference) (SCWIT), Tower of London-Drexel version (ToL), Boston Naming Test (BNT), Free and Cued Selective Reminding Test (FCSRT), Rey-Osterrieth Complex Figure (copy, memory at 3 and 30 min, and recognition) (ROCF), Judgment of Line Orientation (JLO), and the Visual Object and Space Perception Battery (VOSP) (subtests object decision, progressive silhouettes, discrimination of position, and number location). This battery was co-normed in our country and has been specifically recommended to assess patients with suspected AD and other causes of dementia [36, 37]. Raw scores are transformed into age-, education-, and sex-adjusted scaled-scores (mean 10, standard deviation 3) [38, 39]. For this study, we used the word version of 16 items of the FCSRT. In this version, sixteen written words are shown with a semantic cue. Each item belongs to a different semantic category, and items are presented in sheets of four items. The examinee is asked to read aloud and identify the name of each item when the neuropsychologist says the category cue. After all the items have been read and identified, a free recall for 90 s is attempted. Items not spontaneously remembered are cued by the examiner. Free and cued recall is repeated three times. The total free recall and total recall are the sums of the words remembered during the learning trials (free recall: maximum score 48; total recall is the sum of free and cued recall; maximum score 48). Finally, a 30-minutes delayed free and cued recall is conducted (free recall, maximum score 16; total delayed recall, the sum of free and cued recall, maximum score 16).

Patients were categorized into four different groups as follows:

MCI-AD (n=71): These individuals presented with subjective memory complaints and had a FCSRT (delayed recall or total recall scaled-score) equal to or below 6 The FCSRT is the test recommended explicitly by the International Working Group for the diagnosis of prodromal AD and the patients had no significant impairment in social and/or occupational function that would merit a diagnosis of major neurocognitive disorder or dementia [40]. Further, the presence of AD pathology was considered when the diagnosis was supported by at least one biomarker (temporoparietal hypometabolism in FDG-PET, and/or altered levels of amyloid-β₁₄₂ and p-tau in the cerebrospinal fluid (CSF) analysis. CSF biomarkers were measured in 23 patients (32.4%). Mean follow-up of this group was 36.56 ´ 26.00 months, confirming the diagnosis.

Pre-MCI due to AD (Pre-MCI AD) (n=39): These individuals presented with memory complaints but had a FCSRT (delayed recall or total recall scaled-score) equal to or above 7. There was no significant impairment in social and occupational function that would merit a diagnosis of major neurocognitive disorder or dementia [41]. Further, AD was considered since the diagnosis was supported by at least one biomarker (temporoparietal hypometabolism in FDG-PET, and/or altered levels of amyloid-β₁₄₂ and p-tau in the CSF analysis. CSF was performed in 17 (43.6%) patients. Mean follow-up on this group was 37.66 ´ 24.14 months (45.68 ´ 24.69 in those patients with no CSF available).

Subjective Memory Complaints (SMC:NN (n=32): Those with subjective memory complaints thought to be non-neurodegenerative presented with subjective memory complaints but had a FCSRT (delayed recall or total recall scaled-score) equal to or above 7. There was no significant impairment in social and occupational function that would merit a diagnosis of major neurocognitive disorder or dementia [41]. Further, AD was not considered since the diagnosis was not supported by at least one biomarker (temporoparietal hypometabolism in FDG-PET or altered levels of amyloid-β₁₄₂ and p-tau in the CSF analysis). CSF was performed in 8 cases (25.0%), with normal values. Mean follow-up on this group was 36.56 ´ 26.00 months (41.87 ´ 26.86 in those with no CSF available), confirming the absence of cognitive or functional decline.

MCI without evidence of neurodegeneration (MCI-NN) (n=13): These individuals had initial memory complaints and an abnormal FCSRT and on follow-up (delayed recall or total recall) showed a scaled-score equal to or below 6 on baseline but normal functionality upon an average follow-up of 36 ´ 27.43 months and no initial abnormalities on PET or CSF biomarkers. CSF biomarkers were performed in 4 cases (30.8%) in this group, in all cases with normal values of tau and amyloid-β_142 . In patients in which no CSF was performed, mean follow-up was 41.00 ´ 29.45 months.

Patients with significant vascular lesions (Fazekas rating scale for white matter lesions ≥2) or clinical symptoms suggestive of a neurodegenerative disorder not related with AD (e.g., parkinsonism, prominent behavioral disorder, etc.) were excluded. The LASSI-L was not used for diagnosis. In all cases, LASSI-L was administered at the time of the first assessment.

LASSI-L

Patients are asked to remember two semantically competing lists of 15 words each. The words belong to three common semantic categories (fruits, musical instruments, and articles of clothing), are presented on cards (one word per card) and read aloud by the participants. A first list (list A) of 15 words is administered. The participant reads aloud the words presented one at a time at 4-s intervals. After the 15 words are read, the neuropsychologist asks the participant to recall the words. Free recall takes 60 s (Free Recall 1 list A, FRA1). Then, the semantic cues are provided (20 s for each category) (Cued Recall list A 1, CRA1). Then, the first list is administered again to maximize encoding and storage. Afterwards, a cued recall (Cued Recall list A 2, CRA2) is performed. After the second cued recall of the first list, a new list with 15 words (list B) belonging to the same semantic categories is subsequently presented. Free and cued recall performances are again assessed using the same procedure (Free Recall 1 list B, FRB1; Cued Recall 1 list B, CRB1). Then, list B is administered again, and another cued recall is conducted which provides a unique opportunity to examine the failure to recover from proactive interference (frPSI) (Cued Recall 2 list B, CRB2). Afterwards, the patient is asked to remember list A with free and cued recall trials (Short-Delay Free Recall list A, SdFRA; Short-Delay Cued Recall list A; SdCRA). Finally, after 20 min, a delayed free recall of both lists is obtained (Delayed Recall, DR). The scores are interpreted as follows: FRA1 and CRA1 are considered the scores associated with initial learning; CRA2 measures maximal storage; FRB1 and CRB1 are the scores influenced by proactive semantic interference; CRB2 represents the recovery from proactive semantic interference; and SdFRA and SdCRA are susceptible to the effects of retroactive semantic interference [4]. The administration procedure and scores of the LASSI-L are summarized in Fig. 1. An additional index for the evaluation of recovery from interference was calculated as follows: (CRB2 – CRB1).

Fig. 1

LASSI-L administration procedure. The characteristics of the tests, scores, and main memory processes linked to each score are shown. Scores associated with the List A are depicted in blue, while scores associated with List B are shown in orange.

The FCSRT and LASSI-L were administered during distinct assessment sessions.

FDG-PET acquisition, preprocessing, and analysis

PET images were acquired according to the European standards for brain FDG-PET imaging [42]. Statistical Parametric Mapping (v12) was used for preprocessing and voxel-based brain mapping analysis. Images were normalized to the Montreal Neurological Institute space using an FDG-PET template specifically validated for neurodegenerative disorders [43] and were smooth at full width at half maximum of 12 mm. The cerebellum was used for spatial normalization. A multiple regression analysis was conducted to evaluate the association between LASSI-L scores and brain metabolism, controlling by age, sex, and years of education. A p-value <0.001 and a cluster-lever FWE-corrected p<0.05 were applied.

The topology of metabolic brain networks was evaluated using several graph theory parameters. The whole-brain connectivity matrix was created considering the 116 regions from the AAL atlas [44]. These regions corresponded to the nodes of the graph, and the connections between them were the edges. Edges were calculated with the Pearson correlation coefficient between pairs of regions. Negative correlations were replaced by their absolute values. The weighted connectivity matrices were binarized using network densities. Metabolic connectivity analyses were conducted at the group-level. Thus, a comparison between groups with high (first tercile) and low (third tercile) performance on CRB2 was performed to evaluate the network measures associated with frPSI. Patients within the third tercile of CRB2 score were regarded as impaired and showing a frPSI. Between-group network analyses were conducted on the binary undirected graphs through 5000 non-parametric permutations across a range of densities from 10% to 35%, in steps of 1%. The following global measures were calculated: global efficiency, characteristic path length, clustering, and modularity. The following node indices were computed: node degree and closeness centrality. A two-tailed false-discovery rate corrected p-value <0.05 was used. The network analyses were carried out using BRAPH software (http://braph.org/) [45].

Statistical analysis

Statistical analysis was conducted using SPSS 24.0. Descriptive results are shown as absolute frequency (percentage) or mean ´ standard deviation. We used ANCOVA to evaluate differences between the diagnostic groups controlling by age and years of education. Homogeneity of variance was checked with the Levene test. Post-hoc analyses were corrected for multiple comparisons with the Bonferroni test. Effect sizes were evaluated using eta squared (η²), and regarded as small (0.010), moderate (0.058), and large (0.137). A p-value <0.05 was considered statistically significant. ANCOVA results were also used to calculate the adjusted scores for CRB2.

We implemented a Random Forest supervised classification model for discriminating between AD (comprising PreMCI-AD and MCI-AD) and non-AD groups (SMC-NN and MCI-NN). The dataset was randomly divided into training (70%) and test (30%) sets. All the LASSI-L scores were introduced as features. Accuracy, F1-score, precision, recall, and area under the curve were used as evaluation metrics. Random Forest models were also used to rank the different LASSI-L scores according to their importance in the classification using the mean decrease of accuracy measure.

Automatic Linear Modeling (LINEAR) was conducted to identify the most meaningful neuropsychological tests predictors of LASSI-L scores [46], specifically for the prediction of CRB2 (as a measure of recovery from PSI) and SdCRA (indicative of RI). All patients were introduced for these analyses. This method avoids several issues of the standard linear regression analysis, permitting to search of all possible models for best subset modelling. Best subsets were selected applying the Akaike information criterion. To improve the stability of the models, we used the bootstrap aggregation method. Previous studies have used this procedure to determine the neuropsychological predictors of a test or a behavior that can be quantified with a score [47, 48]. All the raw scores of the tests included in the neuropsychological protocol were entered to identify the most important predictors of CRB2.

Standard protocol approval, registration, and patient consent

The study was conducted with the approval of our hospital’s ethics committee, and all participants (or their legal representatives) gave written informed consent. No part of the study procedures or analyses was pre-registered in a time-stamped, institutional registry prior to the research being conducted, although decisions regarding design and analysis were decided a priori.

RESULTS

Group comparisons

A significant group effect was found for all the LASSI-L scores, controlling for age and years of education (Table 2, Fig. 2). Post-hoc analysis showed statistically significant differences between SMC and Pre-MCI in CRB1, SdFRA, SdCRA, and DR. Patients with SMC showed higher performance on all the LASSI-L scores in comparison to MCI-AD. Pre-MCI showed better scores than MCI-AD in FRA1, CRA1, CRA2, FRB1, CRB1, CRB2, and DR. MCI-AD showed lower scores than MCI-NN in FRB1, CRB1, CRB2, and DR. Effect sizes were large for all statistically significant scores. None of the LASSI-L subscales evidenced statistically significant differences between MCI-NN versus PreMCI-AD and MCI-NN versus SMC.

Table 2

ANCOVA. Comparison between groups for the LASSI-L scores controlling by the effect of age and years of education. Raw values of each variable are shown

	SMC-NN	PreMCI-AD	MCI-AD	MCI-NN	F (p)	Eta squared
Number of subjects	32	39	71	13	–	–
Age	67.06´8.27	74.59´5.67	73.44´7.14	65.54´4.66	12.05 (<0.001)#	0.193
Sex (females)	17 (53.1%)	23 (59.0%)	30 (42.3%)	8 (61.5%)	3.77 (0.287)*	–
Years of education	13.19´4.68	10.08´4.96	10.48´4.85	13.08´3.52	3.80 (0.012)#	0.070
MMSE	28.09´2.45	26.74´2.57	25.05´2.92	28.61´1.38	10.33 (<0.001)	0.172
FRA1^b,d	7.84´3.09	6.10´2.36	4.51´1.98	7.15´1.62	11.23 (<0.001)	0.184
CRA1^b,d	9.47´2.27	7.38´2.35	5.49´2.32	7.69´2.09	19.25 (<0.001)	0.279
CRA2^b,d,f	11.66´2.63	9.97´2.42	7.90´2.96	11.31´1.43	13.31 (<0.001)	0.211
FRB1^b,d,f	6.53´2.50	4.64´2.19	3.41´1.75	6.31´2.25	12.76 (<0.001)	0.204
CRB1^a,b,d,f	7.38´2.87	5.05´2.62	3.85´2.23	6.69´1.37	11.79 (<0.001)	0.192
CRB2^b,d,f	10.63´2.91	8.15´2.54	6.39´2.87	10.54´1.33	15.03 (<0.001)	0.232
SdFRA^a,b	6.13´3.18	2.85´1.91	2.39´2.03	4.77´1.53	14.25 (<0.001)	0.223
SdCRA^a,b,d	8.22´2.19	5.62´2.43	4.27´2.48	6.23´2.65	15.74 (<0.001)	0.241
DR ^a,b,d,f	17.00´5.46	11.26´4.99	6.42´5.88	15.77´3.89	24.86 (<0.001)	0.334
(CRB2-CRB1)	3.25´1.62	3.10´2.57	2.54´2.16	3.84´1.06	1.91 (0.13)	0.037

*Chi-Squared test; ^#ANOVA. Post-hoc analysis showed statistically significant differences after Bonferroni correction between: SMC versus Pre-MCI (a), SMC versus MCI-AD (b), SMC versus MCI-NN (c), Pre-MCI versus MCI-AD (d), Pre-MCI versus MCI-NN (e), MCI versus MCI-NN (f). FRA1, Free Recall trial 1 list A; CRA1, Cued Recall list A trial 1; CRA2, Cued Recall list A trial 2; FRB1, Free Recall List B trial 1; CRB1, Cued Recall List B trial 2; CRB2, Cued Recall List B trial 2; SdFRA, Short-Delay Free Recall List A; SdCRA, Short-Delay Cued Recall List A; DR, Delayed Recall.

Fig. 2

Raincloud plot showing the distribution and scores of LASSI-L between groups. Left plot represents data from each patient; middle graph represents mean and standard error using a box plot; right graph shows a split-half violin plot to represent the data distribution using the probability density function of observations. SMC-NN, subjective memory complaints; Pre-MCI-AD, pre-mild cognitive impairment; MCI-AD, mild cognitive impairment due to AD; MCI-NN, mild cognitive impairment without evidence of neurodegeneration; FRA1, Free Recall trial 1 list A; CRA1, Cued Recall list A trial 1; CRA2, Cued Recall list A trial 2; FRB1, Free Recall List B trial 1; CRB1, Cued Recall List B trial 2; CRB2, Cued Recall List B trial 2; SdFRA, Short-Delay Free Recall List A; SdCRA, Short-Delay Cued Recall List A; DR, Delayed Recall.

Random Forest classification between AD and non-AD groups

The resulting model obtained an accuracy of 0.870, F1-score of 0.865, precision of 0.866, recall of 0.870, and area under the curve of 0.890. The three most important scores were: DR (mean decrease in accuracy 0.052), SdFRA (0.027), and CRB2 (0.022). Confusion matrix is shown in Supplementary Table 1.

Neuropsychological predictors of failure to recover from semantic interference (frPSI)

For the prediction of CRB2, precision of the models (defined as the number of true positives divided by the number of true positives plus de number of false positives) was 64.1%. Importance of the neuropsychological tests for the prediction were: FCSRT total free recall (0.14), FCSRT total recall (0.11), FCSRT delayed total recall (0.09), FCSRT delayed free recall (0.09), SDMT (0.08), BNT (0.05), Stroop part A (0.05), Stroop part C (0.05), Stroop part B (0.04), and TMT-A (0.04).

For the prediction of SdCRA, precision of the models was 44.8%. Importance of the neuropsychological tests for the prediction were: FCSRT delayed free recall (0.15), total recall (0.14), FCSRT free recall trial 1 (0.09), FCSRT delayed total recall (0.09), FCSRT total recall (0.09), BNT (0.04), Stroop part C (0.04), Corsi forward (0.04), ROCF recognition memory (0.03), and Stroop part A (0.03).

Voxel-based mapping analysis of semantic interference

We examined the correlations of CRB2 and SdCRA. CRB2 was positively correlated with bilateral precuneus, superior, middle and inferior temporal gyri, fusiform, angular, superior and inferior parietal lobule, superior, middle and inferior occipital gyri, lingual, and posterior cingulate (Fig. 3).

Fig. 3

Whole-brain FDG-PET analysis. Multiple regression analysis showing brain regions associated with CRB2 in blue. SPM maps are displayed on an MRI template (neurological orientation, axial view) (p-value <0.001 and FWE-cluster level <0.05).

There were no statistically significant clusters for the correlation between (CRB2-CRB1) index for the prespecified threshold. We conducted an exploratory analysis reducing the statistical threshold to uncorrected p-value <0.005 and k=200 voxels, the (CRB2-CRB1) index was negatively correlated with a cluster involving the right middle temporal gyrus and the right fusiform gyrus.

Similarly, SdCRA was correlated with bilateral precuneus, superior, middle, and inferior temporal gyri, fusiform, angular and supramarginal gyri, superior and inferior parietal lobules, lingual gyrus, and superior, middle, and inferior occipital gyri (Fig. 4). Complete statistics about voxel-based analysis are depicted in Supplementary Table 2.

Fig. 4

Whole-brain FDG-PET analysis. Multiple regression analysis showing brain regions associated with SdCRA in green. SPM maps are displayed on an MRI template (neurological orientation, axial view) (p-value <0.001 and FWE-cluster level <0.05).

Metabolic connectivity

Weighted correlation matrices are shown in Fig. 5. Visual analysis of the matrices showed that the group with impaired CRB2 showed lower correlations, indicating a lower connectivity weight in the networks involving the parietal (precuneus, inferior and parietal lobules) and temporal lobes (middle, inferior temporal gyri). The group with frPSI (impaired CRB2) displayed a decrease in nodal degree and closeness centrality in the posterior cingulate. There were no statistically significant changes in the global measures (global efficiency, characteristic path length, clustering, and modularity) (Fig. 6).

Fig. 5

Weighted correlation matrices for the groups with normal (left) and very low recovery (right) from proactive semantic interference.

Fig. 6

Comparison between high and low performance in CRB2 in nodal graph measures. A) Node degree; B) Closeness centrality. Network densities are shown on the x-axis (from 5% to 35%, in steps of 1%). Differences between groups are shown on the y-axis (blue line with red points). The 95% confidence intervals (in violet) are shown at p<0.05 (two-tailed) FDR-corrected.

DISCUSSION

The aim of our study was to disentangle the neuropsychological mechanisms and neural basis of frPSI and RSI. To this end, we studied a large cohort of patients that underwent a comprehensive cognitive examination and FDG-PET imaging, and were assessed with the LASSI-L.

We found statistically significant differences in all the LASSI-L scores with large effect sizes. These findings are noteworthy, as all patients were examined because of memory issues but with no functional impairment in daily living activities. All scores showed differences between MCI-AD and SMC, which is consistent with previous research [5]. Interestingly, scores associated with proactive semantic interference, RSI and delayed recall were impaired in Pre-MCI-AD compared to SMC. Pre-MCI-AD also showed higher scores than MCI-AD, which suggests that LASSI-L could also be useful for monitoring in the earliest stages of AD [49]. In addition, the early impairment of RSI and the absence of differences between Pre-MCI and MCI-AD suggests that this could also be an early marker. The role of RSI in the detection of the early stages has been recently suggested in a study in offspring of patients with late-onset AD, in which RSI was correlated with amyloid deposition in the left temporal lobe measured by PiB-PET [26]. Another interesting finding was the absence of differences between MCI-NN and SMC-NN on LASSI-L measures. Since the MCI-NN group was diagnosed according to the results of the FCSRT, but no signs of AD or neurodegeneration were observed according to CSF biomarkers and neuropsychological follow-up, this suggests that this group could have been erroneously classified as impaired according to the FCSRT or that the initial low FCSRT score at baseline represented a transitory phenomenon other than AD. Interestingly, discrepancies between FCSRT and LASSI-L were also found in a previous study, in which the LASSI-L outperformed the results of the FCSRT [6]. The greater facilitation of learning in the LASSI-L (e.g., identification as a memory task involving three-semantic categories before the test administration, and use of more prototypical semantically related words on competing word lists could explain the higher specificity of the LASSI-L [6]. However, further studies with a face-to-face comparison between memory tests are needed to evaluate the pros and cons of each one.

Another remarkable finding of our study is the discrimination between AD and non-AD groups. The different metrics (e.g., accuracy, F1-score, etc.) showed a good classification capacity of the LASSI-L scores. These values are especially notable considering that all patients consulted by memory loss. In addition, the most important variables in the classification included DR, SdFRA, and CRB2. DR is one of the best scores for the discrimination because several memory processes are necessary to achieve adequate performance. Learning during the LASSI-L is performed under the effects of proactive interference, and recall is also sensitive to retroactive interference. Thus, these processes also play a role in DR. However, previous studies have shown still good discrimination capacity by omitting the delayed recall, which could be interesting in clinical settings to reduce the administration time from 30 to 8 min [50].

The automatic linear analysis identified the other memory tests (FCSRT, ROCF) as the best predictors of the performance on CRB2 and SdCRA. In addition, the importance of memory tasks was greater than attention tests. Specifically, Stroop test, which is associated with inhibition capacity [51], had lower importance than memory scores. Although both CRB2 and SdCRA were mainly associated with FCSRT scores, total free recall showed more importance in predicting CRB2 score, than delayed free recall in SdCRA. This suggests a more important role of learning and storage difficulties after previous exposure to a semantically similar target list in CRB2 performance/frPSI. In contrast, the delayed recall was more predictive for SdCRA, which could be more associated with accelerated forgetting. The greater influence of memory abilities than attention and executive functioning suggests that the paradigm used in the LASSI-L may reduce the capacity of the attention system and working memory to compensate for the memory deficits [52]. This supports the conception of the semantically competing target lists employed by the LASSI-L as a “stress test” which taxes the cognitive system due to demands on source memory, attentional systems and working memory and the inhibition of previously learned material [1].

The lower importance of attention and executive functioning is consistent with the results obtained from the administration of the LASSI-L to patients with multiple sclerosis [53]. In these patients, that show a cognitive profile mainly characterized by the attention and executive deficits [54], frPSI was not prominently impaired as in AD. However, we cannot exclude that executive function deficits could be more relevant in the frPSI in other disorders or that measures such as the Stroop-Color Word Test, Trails B and Tower of London do not fully capture all the complexities of executive function. Other attention and executive tests such as the Hayling test (as a measure of inhibition capacity), N-Back (for working memory and concentration), or dual tasks paradigms may be interesting to correlate with the LASSI-L scores in future works [55–57]. In this regard, studies using LASSI-L in other settings showing different underlying characteristics of memory dysfunction (e.g., behavioral variant frontotemporal dementia, Parkinson’s disease) may be interesting to understand further the neural basis of LASSI-L and the cognitive processes involved in the test performance [58].

The association between both CRB2 and SdCRA with the bilateral temporoparietal lobes is particularly relevant because these regions are characteristically impaired in the early stages of AD. Temporoparietal hypometabolism is the most frequent finding in FDG-PET imaging in patients with AD presenting with memory loss [27]. The medial and posterior temporal lobe, precuneus, posterior cingulate and parietal lobules are regions consistently associated with episodic memory in functional neuroimaging studies using PET or fMRI [59]. These regions are interconnected with each other and with the entorhinal cortex and the posterior hippocampus, contributing to episodic memory and recall. Network analysis also revealed interesting findings. After multiple comparisons corrections, all the statistically significant findings were detected in nodal measures, which suggest the global preservation of connectivity in these patients due to the early stages of the disease. Regarding nodal measures, node degree (a measure of the number of connections from each node to the rest of the network) was decreased specifically in the posterior cingulate. Similarly, closeness centrality was also reduced in the same region. Topological centrality refers to the capacity of a node to influence other elements of the network based on the characteristics of the connections [60]. These findings suggest that the influence on brain function of the posterior cingulate was altered in those patients with frPSI, suggesting a key role of posterior cingulate in the pathophysiology of the frPSI. The posterior cingulate is a region highly connected and a central part of the default mode network. Studies in AD has detected posterior cingulate involvement in the earliest stages of the disease. Amyloid deposition and hypometabolism are prominent in this region. In addition, default mode network connectivity alterations may be found in young people carrying the APOE ɛ4 genotype [61]. Our network analysis findings and voxel-based brain mapping suggest a central role of posterior cingulate in frPSI, with the participation of other important regions of the temporal and parietal lobes. In this regard, precuneus and posterior cingulate are strongly connected with posterior hippocampus [62], which has also been previously associated with RI [20]. Overall, this is consistent with the view that the posterior cingulate links neocortical areas with the middle temporal memory system [63]. During AD progression, this system is gradually damaged, which causes insidious but inexorable memory degradation. The correlation between LASSI-L scores with early regions involved in AD (posterior cingulate, precuneus, etc.) and many other regions in the parieto-temporal lobes suggest that the test may be useful to detect the first stages but also to monitor the memory decline.

Our study has several strengths. First, our patients were recruited from our Department of Neurology, which has direct and easy access from primary care [64]. This suggests that our findings are representative for patients with memory consulting with complaints. Second, LASSI-L was not used for the diagnosis, which prevents the circularity of the analysis. Third, FDG-PET was analyzed with a voxel-based approach, which is not biased to any specific structure and permits a comprehensive assessment of all brain regions. In this regard, we used a strict correction for multiple comparisons for PET analysis. Forth, PET connectivity analyses were also conducted, which may provide a complementary and novel information to further understand brain-behavior relationships. However, we should also acknowledge some limitations. On the one hand, the group MCI-NN has a small sample size, which increases the risk of type II errors. Second, we only focused on the main scores of the LASSI-L (based on the number of correct words). Although these scores are used to evaluate some cognitive processes (e.g., CRB2 for recovery from proactive interference, SdFRA and SdCRA for retroactive interference), these scores are not independent. In this regard, encoding and retrieval processes are needed for the optimal performance of the test. Thus, the LASSI-L scores do not provide a totally specific measure of each memory process. However, these scores are at least a proxy for assessing episodic memory under proactive and retroactive semantic interference conditions and the recovery from these effects. We tried to avoid this limitation by evaluating the index (CRB2 – CRB1) as a measure of the recovery from proactive semantic interference. This index showed lower values in MCI-AD than in the other groups, but it does not reach the statistical significance as in previous studies comparing with controls [5]. Similarly, an exploratory analysis uncorrected for multiple comparisons showed a correlation between this index and temporal lobe metabolism. Overall, these findings seem to support the predominant role of temporal regions in the frPSI in early stages of AD. Future studies with larger sample sizes and longitudinal data could be of interest to detect patients with specific impairment of each score or combination of scores. This type of analysis, which has been recently performed in the picture version of the FCSRT with the Stages of Objective Memory Impairment system [65], could be useful to evaluate the dynamics of the memory loss in the AD continuum and may be a good model to corroborate the neural underpinnings found for the LASSI-L in our study [66, 67]. Third, the assessment of intrusions generated during proactive semantic interference, frPSI and RSI were not evaluated. They may be of interest in future works given their relationship to amyloid deposition as well as progression over time [68, 69]. Finally, the diagnostic group classification was supported by several biomarkers and the follow-up, but amyloid biomarkers were available only in a subset of patients. In addition, in this study we did not include a healthy control group, which would be useful to evaluate the utility of the LASSI-L for the characterization of the groups SMC-NN and MCI-NN.

In conclusion, our study suggests that frPSI and RSI are related to the dysfunction of the temporoparietal lobes, specifically the medial temporal structures, precuneus, and posterior cingulate. Episodic memory dysfunction is the main neurocognitive mechanism underlying these phenomena, which probably reflects the difficulties in the learning and storage of new information. Conversely, inhibition and other attention/executive functioning subdomains, and frontal lobes, seems to have a lesser role. These findings support the role of the LASSI-L in the detection, monitoring and outcome prediction in the early stages of AD.

Footnotes

ACKNOWLEDGMENTS

JAMG is supported by Instituto de Salud Carlos III through the project INT20/00079 (co-funded by European Regional Development Fund A way to make Europe). MVS is supported by Instituto de Salud Carlos III through a predoctoral contract (FI20/000145) (co-funded by European Regional Development Fund “A way to make Europe”). DAL and RCC are supported by NIH-NIA through the projects 1R01AG061106 and 1R01AG055638, respectively.

Authors’ disclosures available online ().

The supplementary material is available in the electronic version of this article: .

References

Loewenstein

, Curiel

, Duara

, Buschke

(2018) Novel cognitive paradigms for the detection of memory impairment in preclinical Alzheimer’s disease. Assessment 25, 348–359.

Loewenstein

, Acevedo

, Schram

, Ownby

, White

, Mogosky

, Barker

, Duara

(2003) Semantic interference in mild Alzheimer’s disease: Preliminary findings. Am J Geriatr Psychiatry 11, 252–255.

Crocco

, Curiel

, Acevedo

, Czaja

, Loewenstein

(2014) An evaluation of deficits in semantic cueing and proactive and retroactive interference as early features of Alzheimer’s disease. Am J Geriatr Psychiatry 22, 889–897.

Loewenstein

, Curiel

, Greig

, Bauer

, Rosado

, Bowers

, Wicklund

, Crocco

, Pontecorvo

, Joshi

, Rodriguez

, Barker

, Hidalgo

, Duara

(2016) A novel cognitive stress test for the detection of preclinical Alzheimer’s disease: Discriminative properties and relation to amyloid load. Am J Geriatr Psychiatry 24, 804–813.

Matias-Guiu

, Curiel

, Rognoni

, Valles-Salgado

, Fernandez-Matarrubia

, Hariramani

, Fernandez-Castro

, Moreno-Ramos

, Loewenstein

, Matias-Guiu

(2017) Validation of the Spanish version of the LASSI-L for diagnosing mild cognitive impairment and Alzheimer’s disease. J Alzheimers Dis 56, 733–742.

Matias-Guiu

, Cabrera-Martín

, Curiel

, Valles-Salgado

, Rognoni

, Moreno-Ramos

, Carreras

, Loewenstein

, Matias-Guiu

(2018) Comparison between FCSRT and LASSI-L to detect early stage Alzheimer’s disease. J Alzheimers Dis 61, 103–111.

Sanchez

, Abulafia

, Duarte-Abritta

, de Guevara

MSL

, Castro

, Drucaroff

, Sevlever

, Nemeroff

, Vigo

, Loewenstein

, Villarreal

, Guinjoan

(2017) Failure to recover from proactive semantic interference and abnormal limbic connectivity in asymptomatic, middle-aged offspring of patients with late-onset Alzheimer’s disease. J Alzheimers Dis 60, 1183–1193.

Curiel

Cid RE

, Loewenstein

, Roselli

, Matias-Guiu

, Piña

, Adjouadi

, Cabrerizo

, Bauer

, Chan

, DeKosky

, Golde

, Greig-Custo

, Lizarraga

, Peñate

, Duara

(2019) A cognitive stress test for prodromal Alzheimer’s disease: Multiethnic generalizability. Alzheimers Dement (Amst) 11, 550–559.

Capp

, Curiel Cid

, Crocco

, Stripling

, Kitaigorodsky

, Sierra

, Melo

, Loewenstein

(2020) Semantic intrusion error ratio distinguishes between cognitively impaired and cognitively intact African American older adults. J Alzheimers Dis 73, 785–790.

10.

Friedman

, Miyake

(2004) The relations among inhibition and interference control functions: A latent-variable analysis. J Exp Psychol Gen 133, 101–135.

11.

Badre

, Wagner

(2007) Left ventrolateral prefrontal cortex and the cognitive control of memory. . Neuropsychologia 45, 2883–2901.

12.

Corbett

, Duarte

(2020) How proactive interference during new associative learning impacts general and specific memory in young and old. J Cogn Neurosci 32, 1607–1623.

13.

Lustig

, Jantz

(2015) Questions of age differences in interference control: When and how, not if? Brain Res 1612, 59–69.

14.

Archambeau

, Forstmann

, Van Maanen

, Gevers

(2020) Proactive interference in aging: A model-based study. Psychon Bull Rev 27, 130–138.

15.

Wickens

(1970) Encoding categories of words: An empirical approach to meaning. Psychol Rev 77, 1–15.

16.

Kiegl

, Bäuml

KHT

(2021) Buildup and release from proactive interference: Cognitive and neural mechanisms. Neurosci Biobehav Rev 120, 264–278.

17.

Cushamn

, Como

, Booth

, Caine

(1988) Cued recall and release from proactive interference in Alzheimer’s disease. J Clin Exp Neuropsychol 10, 685–692.

18.

Binetti

, Magni

, Padovani

, Cappa

, Bianchetti

, Trabucchi

(1995) Release from proactive interference in early Alzheimer’s disease. Neuropsychologia 33, 379–384.

19.

Loewenstein

, Acevedo

, Luis

, Crum

, Barker

, Duara

(2004) Semantic interference deficits and the detection of mild Alzheimer’s disease and mild cognitive impairment without dementia. J Int Neuropsychol Soc 10, 91–100.

20.

Wagner

, Gussew

, Köhler

, de la Cruz

, Smesny

, Reichenbach

, Bär

(2016) Resting state functional connectivity of the hippocampus along the anterior-posterior axis and its association with glutamatergic metabolism. Cortex 81, 104–117.

21.

Curiel

, Crocco

, Acevedo

, Duara

, Agron

, Loewenstein

(2013) A new scale for the evaluation of proactive and retroactiveinterference in mild cognitive impairment and early Alzheimer’sdisease. Aging Sci 1, 102.

22.

Loewenstein

, Curiel

, Wright

, Sun

, Alperin

, Crocco

, Czaja

, Raffo

, Penate

, Melo

, Capp

, Gamez

, Duara

(2017) Recovery from proactive semantic interference in mild cognitive impairment and normal aging: Relationship to atrophy in brain regions vulnerable to Alzheimer’s disease. J Alzheimers Dis 56, 1119–1126.

23.

Loewenstein

, Curiel

, DeKosky

, Rosselli

, Bauer

, Grieg-Custo

, Penate

, Li

, Lizarraga

, Golde

, Adjouadi

, Duara

(2017) Recovery from proactive semantic interference and MRI volume: A replication and extension study. J Alzheimers Dis 59, 131–139.

24.

Curiel Cid

, Crocco

, Duara

, Garcia

, Rosselli

, DeKosky

, Smith

, Bauer

, Chirinos

, Ajdouadi

, Barker

, Loewenstein

(2020) A novel method of evaluating semantic intrusion errors todistinguish between amyloid positive and negative groups on theAlzheimer’s disease continuum.. J Psychiatr Res 124, 131–136.

25.

Crocco

, Loewenstein

, Curiel

, Alperin

, Czaja

, Harvey

, Sun

, Lenchus

, Raffo

, Peñate

, Melo

, Sang

, Valdivia

, Cardenas

(2018) A novel cognitive assessment paradigm to detect pre-mild cognitive impairment (PreMCI) and the relationship to biological markers of Alzheimer’s disease. J Psychiatr Res 96, 33–38.

26.

Abulafia

, Loewenstein

, Curiel-Cid

, Duarte-Abritta

, Sanchez

, Vigo

, Castro

, Drucaroff

, Vazquez

, Sevlever

, Nemeroff

, Guinjoan

, Villarreal

(2019) Brain structural and amyloid correlates of recovery from semantic interference in cognitively normal individuals with or without family history of late-onset Alzheimer’s disease. J Neuropsychiatry Clin Neurosci 31, 25–36.

27.

Chetelat

, Arbizu

, Barthel

, Garibotto

, Law

, Morbelli

, van de Giessen

, Agosta

, Barkhof

, Brooks

, Carrillo

, Dubois

, Fjell

, Frisoni

, Hansson

, Herholz

, Hutton

, Jack

Jr , Lammerstma

, Landau

, Minoshima

, Nobili

, Nordberg

, Ossenkoppele

, Oyen

WJG

, Perani

, Rabinovici

, Scheltens

, Villemagne

, Zetterberg

, Drzezga

(2020) Amyloid-PET and 18F-FDG-PET in the diagnostic investigation of Alzheimer’s disease and other dementias. Lancet Neurol 19, 951–962.

28.

Strom

, Iaccarino

, Edwards

, Lesman-Segev

, Soleimani-Meigooni

, Pham

, Baker

, Landau

, Jaugust

, Miller

, Rosen

, Gorno-Tempini

, Rabinovici

, La Joie

; The Alzheimer’s Disease Neuroimaging Initiative (2021) Cortical hypometabolism reflects local atrophy and tau pathology in symptomatic Alzheimer’s disease. Brain 145, 713–728.

29.

Matias-Guiu

, Cabrera-Martín

, Valles-Salgado

, Perez-Perez

, Rognoni

, Moreno-Ramos

, Carreras

, Matías-Guiu

(2017) Neural basis of cognitive assessment in Alzheimer disease, amnestic mild cognitive impairment, and subjective memory complaints. Am J Geriatr Psychiatry 25, 730–740.

30.

Yakushev

, Drzezga

, Habeck

(2017) Metabolic connectivity: Methods and applications. Curr Opin Neurol 30, 677–685.

31.

Sala

, Perani

(2019) Brain molecular connectivity in neurodegenerative diseases: Recent advances and new perspectives using positron emission tomography. Front Neurosci 13, 617.

32.

Grober

, Hall

, Lipton

, Zonderman

, Resnick

, Kawas

(2008) Memory impairment, executive dysfunction, and intellectual decline in preclinical Alzheimer’s disease. J Int Neuropsychol Soc 14, 266–278.

33.

, Nation

, Alzheimer’s Disease Neuroimaging Initiative (2018) Neuropsychological profiles and trajectories in preclinical Alzheimer’s disease. J Int Neuropsychol Soc 24, 693–702.

34.

Mickes

, Wixted

, Fennema-Notestine

, Galasko

, Bondi

, Thal

, Salmon

(2007) Progressive impairment on neuropsychological tasks in a longitudinal study of preclinical Alzheimer’s disease. Neuropsychology 21, 696–705.

35.

Medina

, Woo

, Rodriguez-Agudelo

, Chaparro Maldonado

, Yi

, Coppola

, Zhou

, Chui

, Ringman

(2021) Reaction time andresponse inhibition in autosomal dominant Alzheimer’s disease. Brain Cogn 147, 105656.

36.

Peña-Casanova

, Blesa

, Aguilar

, Gramunt-Fombuena

, Gomez-Anson

, Oliva

, Molinuevo

, Robles

, Barquero

, Antúnez

, Martínez-Parra

, Frank-García

, Fernandez

, Alfonso

, Sol

(2009) Spanish Multicenter Normative Studies (NEURONORMA Project): Methods and sample characteristics. Arch Clin Neuropsychol 24, 307–319.

37.

Sanchez-Benavides

, Peña-Casanova

, Casals-Coll

, Gramunt

, Molinuevo

, Gómez-Ansón

, Aguilar

, Robles

, Antúnez

, Martínez-Parra

, Frank-García

, Fernandez-Martínez

, Blesa

(2014) Cognitive and neuroimaging profiles in mild cognitive impairment and Alzheimer’s disease: Data from the Spanish Multicenter Normative Studies (NEURONORMA Project). J Alzheimers Dis 41, 887–901.

38.

Peña-Casanova

, Gramunt-Fombuena

, Quiñones-Ubeda

, Sanchez-Benavides

, Aguilar

, Badenes

, Molinuevo

, Robles

, Barquero

, Payno

, Antúnez

, Martínez-Parra

, Frank-García

, Fernandez

, Alfonso

, Sol

, Blesa

;NEURONORMA Study Team (2009) Spanish Multicenter Normative Studies (NEURONORMA Project): Norms for the Rey-Osterrieth complex figure (copy and memory), and free and cued selective reminding test. Arch Clin Neuropsychol 24, 371–393.

39.

Peña-Casanova

, Sanchez-Benavides

, de Sola

, Manero-Borras

, Casals-Coll

(2012) Neuropsychology of Alzheimer’s disease. Arch Med Res 43, 686–693.

40.

Dubois

, Feldman

, Jacova

, Hampel

, Molinuevo

, Blennow

, DeKosky

, Gauthier

, Selkoe

, Bateman

, Cappa

, Crutch

, Engelborghs

, Frisoni

, Fox

, Galasko

, Habert

, Jicha

, Nordberg

, Pasquier

, Rabinovici

, Robert

, Rowe

, Salloway

, Sarazin

, Epelbaum

, de Souza

, Vellas

, Visser

, Schenider

, Stern

, Scheltens

, Cummings

(2014) Advancing researchdiagnostic criteria for Alzheimer’s disease: The IWG-2 criteria. Lancet Neurol 13, 614–629.

41.

American Psychiatric Association (2013) Diagnostic and statistical manual of mental disorders (5th ed.).

42.

Varrone

, Asenbaum

, Vander Borght

, Booij

, Nobili

, Nagren

, Darcourt

, Kapucu

, Tatsh

, Bartenstein

, Van Laere

; European Association of Nuclear Medicine Neuroimaging Committee (2009) EANM Procedure guidelines for PET brain using [18F]FDG, version 2. Eur J Nucl Med Mol Imaging 36, 2103–2110.

43.

Della Rosa

, Cerami

, Gallivanone

, Prestia

, Caroli

, Castiglioni

, Gilardi

, Frisoni

, Friston

, Ashburner

, Perani

, EADC-PET Consortium (2014) A standardized [18F]-FDG-PET template for spatial normalization in statistical parametric mapping of dementia. Neuroinformatics 12, 575–593.

44.

Tzourio-Mayozer

, Landeau

, Papathanassiou

, Crivello

, Etard

, Delcroix

, Mazoyer

, Joliot

(2002) Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain. Neuroimage 15, 273–289.

45.

Mijalkov

, Kakaei

, Pereira

, Westman

, Volpe

(2017) BRAPH: A graph theory software for the analysis of brain connectivity. PLoS One 12, e0178798.

46.

Yang

(2013) The case for being automatic: Introducing the automatic linear modeling (LINEAR) procedure in SPSS statistics. Multiple Linear Regression Viewpoints 39, 27–37.

47.

Flanagan

, Lagarde

, Hahn

, Guichart-Gomez

, Sarazin

, Hornberger

, Bertoux

(2018) Executive and social-cognitive determinants of environmental dependency syndrome in behavioral frontotemporal dementia. Neuropsychology 32, 377–384.

48.

Delgado-Álvarez

, Cabrera-Martín

, Pytel

, Delgado-Alonso

, Matías-Guiu

(2021) Design and verbal fluency in Alzheimer’s disease and frontotemporal dementia: Clinical and metabolic correlates. J Int Neuropsychol Soc, doi: 10.1017/S1355617721001144.

49.

Kitaigorodsky

, Curiel Cid

, Crocco

, Gorman

, Gonzalez-Jimenez

, Greig-Custo

, Barker

, Duara

, Loewenstein

(2021) Changes in LASSI-L performance over time among older adults with amnestic MCI and amyloid positivity: A preliminary study. J Psychiatr Res 143, 98–105.

50.

Crocco

, Curiel-Cid

, Kitaigorodsky

, Greig-Custo

, Zheng

, Duara

, Loewenstein

(2020) A brief version of the LASSI-L detects prodromal Alzheimer’s disease states. J Alzheimers Dis 78, 789–799.

51.

Matias-Guiu

, Cabrera-Martín

, Valles-Salgado

, Rognoni

, Galan

, Moreno-Ramos

, Carreras

, Matías-Guiu

(2019) Inhibition impairment in frontotemporal dementia, amyotrophic lateral sclerosis, and Alzheimer’s disease: Clinical assessment and metabolic correlates. Brain Imaging Behav 13, 651–659.

52.

Darby

, Brickhouse

, Wolk

, Dickerson

, Alzheimer’s Disease Neuroimaging Initiative (2017) Effects of cognitive reserve depend on executive and semantic demands of the task. J Neurol Neurosurg Psychiatry 88, 794–802.

53.

Matias-Guiu

, Cortes-Martínez

, Curiel

, Delgado-Álvarez

, Fernandez-Oliveira

, Pytel

, Montero

, Moreno-Ramos

, Loewenstein

, Matías-Guiu

(2020) Memory impairment in relapsing-remitting multiple sclerosis using a challenging semantic interference task. Front Neurol 11, 309.

54.

Matias-Guiu

, Cortes-Martínez

, Valles-Salgado

, Oreja-Guevara

, Pytel

, Montero

, Moreno-Ramos

, Matias-Guiu

(2017) Functional components of cognitive impairment in multiple sclerosis: A cross-sectional investigation. Front Neurol 68, 643.

55.

Foley

, Kaschel

, Logie

, Della Sala

(2011) Dual-task performance in Alzheimer’s disease, mild cognitive impairment, and normal ageing. Arch Clin Neuropsychol 26, 340–348.

56.

Perez-Perez

, Matías-Guiu

, Caceres-Guillen

, Rognoni

, Valles-Salgado

, Fernandez-Matarrubia

, Moreno-Ramos

, Matías-Guiu

(2016) The Hayling Test: Development and normalization of the Spanish version. Arch Clin Neuropsychol 31, 411–419.

57.

Stark

, Avery

, Rosenberg

, Greene

, Gao

, Scheinost

, Constable

, Chun

, Yoo

(2021) Using functional connectivity models to characterize relationships between working and episodic memory. Brain Behav 11, e02105.

58.

Fernandez-Matarrubia

, Matías-Guiu

, Cabrera-Martín

, Moreno-Ramos

, Valles-Salgado

, Carreras

, Matías-Guiu

(2017) Episodic memory dysfunction in behavioral variant frontotemporal dementia: A clinical and FDG-PET study. J Alzheimers Dis 57, 1251–1264.

59.

Cabeza

, Nyberg

(2000) Imaging cognition II: An empirical review of 275 PET and fMRI studies. J Cogn Neurosci 12, 1–47.

60.

Fornito

, Zalesky

, Bullmore

(2016) Fundamentals of brain network analysis. Elsevier Academic Press.

61.

Leech

Sharp

(2014) The role of the posterior cingulate in cognition and disease. Brain 137, 12–32.

62.

Strange

, Witter

, Lein

, Moser

(2014) Functional organization of the hippocampal longitudinal axis. Nat Rev Neurosci 15, 655–669.

63.

Rolls

(2019) The cingulate cortex and limbic systems for emotion, action, and memory. Brain Struct Funct 224, 3001–3018.

64.

Matias-Guiu

, García-Azorín

, García-Ramos

, Basoco

, Elvira

, Matias-Guiu

(2015) Study of outpatient neurological care in the Region of Madrid: The impact of implementing free choice of hospital. Neurologia 30, 479–487.

65.

Grober

, Veroff

, Lipton

(2018) Temporal unfolding of declining episodic memory on the Free and Cued Selective Reminding Test in the predementia phase of Alzheimer’s disease: Implications for clinical trials. Alzheimers Dement (Amst) 10, 161–171.

66.

Grober

, Papp

, Rentz

, Sperling

, Johnson

, Amariglio

, Schultz

, Lipton

, Ezzati

(2021) Neuroimaging correlates of Stages of Objective Memory Impairment (SOMI) system. Alzheimers Dement (Amst) 13, e12224.

67.

Grober

, Lipton

, Sperling

, Papp

, Johnson

, Rentz

, Veroff

, Aisen

, Ezzati

(2022) Associations of stages of objective memory impairment with amyloid PET and structural MRI: The A4 Study. Neurology 98, e1327–1336.

68.

Torres

, Rosselli

, Loewenstein

, Curiel

, Velez Uribe

, Lang

, Arruda

, Penate

, Vaillancourt

, Greig

, Barker

, Bauer

, Duara

(2019) Types of errors on a semantic interference task in mild cognitive impairment and dementia. Neuropsychology 33, 670–684.

69.

Zheng

, Curiel Cid

, Duara

, Kitaigorodsky

, Crocco

, Loewenstein

(2021) Semantic intrusion errors as a function of age, amyloid, and volumetric loss: A confirmatory path analysis. Int Psychogeriatr, doi: 10.1017/S1041610220004007.