A Brief Version of the LASSI-L Detects Prodromal Alzheimer’s Disease States

Abstract

Background:

The Loewenstein-Acevedo Scales for Semantic Interference and Learning (LASSI-L) is an increasingly utilized cognitive stress test designed to identify early cognitive changes associated with incipient neurodegenerative disease.

Objective:

To examine previously derived cut-points for cognitively unimpaired older adults that were suggestive of performance impairment on multiple subscales of the LASSI-L. These cut-points were applied to a new sample of older adults who were cognitive healthy controls (HC: n = 26) and those on the Alzheimer’s disease (AD) continuum from early stage mild cognitive impairment (EMCI: n = 28), late stage MCI (LMCI: n = 18) to mild AD (AD: n = 27).

Methods:

All participants were administered the LASSI-L. All cognitively impaired participants were PET amyloid positive which likely reflects underlying AD neuropathology, while cognitively normal counterparts were deemed to have amyloid negative scans.

Results:

There was a monotonic relationship between the number of deficits on LASSI-L subscales and independent classification of study groups with greater severity of cognitive impairment. Importantly, taken together, impairment on maximum learning ability and measures of proactive semantic interference (both reflected by cued recall and intrusion errors) correctly classified 74.1% of EMCI, 94.4% of LMCI, and 96.3% of AD. Only 7.7% of HC were incorrectly classified as having impairments.

Conclusion:

A modest number of LASSI-L subscales taking approximately 8 minutes to administer, had excellent discriminative ability using established cut-offs among individuals with presumptive stages of AD. This has potential implications for both clinical practice and clinical research settings targeting AD during early prodromal stages.

Keywords

Amyloid biomarkers cognitive screening intrusion errors mild cognitive impairment neuroimaging preclinical Alzheimer’s disease semantic interference structural MRI

INTRODUCTION

There are a number of cognitive measures that have been employed to assess amnestic mild cognitive impairment (aMCI) and dementia. However, there is a considerably more limited number of cognitive indices that have shown sensitivity to the earliest stages of preclinical Alzheimer’s disease (AD) which have been also been specifically related to AD biomarkers such as amyloid-β (Aβ), or other AD patterns of neurodegeneration on magnetic resonance imaging MRI [1 –3].

Moreover, emerging AD trials aimed at early intervention may encounter limitations in using traditional cognitive assessment paradigms and tools, such as the ADAS-Cog [4 –6] developed for dementia clinical trials, because even the addition of additional subtests, these instruments may be not be adequately sensitive to and not specific to the early cognitive deficits associated with AD [4 –8]. In fact, Kueper et al. (2018) points to other metrics such as difficulties with susceptibility to proactive interference on competing semantically related word lists as early indicators of early AD which may not be picked up on the ADAS-Cog [5].

Further studies and reviews have pointed to potential limitations of traditional neuropsychological testing to detect the earliest cognitive changes associated with preclinical AD [1 , 9]. Proactive semantic interference (PSI) has shown to be a sensitive indicator of early AD and has a number of postulated mechanisms in the brain [10 –12].

Loewenstein and colleagues have found that specific breakdowns associated with “cognitive stress tests” measuring PSI as well as the failure to recover from semantic interference (frPSI) when the new competing list is presented a second time for recall on the Loewenstein-Acevedo Scales for Semantic Interference and Learning (LASSI-L) represents a cognitive marker of preclinical AD [13, 14]. Subsequent studies found that performance deficits on the LASSI-L have proven superior to several traditionally used memory tests (e.g., list learning measures, delayed paragraph recall), in studies that relate these early cognitive changes to biological markers of AD such as in vivo amyloid imaging and neurodegeneration measured by MRI, functional MRI, and fluorodeoxyglucose positron emission tomography/ computed tomography (PET/CT). Using receiver operator curve (ROC) analyses, Matias-Guiu et al. [18] found that the LASSI-L was superior to the Free and Cued Selective Reminding Test (FCSRT), in detecting MCI patients with suspected AD. These studies have been using different cohorts in Miami [15, 16] as well as independent cohorts in Madrid [17, 18] as well as Buenos Aires [19].

In addition to number correct on cued recall trials susceptible to PSI, another actively studied cognitive deficit that has been associated with AD brain pathology includes semantic intrusion errors. In 2018, Loewenstein and colleagues [14] found that semantic intrusion errors occurring on the LASSI-L subscales that tap PSI and frPSI appeared to be particularly specific to AD. Semantic intrusion errors are much more prevalent in MCI patients with presumptive AD (amyloid positive on PET/CT imaging) versus those MCI who had suspected non-AD pathology (amyloid negative) or MCI due to other non-AD etiologies (e.g., diffuse Lewy body disease, frontotemporal degeneration, vascular cognitive impairment). In addition, the percentage of intrusion errors made in relation to total correct responses on LASSI-L subscales sensitive to PSI and frPSI could successfully distinguish amyloid positive from amyloid negative aMCI and dementia groups [20].

An advantage of the LASSI-L is that it provides a comprehensive evaluation of PSI, frPSI, vulnerability to retroactive interference (RSI) and delayed recall. The disadvantage is that a consideration of multiple subscales in a clinical or research setting using a specific cut-off for impairment, can give increasing rise to spurious errors of inference for persons who are actually unimpaired.

The current investigation examined whether using individual versus combined subscales of the LASSI-L provided more robust discrimination among individuals with normal cognition from those with presumptive AD in the early, MCI, late MCI, and early dementia stage. The study was designed to provide information as to the most effective subscales of the LASSI-L that could also reduce administration time.

METHODS

Participants

In the present investigation, we examined 99 participants enrolled in the 1Florida Alzheimer’s Disease Research Center (ADRC), aged 54 to 98 from the Clinical Core site, at Mount Sinai Medical Center, Miami Beach, Florida. The 1Florida ADRC baseline evaluation entailed extensive clinical and neuropsychological evaluations including the LASSI-L, a novel cognitive stress paradigm, which was not used for diagnostic classification. Participants were tested in their dominant and preferred language (English or Spanish) by trained examiners who were fluent in the language of administration. All participants underwent neuroimaging including MRI brain scans to assess regional brain volumes and amyloid PET/CT to assess for the presence of fibrillar amyloid plaques.

Diagnostic groups were classified using the following criteria:

Healthy cognitively normal group (HC; n = 26)

Participants were classified as HC if there were: 1) no subjective cognitive complaints made by the participant and/or a collateral informant; 2) no evidence by clinical evaluation that there was a history of memory or another cognitive decline after an extensive interview with the participant and their informant; 3) Global Clinical Dementia Rating (CDR) [21] scale score of 0; 4) all memory, e.g., Hopkins Verbal Learning Test, Revised (HVLT-R) [22] and delayed paragraph recall from the National Alzheimer’s Coordinating Center Uniform Data Set (NACC UDS) [23], and non-memory measures e.g., Category Fluency [24], Trails A and B [25], WAIS-IV Block Design subtest [26], were less than 1.0 SD below normal limits for age, education, and language group; 5) had a negative amyloid scan (Amy-) as read by an experienced rater (RD).

Early mild cognitive impairment with prodromal Alzheimer’s disease (EMCI; n = 28)

Participants in the EMCI group presented with the following: 1) fulfilled Petersen’s criteria for MCI [27], 2) subjective cognitive complaints reported by the participant and/or collateral informant; 3) clinical course and history consistent with AD; 4) Global CDR scale score of 0.5 and CDR sum of boxes ≤2.0; 5) impaired delayed recall (i.e., 1.5 SD or greater, below the mean, accounting for age, education, and language of testing) for either the HVLT-R or delayed paragraph recall from the NACC UDS and/or 1.5 SD below expected levels on non-memory measures; 6) no evidence of impairment in independent activities of daily living after extensive interview by the patient and the informant; 7) no evidence of other significant neuropathological findings on MRI scans, except neurodegeneration, assessed as atrophy in AD prone areas such as, the medial temporal lobes or parietal lobes; 8) amyloid scans read by an experienced rater (RD) as amyloid positive (Amy+).

Late mild cognitive impairment with prodromal Alzheimer’s disease (LMCI; n = 18)

Participants in the LMCI group presented with the following: 1) fulfilled Petersen’s criteria [27] for MCI; 2) subjective cognitive complaints reported by the participant and/or collateral informant; 3) clinical course and history consistent with AD; 4) Global CDR scale score of 0.5 and CDR sum of boxes ≥2.5; 5) impaired delayed recall (i.e., 1.5 SD or greater, below the mean, accounting for age, education, and language of testing) for either the HVLT-R and/or delayed paragraph recall from the NACC UDS and/or impaired performance below expectations on non-memory measures, 1.5 SD or greater, below the mean; 6) no evidence of impairment in independent activities of. daily living after extensive interview by the patient and the informant; 7) with no evidence of other significant neuropathological findings on MRI, such as atrophy of the medial temporal lobes or parietal lobes; 8) amyloid scans read by an experienced rater (RD) as Amy+.

Mild Alzheimer’s disease (AD; n = 27)

Participants in the AD group presented with the following: 1) fulfilled criteria for dementia [39]; 2) subjective cognitive complaints reported by the participant and/or collateral informant; 3) clinical course and history consistent with AD; 4) Global CDR scale score of 1.0; 5) impaired delayed recall (i.e., 1.5 SD or greater, below the mean, accounting for age, education, and language of testing) for either the HVLT-R or delayed paragraph recall from the NACC UDS and/or 1.5 SD below expected levels on non-memory measures as described for the cognitively normal (CN) group; 6) impairment in the ability to perform independent activities of daily living and functional decline reported by the informant; 7) no evidence of other significant neuropathological findings on MRI scans, except neurodegeneration, assessed as atrophy in typically visualized in AD prone areas, such as the medial temporal lobes or parietal lobes; 8) amyloid scans read by an experienced rater as Amy+.

Neuropsychological measure

Loewenstein-Acevedo Scales for Semantic Interference and Learning (LASSI-L)

The LASSI-L was not used for diagnostic determination in this study to avoid potential issues of circularity. This cognitive stress test represents a novel paradigm that employs controlled learning and cued recall in an effort to maximize the storage of a list of to-be-remembered target words belonging to three semantic categories (fruits, articles of clothing, and musical instruments [28]). Participants were tested in their preferred language (English or Spanish). The LASSI-L has been previously shown to be culturally fair and valid in either language [29]. The full LASSI-L that may take 20 or more minutes to administer.

During the administration of the LASSI-L, the examinee is instructed to remember a list of 15 common words, which is presented twice. After the second presentation, semantic cues prompt the recall of the target List A words, which is a measure of maximum storage capacity (Trial A2). A unique aspect of the LASSI-L paradigm is the presentation of a second competing list of to-be-remembered words that is presented in the same manner as the first list. That is, immediately following the second cued recall trial of List A, List B is presented. As in List A, there are two presentations of List B. The second list introduces different words with shared semantic categories, which elicits a considerable amount of proactive semantic interference (PSI as measured on Trial B1). Unlike other traditional memory paradigms, the second presentation and subsequent cued recall of this second list of words measures the failure to recover from the effects of proactive semantic interference (frPSI on Trial B2).

For the current study, we focused on the most sensitive subscales of the LASSI-L (i.e., number of intrusion errors on Cued B1 and Cued B2) that have been associated with AD neuropathology (amyloid load) and compared the percentage of individuals in each diagnostic group that made these errors. These intrusion errors primarily consist of words from the first list (List A) of semantically similar target items or other non-target words that share a similar semantic category. Intrusion errors produced on the Cued B1 and Cued B2 subscales are extremely sensitive to PSI and frPSI deficits thought to reflect deficits in source memory and inhibitory control [3 , 30]. We examined other LASSI-L subscales as part of secondary analyses.

Derivation of LASSI-L Cut-Points: We relied on previously derived LASSI-L cut-points that were established for a CN older adult group in a NIH-funded community-based study (Detect PAD; Loewenstein, PI) who were classified as CN by an extensive clinical interview, and normal performance on neuropsychological measures (memory and non-memory) at baseline. In addition, to be considered CN, these participants were consistently classified as CN during a two-to-three-year follow-up period.

Previous research has shown that when multiple memory tests or subtests within the same cognitive domain are examined as a group, spurious errors of inference may occur. For example, Loewenstein and colleagues [31] found that in CN older adults (determined by clinical evaluation and neuropsychological testing), 30% or more CN patients were erroneously classified as cognitively impaired when the results of 6 memory tests were considered, using a 1.5 SD cut-off for impairment. Thus, when evaluating a broad range of memory measures (i.e., six or more), Loewenstein and colleagues [31] recommended more stringent cut-offs (2.0 SD below expected levels or greater) to reduce the possibility of inaccurate classification. In line with these recommendations, abnormal performance on a particular LASSI-L measure for CN participants in the current study was defined as the 6^th percentile or lower, relative to the other CN individuals in the sample.

MRI assessment

All participants described above underwent structural MR imaging using a Siemens Skyra 3T scanner at Mount Sinai Medical Center, Miami Beach, Florida. Brain parcellation was obtained using a 3D T1-weighted sequence (MPRAGE) with 1.0 mm isotropic resolution using FreeSurfer Version 6.0 software (https://surfer.nmr.mgh.harvard.edu). We calculated the volumes of bilateral AD prone regions as specified by Dickerson and colleagues [32], and in our previous work [15, 16]. A composite volume of several bilateral regions was created, including the hippocampus, entorhinal cortex, amygdala, parahippocampal gyrus, inferior temporal lobule, temporal pole, supramarginal, superior parietal, precuneus, rostral middle frontal, and superior frontal areas. This composite volume was then normalized by dividing by the total intracranial volume.

Amyloid imaging

PET/CT imaging was obtained using a 3D Hoffmann brain phantom to establish a standardized acquisition and reconstruction method. Participants were infused with [18-F] florbetaben 300 MBQ over a 3-min period. Scanning commenced at 70 min after the infusion, for a duration of 20 min. We scanned all participants on a Siemens Biograph 16 PET/CT scanner operating in 3D mode (55 slices/frame, 3 mm slice thickness 128×128 matrix). The PET data was reconstructed into a 128×128×63 (axial) matrix with voxel dimensions of 0.21×0.21×0.24 cm. A small number of participants had Florbetapir as their amyloid tracer. Reconstruction was performed using manufacturer-supplied software and included corrections for attenuation, scatter, random coincidences, and dead time. Images for regional analyses were processed using Fourier analysis followed by direct Fourier reconstruction. Images were smoothed with a 3 mm Hann filter. Following reconstruction, image sets were inspected and, if necessary, corrected for inter-frame motion. Images were obtained from the top of the head to the top of the neck and CT data was employed for initial attenuation correction and image reconstruction in the sagittal, axial and coronal planes.

The PET/CT scans, including the outline of the skull, co-registered linearly (i.e., trilinear interpolation) with 12 degrees of freedom, onto the volumetric MRI scan using a T1-weighted (MP-RAGE) [33, 34]. Region-of-interest (ROI) boundaries were defined manually using the structural MRI for anatomical reference, and criteria that have been proven to provide highly reproducible outcomes [35]. This registration process ensured that the PET/CT image had the same accurate segmentation and parcellation as in the MRI scan. In 77.1% of the subjects in this study the ligand used for PET scanning was [18-F] Florbetaben and in the remainder [18-F] Florbetapir (22.9%) was used. Average activity was calculated in the ROIs corresponding to cerebellar gray matter and cerebral cortical regions. A standardized uptake value ratio (SUVR) was calculated for each region, as the ratio of regional uptake to the uptake in the cerebellar gray matter, as the reference region. A composite volume-weighted mean SUVR of 5 bilateral cortical regions (frontal, temporal, parietal, anterior and posterior cingulate cortex), was created [36].

Visual ratings of amyloid PET/CT images

All Aβ-PET scans were interpreted by an experienced reader (RD) who was blind to the cognitive and clinical diagnosis, using a methodology similar to that described by Seibyl and colleagues [37]. A final dichotomous (Amy+ versus Amy–) diagnosis was rendered. The agreement between two readers was 93.2% for positive scans and 100% for negative scans [38].

Statistical analyses

For initial comparisons between diagnostic groups (illustrated in Table 1) we employed a series of one-way analyses of variance for interval level variables. Following a statistically significant F-value of p < 0.05, we utilized the Tukey’s Honestly Significant Difference Test for mean comparisons. Adjusting for covariates such as the Mini-Mental State Examination (MMSE) did not influence the obtained results. Fisher’s Exact Test were used to test the independence between two categorical variables and this procedure is optimal in analyses where certain cells have small expected cell size.

Table 1

Demographic characteristics of study groups

	HC (n = 26)	EMCI (n = 28)	LMCI (n = 18)	AD (n = 27)	F or χ²	p
Age (52-93 y)	70.23	71.36	76.00	73.11	2.19	0.094
	(54–82)	(56–88)	(63–93)	(52–91)
	(SD = 5.9)	(SD = 8.2)	(SD = 7.9)	(SD = 8.8)
Education (5–22 y)	15.92	15.29	14.89	13.89	1.65	0.184
	(12–19)	(7–20)	(8–22)	(5–20)
	(SD = 2.6)	(SD = 3.2)	(SD = 4.0)	(SD = 3.9)
MMSE (19–30)	29.23^a	27.12^b	25.28^c	23.33^d	40.34	<0.001
	(27–30)	(24–30)	(22–30)	(19–29)
	(SD = 0.9)	(SD = 1.9)	(SD = 2.0)	(SD = 2.8)
HVLT-R Total (2–35)	25.62^a	18.50^b	14.56^c	13.74^c	28.48	<0.001
	(19–35)	(8–27)	(4–21)	(2–27)
	(SD = 4.7)	(SD = 5.7)	(SD = 4.3)	(SD = 5.1)
HVLT-R Delay (0–12)	7.12^a	2.54^b	0.88^b	0.41^b	21.85	<0.001
	(5–12)	(0–11)	(0–5)	(0–9)
	(SD = 3.2)	(SD = 3.0)	(SD = 1.6)	(SD = 18)
NACC Passage Delayed (0–25)	12.0	7.24	5.41	1.57	22.90	<0.001
	(6–20)	(1–15)	(0–25)	(0–10)
	(SD = 3.8)	(SD = 4.2)	(SD = 6.5)	(SD = 2.5)
Sex (% female)	65.1%	57.1%	44.4%	59.3%	1.42	0.71
Total MRI Volume in AD Prone	0.1122^a	0.1064^ab	0.1044^b	0.1039^b	4.52	0.005
Regions (0.0769–0.1247)	(0.0986–0.1247)	(0.0932–0.1232)	(0.0874–0.1210)	(0.0769–0.1237)
	(SD = 0.007)	(SD = 0.007)	(SD = 0.011)	(SD = 0.011)
Total LASSI-L Impairments (0–7)	0.087^a	1.67^b	3.17^c	4.29^d	45.85	<0.001
	(0–1)	(0–5)	(0–5)	(0–7)
	(SD = 0.29)	(SD = 1.5)	(SD = 1.4)	(SD = 1.6)

Means with different alphabetic superscripts are statistically significant at p < 0.05 by the Tukey Honestly Significant Difference Test. Ranges for each diagnostic group are provided below the mean. HC, healthy cognitive participants (amyloid-); EMCI, early mild cognitive impairment (amyloid+); LMCI, late mild cognitive impairment (amyloid positive+); AD, Alzheimer’ disease with mild dementia (amyloid positive+).

RESULTS

As indicated in Table 1, there were no statistically significant differences between HC, EMCI, LMCI, or AD cases with regards to age, level of educational attainment, sex, or language that the evaluation was conducted (English or Spanish). As expected, AD participants had the lowest mean MMSE scores, followed by LMCI, followed by EMCI, and then HC, who had with the highest average MMSE scores. Total structural brain volumes in AD prone regions were greater for HC, compared to LMCI and AD groups. MCI and AD groups who were Amy+ did not differ on these measures. Further, in general, AD participants had the highest mean intrusion errors on LASSI-L subscales, followed by LMCI, EMCI and HC, who had the lowest mean LASSI-L intrusion errors and the lowest mean number of LASSI-L impairments in aggregate. These latter two analyses were replicated using a Mann Whitney Test of Ranks, given heterogeneity of variance and unequal cell sizes, yielding equivalent results. The results presented in Table 1 also suggested a monotonic relationship between total number of LASSI-L intrusions and increasing cognitive impairment from HC to EMCI, LMCI and AD.

As depicted in Table 2, using previously derived cut-points on a normal sample, a series of Fisher’s Exact Tests were all statistically significant. HC participants exhibited 0% deficits in LASSI-L maximum learning (Cued A2), Proactive Semantic Interference (PSI) (Cued B1 Cued Recall; Cued B1 Intrusions), failure to recover from proactive semantic interference (frPSI) (Cued B2 Recall; Cued B2 Intrusions), or total delayed recall. HC participants evidenced impairments on Cued A3 recall or Cued A3 intrusions (measures of retroactive semantic interference) in 3.8% of cases. In contrast, over 80% of participants in the LMCI or AD groups had initial learning deficits (Cued A2) which were also observable in over a third of EMCI participants. In total, 95.8% of AD participants, 94.4% of LMCI, 70.8% of EMCI, and 8.7% of HC individuals evidenced one or more LASSI-L deficits.

Table 2

Percentage of participants with impairments using previously established cut-offs to distinguish between amyloid negative healthy cognitive versus amyloid positive early MCI, late MCI and mild dementia

	HC (n = 26)	EMCI (n = 28)	LMCI (n = 18)	AD (n = 27)	Fisher’s Exact Test	p	Cut-Points
Max. Learning
A2 Cued Recall	0.0%	35.7%	83.3%	89.0%	60.50	<0.001	≤10
PSI
B1 Cued Recall	0.0%	10.7%	5.6%	48.1%	22.69	<0.001	≤2
B1 Cued Intrusion	0.0%	32.1%	61.1%	37.0%	23.41	<0.001	≥7
frPSI
B2 Cued Recall	0.0%	3.6%	16.7%	70.4%	44.79	<0.001	≤5
B2 Cued Intrusion	0.0%	28.6%	38.9%	33.3%	14.95	0.001	≥6
RSI
A3 Cued Recall	3.8%	21.4%	22.2%	48.1%	4.49	0.002	≤4
A3 Cued Intrusion	3.8%	21.4%	22.2%	14.8%	14.43	<0.001	≥9
Delayed Recall	0.0%	29.2%	66.7%	79.2%	40.89	<0.001	≤11
Any LASSI-L Impairment	8.7%	70.8%	94.4%	95.8%	51.30	<0.001	NA

Given the high specificity for HC participants (100%) on tests measuring maximum learning, PSI, and frPSI shown in Table 2, we relaxed the cut-points for impairment by one unit to determine the extent to which performance on an individual subscale or combinations of subscales measuring maximum learning, PSI, and frPSI could differentiate between study groups.

As indicated in Table 3, all Chi-square analyses using the Fisher’s Exact Test were statistically significant. When one or more impairments were observed on Cued A2 (Maximum Learning), and Cued B1 (either Cued B1 Recall or Cued B1 Intrusions: susceptible to PSI), 7.7% of HC, 71.4% of EMCI, 94.4% of LMCI, and 96.3% of AD participants exhibited deficits.

Table 3

Percentage of participants with impairments using more liberal cut-offs to distinguish between amyloid negative healthy cognitive versus amyloid positive early MCI, late MCI and mild dementia on measures of maximum learning, proactive semantic interference (PSI) and failure to recover from PSI (frPSI)

	HC (n = 26)	EMCI (n = 28)	LMCI (n = 18)	AD (n = 27)	Fisher’s Exact Test	p	Cut-Points
Max. Learning
A2 Cued Recall	3.8%	46.4%	83.3%	82.6%	55.35	<0.001	≤11
PSI
B1 Cued Recall	0.0%	17.8%	16.7%	59.3%	27.14	<0.001	≤3
B1 Cued Intrusion	7.7%	53.6%	66.7%	40.7%	20.27	<0.001	≥6
Max. Learning &PSI	7.7%	71.4%	94.4%	96.3%	59.36	<0.001	NA
frPSI
B2 Cued Recall	0.0%	14.3%	27.8%	74.1%	40.55	<0.001	≤6
B2 Cued Intrusion	3.8%	42.9%	66.7%	44.4%	22.56	<0.001	≥5
Max. Learning, PSI &frPSI	11.5%	75.0%	94.4%	96.3%	54.48	<0.001	NA

This indicates that the majority of HC participants were classified as unimpaired while the majority of participants with EMCI, LMCI, and AD had identified cognitive impairments. When impairment on these aforementioned subscales plus Cued B2 Recall and Cued B2 Intrusions (susceptible to frPSI) were considered, 11.5% of HC, 75.0% of EMCI, 94.4% of LMCI, and 96.3% of AD exhibited impairment, suggesting that the administration of additional LASSI-L scales tapping into frPSI may not be necessary to improve the LASSI-L’s ability to efficiently differentiate among groups (Fig. 1). Table 4 indicates that measures of RSI and delayed recall alone, or in combination were not nearly as effective in classifying diagnostic groups as compared to subscales that measure maximum learning and proactive semantic interference, with 17.4% of HC exhibiting one or more deficit and 54.2% of EMCI, 77.8% of LMCI, and 88.3% of AD participants exhibiting impairment.

Fig.1

Comparison of Participants with Impairments by Scales [(1) Maximum Learning and Proactive Semantic Interference (PSI), and (2) Maximum Learning, PSI, and Failure to Recover from PSI (frPSI)] to Distinguish Between HC (amyloid negative) versus early MCI, late MCI, and mild AD (amyloid positive).

Table 4

Using more liberal cut-offs to distinguish between amyloid negative healthy cognitive versus amyloid positive early-stage MCI, late-stage MCI and mild Dementia on Retroactive Semantic Interference (RSI) and Delayed Recall

	HC (n = 26)	EMCI (E = 28)	LMCI (n = 18)	AD (n = 27)	Fisher’s Exact Test	p	Cut-Points
RSI
A3 Cued Recall	7.7%	32.1%	50.0 %	5.56%	16.40	0.001	≤5
A3 Cued Intrusions	3.8%	28.6 %	33.3 %	14.8%	8.52	0.030	≥8
Delayed Recall	4.3%	29.2%	72.2 %	79.2%	37.11	<0.001	≤2
RSI &Delayed Recall	17.4%	54.2%	77.8%	83.3%	35.13	<0.001	NA

There were cases 23 cases of delayed Recall for the HC and 26 cases for the other measures which accounted for the total impairment score for HC to be slightly greater than the sum impairment on component measures.

We could not directly compare the LASSI-L to the HVLT-R or delayed NACC passages since these later measures were actually employed to distinguish between diagnostic groups and lead to circular reasoning. However, we correlated these measures to Cued A2 Maximum recall, Cued B1 total Recall and Cued B1 Intrusions. As expected Cued A2 of the LASSI-L was correlated with the HVLT-R across the entire sample (r = 0.66; p < 0.001), HVLT-Delayed Recall (r = 0.59; p < 0.001), and NACC Delayed passages (r = 0.70; p < 0.001). The correlations were reduced for Cued B1 of the LASSI-L (r = 0.64; p < 0.001; r = 0.48; p < 0.001; and r = 0.62; p < 0.001). Correlations were lowest for Cued B1 intrusions (r = –0.36; p < 0.001; r = –0.45; p < 0.001; and r = –0.28; p < 0.008). When only impaired groups were considered, correlations were considerably reduced, most noticeably for Cued B1 intrusions and HVLT-Total, (r = –0.024; p = 0.84), HVLT-Delayed (r = –0.24; p = 0.04), and NACC Delayed Passages (r = 0.012; p = 0.92).

DISCUSSION

There is increasing evidence that the LASSI-L is an effective cognitive stress test for the detection of preclinical AD in the MCI state, or early AD. In this study, we focused on those individuals who had a presumptive underlying diagnosis of AD. Given that the participants in this study who were cognitively impaired with MCI or dementia were also Amy+, these participants fulfill criteria for prodromal AD and AD dementia, using Revised NIA-AA criteria [39, 40].

In the current investigation, HC participants did not evidence any deficits (utilizing previously-derived cut-points), on combined LASSI-L subscales that measure maximum learning and proactive semantic interference (i.e., correct responses and semantic intrusions). These indices involve the first few trials of the LASSI-L which only involves eight minutes of administration time. When employing slightly more liberal cut-points to these subscales, 7.7% of HC had one or more deficits in the above cognitive stress measures, whereas 71.4% of EMCI, 94.4% of LMCI, and 96.3% of those with mild AD dementia evidenced one or more deficits.

In contrast, as indicated in Table 4, 17.6% of HC participants were misclassified as impaired using the LASSI-L combined measures of retroactive semantic interference (correct responses and delayed intrusions) and delayed recall. RSI measures were not as robust as subscales that tap into maximum learning, PSI, and frPSI in differentiating groups with different levels of MCI and AD from healthy controls.

Compared to other list-learning tests, the LASSI-L provides a measure of maximum learning ability and stimulates semantic networks during both the encoding and the retrieval processes. This facilitates the measurement of PSI and frPSI. These cognitive processes can be examined through the number of correct responses provided upon cued semantic recall and/or the occurrence of semantic intrusion errors.

Previous research that has shown that individual LASSI-L subscales, namely PSI and frPSI, are among those indices most related to pathological brain changes in AD [3, 14]. In clinical practice, there is often a reliance on multiple indices of list-learning tests. In the case of the LASSI-L, a practitioner typically takes note of performance each of the LASSI-L subscales as they evaluate performance deficiencies. While consideration of performance on multiple memory tests has its advantages, Loewenstein and colleagues [31] have previously demonstrated that CN individuals will have at least one impaired test score if a sufficient number of memory measures are taken into account.

This raises the questions as to how multiple LASSI-L subscales should be interpreted both individually and collectively. To address this issue, conservative cut-offs were employed, representing impairment (6% or less) for LASSI-L subscales in a group of CN older adults who were consistently classified as CN after over a two-to-three-year period. We subsequently applied these cut-offs to an independent sample of individuals in the current study who were classified as cognitively normal, or diagnosed with early stage MCI, late stage-MCI, and mild dementia, all diagnosed with presumptive AD and all showing amyloid positivity on PET scans.

A major strength of this investigation is the use of formalized operational criteria for the cognitive diagnosis of HC, EMCI, LMCI, and mild AD. The differentiation between early later stage MCI based on CDR sum of boxes us unique aspect of our study. Similar CDR sum of boxes cut-offs have been employed to differentiate early and late stage MCI. [41]. We would have liked to include even earlier MCI or PreMCI patients based on a memory cut-off of 1.0 SD below expected. values. Unfortunately, there were only a handful of such participants in our cohort who were amyloid positive. This is an area of significant interest as we continue to recruit studies. In the current investigation, the use of biomarkers which allowed classification of the participants, using NIA-AA revised criteria for AD for diagnosis of the continuum of AD from early prodromal to AD dementia. A particularly important finding was that the combination of LASSI-L maximum learning and PSI subscales (both the number of correct cued recall responses and semantic intrusions) was highly sensitive to early AD pathology, while also exhibiting excellent specificity using the previously-defined cut-points. It is important to note that the assessment of these cognitive processes was achieved within the first 8 minutes of the standardized administration of the LASSI-L. This is in contrast to the full LASSI-L that may take 20 or more minutes to administer. We realize that are cell sizes that were somewhat modest and intend to continue increasing our N and plan future longitudinal analyses of the data.

In essence, adding additional measures of frPSI, RSI, and delayed recall, which occur later in the administration order of the test, did not enhance sensitivity but merely lowered specificity. As a result, these findings indicate that two LASSI-L subscales measuring maximum learning and PSI could function as an effective brief memory screening that can be completed in approximately 8 minutes and be just as effective as the full administration. Based on these findings, it would be important to investigate this abbreviated form of the LASSI-L among persons with non-AD pathologies, as well as to evaluate its generalizability among different ethnic and cultural groups. Longitudinal studies involving these LASSI-L indices as well as their relationship to tau pathology would also be useful in evaluating predictive utility over time.

In summary, this abbreviated version of the LASSI-L has excellent discriminative properties which would serve well in clinical practice and research as a screening tool for the detection of AD during its early prodromal stages. Previous findings of the high sensitivity, specificity, and strong association with AD biomarkers, and current results showing that a brief subset of LASSI-L measures could readily distinguish amyloid positive EMCI, amyloid positive LMCI from amyloid negative HC participants suggest that this brief subset of LASSI-L measures be useful for early cognitive screening for more comprehensive evaluation or identifying individuals who may be potential candidates for AD clinical trials focused on early intervention.

Footnotes

ACKNOWLEDGMENTS

This research was supported by the National Institute of Aging Grants number 5 P50 AG047726602 1 Florida Alzheimer’s Disease Research Center, 1 R01 5R01AG055638-02 and R01 AG061106-02 University of Miami. This research was also supported by the Ed and Ethel Moore Research Program.

Authors’ disclosures available online ().

References

Silverberg

, Ryan

, Carrillo

, Sperling

, Petersen

, Posner

, Snyder

, Hilsabeck

, Gallagher

, Raber

, Rizzo

, Possin

, King

, Kaye

, Ott

, Albert

, Wagster

, Schinka

, Cullum

, Farias

, Balota

, Rao

, Loewenstein

, Budson

, Brandt

, Manly

, Barnes

, Strutt

, Gollan

, Ganguli

, Babcock

, Litvan

, Kramer

, Ferman

(2011) Assessment of cognition in early dementia. Alzheimers Dement 7, e60–e76.

Harvey

, Cosentino

, Curiel

, Goldberg

, Kaye

, Loewenstein

, Marson

, Salmon

, Wesnes

, Posner

(2017) Performance-based and observational assessments in clinical trials across the Alzheimer’s disease spectrum. Innov Clin Neurosci 14, 30–39.

Loewenstein

, Curiel

, Duara

, Buschke

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

Podhorna

, Krahnke

, Shear

, Harrison

(2016) Alzheimer’s Disease Assessment Scale-Cognitive subscale variants in mild cognitive impairment and mild Alzheimer’s disease: Change over time and the effect of enrichment strategies. Alz Res Therapy 8, 8.

Kueper

, Speechley

, Montero-Odasso

(2018) The Alzheimer’s Disease Assessment Scale-Cognitive Subscale (ADAS-Cog): Modifications and responsiveness in pre-dementia populations. A narrative review. J Alzheimers Dis 6, 423–444.

Karin

, Hannesdottir

, Jaeger

, Annas

, Segerdahl

, Karlsson

, Sjögren

, von Rosen

, Miller

(2014) Psychometric evaluation of ADAS-Cog and NTB for measuring drug response. Acta Neurol Scand 129, 114–122.

Hobart

, Cano

, Posner

, Selnes

, Stern

, Thomas

, Zajicek

(2013) Putting the Alzheimer’s cognitive test to the test II: Rasch Measurement Theory. Alzheimers Dement 9, S10–S20.

Skinner

, Carvalho

, Potter

, Thames

, Zelinski

, Crane

, Gibbons

, Alzheimer’s Disease Neuroimaging Initiative (2012) The Alzheimer’s Disease Assessment Scale- Cognitive-Plus (ADAS-Cog-Plus): An expansion of the ADAS-Cog to improve responsiveness in MCI. Brain Imaging Behav 6, 489–501.

Thomas

, Eppig

, Edmonds

, Jacobs

, Libon

, Au

, Salmon

, Bondi

, Alzheimer’s Disease Neuroimaging Initiative (2018) Word-list intrusion errors predict progression to mild cognitive impairment. Neuropsychology 32, 235–245.

10.

Libon

, Bondi

, Price

, Lamar

, Eppig

, Wambach

, Nieves

, Delano- Wood

, Giovannetti

, Lippa

, Kabasakalian

, Cosentino

, Swenson

, Penney

(2011) Verbal serial list learning in mild cognitive impairment: A profile analysis of interference, forgetting, and errors. J Int Neuropsychol Soc 17, 905–914.

11.

, Liu

, Li

, Zhou

(2020) Effects of exercise on proactive interference in memory: Potential neuroplasticity and neurochemical mechanisms. Psychopharmacology 237, 1917–1929.

12.

Atkins

, Reuter-Lorenz

(2011) Neural mechanisms of semantic interference and false recognition in short-term memory. Neuroimage 56, 1726–1734.

13.

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 disease: Discriminative properties and relation to amyloid load. Am J Geriatr Psychiatry 24, 804–813.

14.

Loewenstein

, Curiel

, DeKosky

, Bauer

, Rosselli

, Guinjoan

, Adjouadi

, Peñate

, Barker

, Goenaga

, Golde

, Greig-Custo

, Hanson

, Li

, Lizarraga

, Marsiske

, Duara

(2018) Utilizing semantic intrusions to identify amyloid positivity in mild cognitive impairment. Neurology 91, E976–E984.

15.

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.

16.

Loewenstein

, Curiel

, DeKosky

, Rosselli

, Bauer

, Grieg-Custo

, Penate

, Li

, Lizagarra

, Golde

, Adjouadi

, Duara

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

17.

Matías-Guiu

, Curiel

, Rognoni

, Valles-Salgado

, Fernández-Matarrubia

, Hariramani

, Fernández-Castro

, Moreno-Ramos

, Loewenstein

, Matías-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.

18.

Matias-Guiu

, Cabrera-Martín

, Curiel

, Valles-Salgado

, Rognoni

, Moreno-Ramos

, Carreras

, Loewenstein

, Matías-Guiu

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

19.

Sánchez

, 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.

20.

Curiel Cid

, Crocco

, Duara

, Garcia

, Rosselli

, DeKosky

, Smith

, Bauer

, Chirinos

, Adjouadi

, Barker

, Loewenstein

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

21.

Morris

(1993) The Clinical Dementia Rating (CDR): Current version and scoring rules. J Neurol 43, 2412–2414.

22.

Benedict

RHB

, Schretlen

, Groninger

, Brandt

(1998) Hopkins Verbal Learning Test - Revised: Normative data and analysis of inter-form and test-retest reliability. Clin Neuropsychol 12, 43–55.

23.

Beekly

, Ramos

, Lee

, Deitrich

, Jacka

, Wu

, Hubbard

, Koepsell

, Morris

, Kukull

, Reiman

, Kowall

, Landreth

, Shelanski

, WelshBohmer

, Levey

, Potter

, Ghetti

, Price

, Hyman

, Petersen

, Sano

, Ferris

, Mesulam

, Kaye

, Bennett

, Yesavage

, Marson

, Beck

, DeCarli

, Cotman

, Cummings

, Thal

, Markesbery

, Gilman

, Trojanowski

, DeKosky

, Chui

, Rosenberg

, Raskind

(2007) The National Alzheimer’s Coordinating Center (NACC) database: The uniform data set. Alzheimer Dis Assoc Disord 21, 249–259.

24.

Binetti

, Magni

, Cappa

, Padovani

, Bianchetti

, Trabucchi

(1995) Semantic memory in Alzheimer’s disease: An analysis of category fluency. J Clin Exp Neuropsychol 17, 82–89.

25.

Reitan

(1958) Validity of the Trail Making Test as an indicator of organic brain damage. Percept Mot Skills 21, 271–276.

26.

Wechsler

(2008) Wechsler Adult Intelligence Scale–Fourth Edition. Pearson, San Antonio, TX.

27.

Petersen

, Caracciolo

, Brayne

, Gauthier

, Jelic

, Fratiglioni

(2014) Mild cognitive impairment: A concept in evolution. J Intern Med 275, 214–228.

28.

Crocco

, Acevedo

, Curiel

, Loewenstein

(2014) An evaluation of deficits in semantic cuing, proactive and retroactive interfere as early features of Alzheimer’s disease. Am J Geriatr Psychiatry 9, 889–897.

29.

Curiel Cid

, Loewenstein

, Rosselli

, Matias-Guiu

, Piña

, Adjouadi

, Cabrerizo

, Bauer

, Chan

, DeKosky

, Golde

, Greig-Custo

, Lizarraga

, Peñate

, Duara

, 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.

30.

Torres

, Rosselli

, Loewenstein

, Curiel

, Vélez 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.

31.

Loewenstein

, Acevedo

, Ownby

, Agron

, Barker

, Isaacson

, Strauman

, Duara

(2006) Using different memory cutoffs to assess mild cognitive impairment. Am J Geriatr Psychiatry 14, 911–919.

32.

Dickerson

, Stoub

, Shah

, Sperling

, Killiany

, Albert

, Hyman

, Blacker

, Detoledo-Morrell

(2011) Alzheimer-signature MRI biomarker predicts AD dementia in cognitively normal adults. Neurology 76, 1395–1402.

33.

Smith

, Jenkinson

, Woolrich

, Beckmann

, Behrens

TEJ

, Johansen-Berg

, Bannister

, De Luca

, Drobnjak

, Flitney

, Niazy

, Saunders

, Vickers

, Zhang

, De Stefano

, Brady

, Matthews

(2004) Advances in functional and structural MR image analysis and implementation as FSL. Neuroimage 22, 1646–1656.

34.

Lizarraga

, Li

, Cabrerizo

, Barker

, Loewenstein

, Duara

, Adjouadi

(2018) A neuroimaging web services interface as a cyber physical system for medical imaging and data management in brain research: Design study. JMIR Med Informatics 6, e26.

35.

Desikan

, Ségonne

, Fischl

, Quinn

, Dickerson

, Blacker

, Buckner

, Dale

, Maguire

, Hyman

, Albert

, Killiany

(2006) An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest. Neuroimage 31, 968–980.

36.

Rowe

, Ackerman

, Browne

, Mulligan

, Pike

, O’Keefe

, Tochon-Danguy

, Chan

, Berlangieri

, Jones

, Dickinson-Rowe

, Kung

, Zhang

, Kung

, Skovronsky

, Dyrks

, Holl

, Krause

, Friebe

, Lehman

, Lindemann

, Dinkelborg

, Masters

, Villemagne

(2008) Imaging of amyloid β in Alzheimer’s disease with 18F-BAY94-9172, a novel PET tracer: Proof of mechanism. Lancet Neurol 7, 129–135.

37.

Seibyl

, Catafau

, Barthel

, Ishii

, Rowe

, Leverenz

, Ghetti

, Ironside

, Takao

, Akatsu

, Murayama

, Bullich

, Mueller

, Koglin

, Schulz-Schaeffer

, Hoffmann

, Sabbagh

, Stephens

, Sabri

(2016) Impact of training method on the robustness of the visual assessment of 18F-florbetaben PET scans: Results from a phase-3 study. J Nucl Med 57, 900–906.

38.

Duara

, Loewenstein

, Lizarraga

, Adjouadi

, Barker

, Greig-Custo

, Rosselli

, Penate

, Shea

YFF

, Behar

, Ollarves

, Robayo

, Hanson

, Marsiske

, Burke

, Ertekin-Taner

, Vaillancourt

, De Santi

, Golde

, St

(2019) Effect of age, ethnicity, sex, cognitive status and APOE genotype on amyloid load and the threshold for amyloid positivity. Neuroimage Clin 22, 101800.

39.

Albert

, DeKosky

, Dickson

, Dubois

, Feldman

, Fox

, Gamst

, Holtzman

, Jagust

, Petersen

, Snyder

, Carrillo

, Thies

, Phelps

(2011) The diagnosis of mild cognitive impairment due to Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 7, 270–279.

40.

Jack

Jr , Bennett

, Blennow

, Carrillo

, Dunn

, Haeberlein

, Holtzman

, Jagust

, Jessen

, Karlawish

, Liu

, Molinuevo

, Montine

, Phelps

, Rankin

, Rowe

, Scheltens

, Siemers

, Snyder

, Sperling

, Contributors (2018) NIA-AA research framework: Toward a biological definition of Alzheimer’s disease. Alzheimers Dement 14, 535–562.

41.

O’Bryant

, Lacritz

, Hall

, Waring

, Chan

, Khodr

, Massman

, Hobson

, Cullum

(2010) Validation of the new interpretive guidelines for the clinical dementia rating scale sum of boxes score in the national Alzheimer’s coordinating center database. Arch Neurol 67, 746–749.