Use of the Spanish English Neuropsychological Assessment Scale in older adult Latines and those at risk for Autosomal Dominant Alzheimer’s Disease

Abstract

OBJECTIVE:

The Spanish English Neuropsychological Assessment Scale (SENAS) is a cognitive battery with English and Spanish versions for use with persons for whom either language is predominant. Few studies have examined its utility outside the normative sample. The current study examined SENAS performance in samples of older adult Latines and Latines with or at risk for autosomal dominant Alzheimer’s disease (ADAD) mutations.

METHOD:

The SENAS was administered to 202 older adults from the Los Angeles Latino Eye Study (LALES) and 29 adults with (carriers) or without (non-carriers) mutations causing ADAD. We examined associations between SENAS, age, education, and language (LALES) and between SENAS, estimated years from familial age of dementia diagnosis, education, language, and acculturation (ADAD). Partial correlations were used to examine differences in correlational strength between estimated years from familial age of dementia diagnosis and SENAS scores among ADAD carriers compared to chronological age and SENAS in the LALES sample. Exploratory t-tests were performed to examine SENAS performance differences between ADAD carriers and non-carriers.

RESULTS:

In an older adult sample (LALES), increased age correlated with worse verbal delayed recall; English fluency and higher education correlated with better naming and visuospatial subtest performance. Among ADAD carriers, verbal and nonverbal delayed recall and object naming subtest performance worsened as they approached their familial age of dementia diagnosis. English fluency and higher U.S.-acculturation were related to better SENAS performance among carriers and non-carriers. Tests of verbal delayed recall and object naming best distinguished ADAD carriers from their familial non-carrier counterparts.

CONCLUSIONS:

Verbal delayed recall and object naming measures appear to be most sensitive to age-related changes in older adult samples and mutation-related changes in distinguishing ADAD carriers from non-carriers. Future research should examine the sensitivity of SENAS in other samples, such as larger samples of symptomatic ADAD carriers and other AD subtypes.

Introduction

Latines¹ are the fastest-growing ethnoracial group in the United States. Recent projections estimate that 1.1 million Latines in the U.S. will be affected by Alzheimer's disease (AD) by 2030 (Wu et al., 2016). Latines are disproportionately affected by disparities associated with greater AD risk, including lower socioeconomic status, educational attainment, and educational quality (Garcia et al., 2018; Weden et al., 2017; Zeki Al Hazzouri et al., 2011). Relative to their non-Latine White counterparts, U.S. Latines face higher rates of vascular health conditions associated with increased dementia risk (e.g., diabetes, hypertension) and appear to develop these conditions at younger ages (Mayeda et al., 2013; Vega et al., 2017). A nationally representative estimate across the United States reported a moderately elevated dementia incidence rate of 19.6 per 1000 person-years among a sample of Latines predominantly of Mexican origin (Mayeda et al., 2016). Moreover, prior research in a sample of Latines (Countries of origin of sample: 47% Mexican, 25% Central American, 15% South American, 7% Cuban, 4% Puerto Rican, 2% unspecified) residing in Southern California found, on average, Latine adults were four years younger at the time of dementia diagnosis relative to their non-Latine White counterparts (Fitten et al., 2015). Thus, the higher reported rates of dementia risk factors and younger age of dementia onset raise the importance of understanding AD risk, diagnosis, and development of cognitive symptoms across the lifespan among Latines (Vega et al., 2017).

At the population level, Latines are heterogeneous, with wide variability in region of origin, educational attainment, socioeconomic status, degree of acculturation, and linguistic factors (e.g., dialect, bilingualism), for example. These features may influence dementia risk and trajectories but remain understudied (Vega et al., 2017; Weden et al., 2017). Language assessments, for example, are an essential component of AD assessment, given the impact on the semantic fluency networks in the course of the disease. However, there are several limitations when measures are administered in English or for those for whom English is not their primary language. Prior research has demonstrated that English literacy and reading level among Latine older adults was a more reliable predictor of baseline cognitive performance above and beyond other demographic factors, including years of education (Manly et al., 2004).

Further, the development of normative data from which to stage cognitive impairment often only accounts for age and years of education, neglecting potentially critical cultural factors, such as primary language, language proficiency, years spent in the U.S., and acculturation (Manly & Espino, 2004; Sayegh, 2015). In addition, cross-language interference in those for whom English is not their primary and dominantly-used language can artificially inflate impairment and lead to a false diagnosis of dementia (Sayegh, 2015). Given that the above factors can influence performance on cognitive measures, developing and validating measures that adequately capture cognitive changes and AD risk for culturally and linguistically diverse populations is critical.

As a result, increasing attention has been paid to the development and validation of neuropsychological test batteries that provide equivalence in Spanish-speaking and English-speaking populations and are also sensitive to age- and disease-related cognitive changes (Bermúdez-Llusá et al., 2019; Karr et al., 2022; Luna-Lario et al., 2015; Morlett Paredes et al., 2021; Rivera Mindt et al., 2021; Sánchez-Benavides et al., 2014; Suárez et al., 2021). The Spanish and English Neuropsychological Assessment Scale (SENAS; Mungas et al., 2004) was developed to provide a reliable and valid cognitive test battery appropriate for use with unilingual and bilingual Spanish and English-speaking populations. Item response theory was employed to develop psychometrically matched measures across scales and language versions to assess cognitive domains important for age-related cognitive changes (e.g., episodic memory, verbal fluency, etc.; Mungas et al., 2000; 2004). Different language versions can be susceptible to differential item functioning and measurement bias, hence the importance of item-level analysis in the measure’s development. SENAS memory subtests were reportedly most sensitive to dementia diagnosis, with moderate sensitivity observed on language, spatial, and executive function measures (Mungas et al., 2005b). However, studies of the SENAS in other Latine samples have been limited.

Regarding psychosocial and cultural influences on SENAS performance, prior SENAS research has shown educational and language effects such that education strongly correlated with semantic memory, and the English-language version correlated with higher SENAS scores. In contrast, the Spanish-language version correlated with lower SENAS scores (Mungas et al., 2005a). Acculturation effects were also observed, but acculturation did not correlate with performance after controlling for education and language (Mungas et al., 2005a).

While the SENAS was developed in a Latine sample ages 60 and older and predominantly of Mexican origin (Mungas et al., 2004; Mungas et al., 2005b), the utility of the SENAS in rare forms of dementia, including Latines with or at risk for autosomal dominant Alzheimer’s disease (ADAD), has not yet been explored. Studies of persons with or at risk for ADAD provide a unique window into the pathogenesis of AD, given that dementia onset is essentially guaranteed if one is a carrier and the staging of dementia can be predicted using information about the mutation type and a carrier family, when available. Variants in one of three genes cause ADAD: PSEN1, PSEN2, and APP (Ringman, 2005), with a 50% heritability risk. For those who inherit the mutation, it causes fully penetrant AD with a relatively predictable age of onset (Ryman et al., 2014). Persons who are determined to carry the mutation are considered “mutation carriers,” and persons who are a part of families with a history of an ADAD mutation but whose carrier status is not known are referred to as “at risk,” whereas those who are a part of a family with a history of an ADAD mutation, but are determined to have not inherited the mutation, are referred to as “non-carriers.” The average age of dementia onset is relatively consistent within carriers from the same family or who have the same mutation (e.g., PSEN1 A431E) but appears to vary across families or different mutation types (Bateman et al., 2012; Ryman et al., 2014). A family-specific age of dementia diagnosis can be estimated by taking the median age of dementia diagnosis for all affected persons in a given family. Using this information, an estimated years from familial age of dementia diagnosis can be calculated for each individual by subtracting the family-specific median age of dementia diagnosis from an individual's age, which estimates how many years away an individual is from dementia diagnosis. A negative estimate indicates the number of years younger the individual is than their expected family-specific age of dementia diagnosis, whereas a positive number indicates they are older than their family-specific age of dementia diagnosis. For example, if an individual is 30 years old and their familial median age of dementia diagnosis is 42, they would have an estimated years from familial age of dementia diagnosis of “-12.” If an individual is diagnosed at the same age as their family’s median age of dementia diagnosis, this would yield an estimated years from familial age of dementia diagnosis of “0” (Medina et al., 2021; Ryman et al., 2014).

Many Latine families with specific PSEN1 ADAD mutations, representing founder effects, have been identified and described in Colombia, Puerto Rico, and Mexico (Athan et al., 2001; Lopera et al., 1997; Murrell et al., 2006; Yescas et al., 2006). The A431E PSEN1 variant from Mexico has a mean age of symptom onset of 40 years and a mean age of dementia diagnosis of 42 years. Another causative ADAD mutation, the V7171 APP substitution, was the first described ADAD mutation (the "London" mutation; Goate et al., 1991) and was subsequently found in diverse populations, including in Mexico (Ringman et al., 2007). Because the development of late-onset AD can be challenging to predict, studying ADAD mutation carriers can provide insight into the symptomatic presentation and rate of decline in a population essentially guaranteed to develop AD without the confounding effects of age-related comorbidities (Murrell et al., 2006; Ringman, 2005; Ryman et al., 2014). Identifying the presymptomatic stages of AD before the accumulation of irreversible damage may provide a viable early intervention window to slow or delay disease progression (Caselli & Reiman, 2014; O’Connor et al., 2020; Ryman et al., 2014). Episodic memory is the most commonly affected domain in AD across all types; however, about 30–40% of patients with ADAD mutations or non-familial, early-onset AD have atypical presentations affecting non-memory domains, including executive, behavioral, language, and visuospatial skills (Balasa et al., 2011; Koedam et al., 2010; Mendez, 2012). Further, early cognitive changes have been detected 10–20 years prior to dementia diagnosis using information about mutation type and familial or mutation-specific age of onset and have been utilized to predict disease progression in late-onset AD (Bateman et al., 2012; Parnetti et al., 2019; Ringman et al., 2005). Thus, with the ability to accurately predict the age of symptom onset and therefore study presymptomatic stages, assessing ADAD mutation carriers in preclinical stages may provide a model for characterizing preclinical and prodromal disease in late-onset, sporadic forms, including markers of early cognitive changes.

Few studies have examined which neuropsychological measures are most sensitive to detecting early impairment among those with or at risk for ADAD (Almkvist et al., 2017; O’Connor et al., 2020; Weston et al., 2018). None to our knowledge have addressed the use of the SENAS in a sample of linguistically and culturally diverse Latines with or at risk for ADAD in the U.S. (Almkvist et al., 2017; O’Connor et al., 2020; Weston et al., 2018). Studies in the Colombian ADAD kindred using a Spanish language version of the Consortium to Establish a Registry for Alzheimer's Disease (CERAD) battery identified tests of Word List Recall, Naming, the Mini-Mental State Exam, orientation to time, constructional praxis, and the Raven's Progressive Matrices as having good discriminative capacity between ADAD carriers and non-carriers early in the course of disease (Ayutyanont et al., 2014). Another study examined the discriminative ability of two complex reaction time tasks, which revealed that ADAD carriers demonstrated slower reaction time as they approached the anticipated age of dementia onset compared to non-carriers (Medina et al., 2021). Tasks that may reliably discriminate carriers from non-carriers have important implications in guiding research on early preclinical cognitive deficits in late-onset forms. However, more research in this area is needed.

The purpose of the current study was threefold: first, to evaluate SENAS performance in a Los Angeles community-based sample of older adult Latines (Los Angeles Eye Study; LALES), assessment age, education, and language effects (Study 1); second, to examine how incipient ADAD affects SENAS scores, including the effects of estimated years from familial age of dementia diagnosis, language, education, and acculturation, as well as which SENAS subtests best distinguished ADAD carriers from familial non-carriers (i.e., persons in a family with risk of ADAD who have not inherited mutations; Study 2). A third exploratory study was also conducted between an older adult community sample (LALES; (Exploratory Study 3) and ADAD carriers to compare whether chronologic age (LALES) or estimated years from familial age of dementia diagnosis (ADAD) was more strongly associated with SENAS performance, holding constant educational attainment.

In the LALES older adult community sample (Study 1), we hypothesized that chronologic age would have a significant effect on the SENAS, such that the older a participant was, the lower their SENAS performance. Regarding education and language effects in the older adult sample, we hypothesized that education would correlate most strongly with verbal memory measures and slightly higher SENAS performance for the English version than the Spanish version, consistent with prior research (Mungas et al., 2005a).

In the ADAD sample (Study 2), we hypothesized that among mutation carriers, the closer people were to their estimated years from familial age of dementia diagnosis, the worse their SENAS performance. In addition, we hypothesized that the estimated years from familial age of dementia diagnosis effects would be most robust relative to education, language (English or Spanish), and acculturation, given the strong effects of ADAD mutations. Among non-carriers, the effects of chronological age, education, language, and acculturation would be significantly associated with the SENAS, with age effects most strongly correlating with verbal and nonverbal memory, language and acculturation effects with memory and language subtests, and education with verbal fluency measures, consistent with prior research (Mungas et al., 2005a). Regarding subtests that would best distinguish ADAD carriers from noncarriers, we hypothesized that verbal and nonverbal memory measures would emerge as the best subtests to distinguish ADAD mutation carriers within 10 years of their estimated familial age of dementia onset from their noncarrier counterparts (i.e., family members at risk of ADAD mutation but determined not to carry mutation after genetic testing).

Finally, for Exploratory Study 3, we hypothesized that correlations between SENAS subtests and estimated years from familial age of dementia diagnosis, controlling for education and chronological age effects, would be stronger in the ADAD mutation carrier sample relative to correlations between SENAS subtests and chronological age (controlling for education effects) in the LALES older adult sample, given that genetic predisposition would have a stronger impact on cognitive performance than the effects of chronological age among older adults. Because estimated years from familial age of dementia diagnosis is an approximation for dementia onset based on the average age of dementia diagnosis within one's family, we hypothesized that this would be more strongly related to SENAS performance above and beyond the effects of assessment age and education. In the older adult sample, chronological age would be more relevant because dementia risk increases with advancing age in the general aging population.

Methods

Participants

Participants included 202 Latine adults enrolled in the Los Angeles Latino Eye Study (LALES, NIH U10-EY011753) and 29 persons affected by or at risk for inheriting ADAD mutations recruited from UCLA and USC as a part of a parent study of ADAD (NIH K08AG22228). The LALES is a population-based study of eye disease prevalence in a sample of Latines of predominantly Mexican ancestry who were at least 40 years old and resided in La Puente, California. La Puente was selected as the study catchment area due to its demographic and socioeconomic characteristics reflecting larger patterns observed among Latines of Mexican origin residing in Los Angeles County, California state, and across the United States more broadly. LALES participants aged 65 and older were selected to complete a bilingual neuropsychological assessment battery, the Spanish English Neuropsychological Assessment Scale (SENAS). Participants provided informed consent and completed a comprehensive clinical examination, an in-home questionnaire consisting of demographic, behavioral, and ocular risk factor assessments, and a survey related to health-related and vision-related quality of life (Varma et al., 2004). The present study included LALES participants fluent in Spanish or English, with at least six years of education, and who completed the SENAS. All 29 ADAD subjects were also Latine, at least 18 years of age, had at least six years of education, and had a Clinical Dementia Rating Scale (CDR) of less than 1, meaning they had scores of no (0) or mild cognitive impairment (0.5), as the focus of the present study was to examine presymptomatic or early stages of cognitive decline. This study was approved by the Institutional Review Boards at the University of Southern California and the University of California at Los Angeles. It was conducted in accordance with the Helsinki Declaration.

Measures

Spanish English Neuropsychological Assessment Scale (SENAS)

The SENAS (Mungas et al., 2004) is a battery simultaneously developed in English and Spanish sensitive to age-related cognitive changes. Due to visit time limitations of the parent study, a partial SENAS battery was administered to LALES participants. The following seven subtests were administered to the LALES and ADAD samples: Word List Learning and Delayed Recall, Nonverbal Learning and Delayed Recall, Object Naming, Phonemic Fluency, Category Fluency, Spatial Localization, and Pattern Recognition. SENAS raw scores were used in all analyses. Although the full 16 subtests of the SENAS were not administered, the seven subtests that were administered addressed each of the five-factor latent variables of cognitive domains found in the full SENAS (see Mungas et al., 2011 supplement for detailed information on the SENAS beyond the scope of this paper). Therefore, domains of episodic memory, semantic memory/language, attention/working memory, and fluency were adequately assessed in the present study and are detailed below. Participants were administered the Spanish or English SENAS version based on an assessment of their preferred language.

Word List Learning & Delayed Recall.

A 15-item word list of common objects purchased at a grocery store is read at a rate of one per second, with a fixed presentation order across five learning trials, followed by a delayed free recall trial.

Nonverbal Learning (Or Spatial Configuration Learning) & Delayed Recall.

A map-like figure is shown, which contains various shapes (akin to "countries on a map") in different colors. The figure is then removed and replaced by the same figure in black-and-white; participants are asked to complete an immediate recall task by pointing to the appropriate color of that shape/region on a palette-like page. This is repeated across seven learning trials, each with increasing difficulty as two regions are added per trial. During the final trial, all 12 colored regions are presented twice. The full black-and-white map and a stimulus page with 12 colored ovals are shown after a 20-minute delay. The ability to match the colors to the appropriate places on the map is scored.

Object Naming.

Colored pictures of objects are shown, and participants are asked to name the objects. Object words were selected based on similar usage frequency in English and Spanish.

Phonemic Fluency.

A target letter is presented, and examinees must name as many words that start with the target letter as possible in 60 seconds. For the SENAS, there are two trials, “F” (allows for words beginning with “ph” that make the “f” sound, e.g., phone) and “L.” Proper nouns are not allowed.

Category Fluency.

The examinee names as many unique words as possible pertaining to a given category in 60 seconds. The full SENAS has three category trials (animals, fruits, and vegetables), but only the animals trial was administered to both the ADAD and LALES participants.

Spatial Localization.

This task examines the visuospatial ability. A stimulus page is presented with a dog burying a bone in a yard with a doghouse and tree as spatial cues in the top half and a target area, the dog in the doghouse with a tree in the bottom half. The subject must point to where the bone is buried in the bottom half. The stimulus and target are presented simultaneously such that the task does not require memory recall. A grid outlines the target area and is scored based on the region of the grid correctly marked to contain the bone. The orientation of the stimuli differs on more complex items.

Pattern Recognition.

This task involves visually discriminating between a series of six black-and-white designs. The target stimulus design has six alternative designs, one identical to the target stimulus, and the correct stimulus must be identified.

Estimated years from familial age of dementia diagnosis.

Within ADAD carriers from the same family, the average age of dementia onset is fairly consistent. Therefore, a time to estimated dementia diagnosis variable, referred to as “estimated years from familial age of dementia diagnosis” can be calculated by subtracting the family-specific median age of dementia diagnosis from an individual’s chronological age (Ryman et al., 2014). Negative estimate values indicate an individual is younger than their family-specific median age of dementia diagnosis. For example, if the family-specific median age of dementia diagnosis was 42 years and an individual was 32 years old, they would receive an estimate of “-10.” An estimate of “0” indicates the individual received a dementia diagnosis at the same age as the average age of dementia diagnosis associated with the mutation within their family.

Clinical Dementia Rating Scale (CDR)

The CDR (Hughes et al., 1982) is used by clinicians to characterize dementia severity based on a scale of 0–3: no impairment (CDR = 0), questionable or mild impairment (CDR = 0.5), mild dementia (CDR = 1), moderate dementia (CDR = 2), and severe dementia (CDR = 3). CDR ratings are based on semi-structured interviews with the patient and a knowledgeable informant, usually a spouse or close family member. Participants undergo brief assessments of memory, orientation, judgment, and problem-solving. In addition, informants are asked about the participant’s abilities in these areas and their functional abilities. A previous multicenter study of the CDR reported interrater reliability of 0.62 (Rockwood et al., 2000). In the present study, CDR ratings were only available for the ADAD sample and were not a component of the LALES study.

Language

For the LALES study, participants completed a self-report questionnaire about their preferred language, which was used to determine which language version of the SENAS was administered. A subset of LALES participants completed a demographic questionnaire about language preference for speaking, reading, and writing. For the ADAD sample, participants’ preferred testing language (English or Spanish) was determined by a self-report questionnaire about language background characteristics, including their preferred language for speaking, reading, and writing.

Acculturation Rating Scale for Mexican Americans-II (ARSMA-II)

The ARSMA-II (Cuellar et al., 1995) is a 30-item measure of acculturation that assesses the following: language use and preference, ethnic identity and classification, cultural heritage and ethnic behaviors, and ethnic interaction. Each item is scored on a five-point Likert scale ranging from 1 (not at all) to 5 (extremely often or almost always). The items are also designed to construct two subscales that measure orientation to Mexican culture (Mexican Orientation Scale; MOS) and Anglo culture (Anglo Orientation Scale; AOS), respectively. The sum of 17 items is divided by 17 to calculate the MOS (coefficient alpha = .88). The remaining 13 items are summed and divided by 13 to calculate the AOS (coefficient alpha = .83). A continuum score is also constructed to indicate total acculturation by subtracting the MOS mean from the AOS mean, with low scores reflecting a greater Mexican cultural orientation and higher scores indicating a greater Anglo cultural orientation (Jimenez et al., 2010).

In the present study, ARSMA-II ratings were only available for the ADAD sample, not the LALES sample. Of note, while most of the ADAD sample was of Mexican origin, there were several participants with origins in other Latin American countries. Thus, participants with origins other than Mexico were instructed to answer the MOS scale with their Latin American region of origin (e.g., Puerto Rico) in mind, and reference to Latine Orientation was utilized in place of Mexican Orientation to be inclusive of all participant’s Latin American origins.

Analyses

First, descriptive statistics were run to examine the characteristics of ADAD and LALES samples for all studies. A cutoff value of .05 was used for all statistical tests. All statistical analyses were performed in R 4.1.1 (R Core Team, 2021), the lme4 package (Bates et al., 2015).

For Study 1 in the older adult sample, Pearson product-moment correlations were tested to identify significant associations between chronological age, years of education, language use (Spanish or English), and SENAS subtests within the LALES sample.

For Study 2 within the ADAD sample, correlations were estimated between estimated years from familial age of dementia diagnosis, years of education, language (Spanish or English), acculturation (Latine orientation, Anglo orientation, and total acculturation), and SENAS subtests. The 29 ADAD participants were nested within 13 different families, and as such statistical models were chosen that addressed the dependence of this sample. Then, t-tests were run between ADAD mutation carriers within ten years of their estimated years from familial age of dementia diagnosis and their non-carrier counterparts to examine differences in SENAS performance in this specialized population, given that most cognitive changes among ADAD mutation carriers occur within 10 years of their average familial age of dementia onset.

Finally, for exploratory Study 3, partial correlation coefficients were estimated to compare the strength of the association between chronologic age and SENAS subtests (LALES) and estimated years from familial age of dementia diagnosis and SENAS subtests (ADAD mutation carriers only), controlling for the effects of education in both groups and chronologic age (in the ADAD mutation carriers only).

Results

Descriptive Results

Skewness and kurtosis of SENAS subtests were acceptable in both the LALES and ADAD samples. Analysis of the residuals suggests that general linear modeling assumptions of homoscedastic and independent and identical normally distributed residuals were met. Table 1 presents the sample characteristics of the ADAD and LALES participants. LALES participants were significantly older (t = -19.62, p <.001) and had fewer years of formal education (t = -4.31, p <.001) than the ADAD sample. LALES participants had a median age of 68 years and nine years of education, whereas ADAD participants had a median age of 36 years and 12 years of education. They did not significantly differ in the proportion of men or women or testing language (68% Spanish for LALES, 76% for ADAD).

Table 1.

Demographic Characteristics of the Study Sample

	LALES Sample (n = 202) Median (IQR) or n	ADAD Sample (n = 29) Median (IQR) or n	ADAD Carriers (n = 16) Median (IQR) or n	ADAD Non-Carriers (n = 13) Median (IQR) or n	t/X 2	p
Demographics
Age (median years, IQR)	68.00(10.00)	36.00(12.00)	34.50(9.75)	39.00(10.00)	−19.62**	<.001
Age Range (years)	[59.00,86.00]	[24.00,60.00]	[24.00,43.00]	[28.00,60.00]	--	--
Estimated Years from Familial Age of Dementia Diagnosis (median years, IQR)	NA	--	−10.50(10.50)	NA	--	--
Estimated Years from Familial Age of Dementia Diagnosis Range (years from familial dementia diagnosis)	NA	[−24.00, 18.00]	[−22.00, −1.0]	[−24.00,18.00]	--	--
Education (median years, IQR)	9.00(5.75)	12.00(5.00)	12.00(3.50)	13.00(7.00)	−4.31**	<.001
Education Range (years)	[6.00,17.00]	[6.00,18.00]	[8.00,18.00]	[6.00,17.00]	--	--
Sex (N Female)	121	23	13	10	0.61	.44
Language (N tested in Spanish)	138	22	13	9	0.04	.85
ARSMA-II MOS+Mean (SD)	--	4.26(0.71)	4.19(0.64)	4.33(0.81)	−.53	.60
ARSMA-II AOS+Mean (SD)	--	2.72(1.08)	2.83(0.92)	2.58(1.28)	.60	.55
ARSMA-II Acculturation Level+Mean (SD)	--	1.66(1.04)	1.69(0.95)	1.62(1.19)	.18	.85
Clinical Characteristics
CDR Rating (n)
0	--	22	11	11	--	--
0.5	--	7	5	2	--	--
A431E PSEN1 Mutation Carrier (n)	--	19	12	7	--	--
Non-A431E Mutation Carriers (n)	--	6	4	2	--	--
BDI-II Total Score (median, IQR)	--	6.00(7.00)	4.50(5.75)	7.00 (7.00)	.37	.72

Note. ADAD = Autosomal Dominant Alzheimer’s Disease; AOS = Anglo Orientation Scale; APP = amyloid precursor protein; ARSMA-II = The Acculturation Rating Scale for Mexican Americans-II; BDI-II = Beck Depression Inventory Second Edition; CDR = Clinical Dementia Rating; IQR = interquartile range; MOS = Mexican Orientation Scale; PSEN1 = presenilin-1; U.S. = United States

^*p < .05;

^**p < .001,

⁺= analyses done within ADAD carrier and non-carrier sample only

Study 1 Sample Characteristics

In the LALES sample, 60% reported being born in Mexico, 23% reported being born in the U.S., and the remaining reported being born in various Latin American countries (El Salvador = 7%, Guatemala = 2%, Nicaragua = 2%, country not specified/did not report = 6%). Among the ADAD sample, 19 reported being born in Mexico, eight reported being born in the U.S., and two did not report their country of birth. Approximately 80% of participants were tested in Spanish. Of the subset of LALES participants who completed self-report questions for their language preference for speaking (n =149), 25 reported a preference for English, 39 reported both languages equally, and 85 preferred Spanish. Of the subset of participants who self-reported in what language they read (n = 141), 28 reported English, 23 reported both equally, and 90 reported Spanish. Among the subset of participants who self-reported the language in which they write better (n =97), 20 reported English, 18 reported both equally, and 59 reported Spanish.

Study 2 Sample Characteristics

Study 2 examined SENAS performance in the ADAD sample, which included ADAD mutation carriers and their non-carrier counterparts (i.e., persons from families with a history of ADAD mutations but determined not to have inherited the mutation).

Among ADAD carriers (n = 16), 13 were tested in Spanish, 12 reported being born and educated in Mexico and being first-generation in the U.S., and four reported being born and educated in the U.S. The median age among carriers was 34.5 years, and the median educational attainment was 12 years. The median age of familial dementia diagnosis was 43 years (range: 39–54 years). Of note, each individual is provided an estimated years from familial age of dementia diagnosis; however, we do not report the precise age for each participant’s family to preserve confidentiality. Regarding dementia ratings, 69% were asymptomatic (i.e., CDR score of 0), and 31% were mildly impaired (i.e., CDR score of 0.5). The median estimated years from familial age of dementia diagnosis was -10.5 years, meaning that, on average, ADAD carriers were 10.5 years younger than their familial median age of dementia diagnosis.

Among ADAD non-carriers (n = 13), nine were tested in Spanish, seven reported being born and educated in Mexico, four reported being born and educated in the U.S., and 11 reported being first-generation in the U.S. Two participants did not report their country of birth, country of education, and generation status. The median age of noncarriers was 39 years, and the median educational attainment was 13 years. CDR ratings among the non-carrier sample revealed that 85% were asymptomatic (i.e., CDR of 0) and 15% were mildly impaired (i.e., CDR of 0.5).

Table 2 presents group differences in performance across SENAS subtests for LALES and ADAD participants. The LALES participants performed significantly worse than the total ADAD sample (i.e., carriers and non-carriers together) on SENAS Word List Delayed Recall (t = -5.43, p < .001), Nonverbal Learning Delayed Recall (t = -2.80, p = .001), Category Fluency (t = -3.98, p <.001), and Phonemic Fluency (t = -6.50, p <.001). However, the LALES and ADAD participants did not differ significantly on SENAS Object Naming, Pattern Recognition, or Spatial Localization (all ps >.10).

Table 2.

Spanish English Neuropsychological Assessment Scale (SENAS) Raw Score Test Performance Characteristics

	LALES Sample (n = 202) M (SD) or %	ADAD Sample (n = 29) M (SD) or %	ADAD Carrier Sample (n = 16) M (SD) or %	ADAD Non-Carrier Sample (n = 13) M (SD) or %	t	p	Cohen’s D
Memory
Word List Learning Delayed Recall Raw	8.51 (3.27)	11.52 (2.71)	11.12 (2.96)	12.00 (2.38)	-5.43**	<.001	1.00
Nonverbal Learning Delayed Recall Raw	6.88 (2.62)	8.55 (2.95)	8.57 (2.85)	8.54 (3.18)	-2.80*	.001	.60
Language
Object Naming Raw	25.81 (6.15)	25.38 (4.55)	24.44 (4.69)	26.54 (4.27)	.46	.65	.07
Category Fluency Raw (Animals)	18.65 (4.97)	23.03 (5.62)	23.50 (4.16)	22.46 (7.17)	-3.98**	<.001	.82
Phonemic Fluency Raw (Letters F & L)	17.66 (8.06)	27.17 (7.25)	25.75 (6.44)	28.92 (8.06)	-6.50**	<.001	1.24
Visuospatial
Pattern Recognition Raw	31.48 (6.47)	33.10 (6.73)	32.00 (4.15)	34.46 (8.98)	-1.20	.24	.24
Spatial Localization Raw	15.57 (4.99)	17.31 (6.33)	16.62 (6.22)	18.15 (6.62)	-1.41	.17	.30

Note. t-tests were run on the full LALES and ADAD samples (i.e., carriers and non-carriers). ADAD = Autosomal Dominant Alzheimer’s Disease; LALES = Los Angeles Latino Eye Study; SENAS = Spanish English Neuropsychological Assessment Scale;

^*p < .05

^**p < .001

LALES SENAS Score Ranges: Word List Learning Delayed Recall (0–15), Nonverbal Delayed Recall (1–12), Object Naming (10–38), Category Fluency (5–34), Phonemic Fluency (1–45), Pattern Recognition (6–47), Spatial Localization (1–28)

ADAD Carrier SENAS Score Ranges: Word List Learning Delayed Recall (4–14), Nonverbal Delayed Recall (4–12), Object Naming (18–34), Category Fluency (18–30), Phonemic Fluency (9–38), Pattern Recognition (26–39), Spatial Localization (1–23)

ADAD Non-Carrier SENAS Score Ranges: Word List Learning Delayed Recall (9–15), Nonverbal Delayed Recall (4–12), Object Naming (20–34), Category Fluency (7–34), Phonemic Fluency (17–47), Pattern Recognition (17–47), Spatial Localization (1–25)

Correlational Analyses

Study 1

Table 3 presents correlations between SENAS subtests, chronological age, language, and education among LALES participants. SENAS subtest intercorrelations ranged from 0.17–0.83. Chronological age was significantly negatively correlated with Word List Delayed Recall (r = -.7, p <.01). Conversely, education was positively correlated with three SENAS subtests: Object Naming (r = .79, p <.01), Pattern Recognition (r = .78, p <.01), and Letter Fluency (r = .61, p <.01). Language was correlated with SENAS Object Naming (r = -.69, p <.01) and Pattern Recognition (r = -.69, p <.01) subtests such that English was more strongly correlated with performance on the two subtests than Spanish. There were no significant sex differences observed.

Table 3.

Pearson Correlations among SENAS and Demographic Variables in Los Angeles Latino Eye Study (LALES) Sample (n = 202).

Variable	1	2	3	4	5	6	7	8	9	10
1. Age (years)	--	--	--	--	--	--	--	--	--	--
2. Education (years)	-.08 [-.65, .54]	--	--	--	--	--	--	--	--	--
3. Sex	-.03 [-.62, .58]	-.38 [-.80, .28]	--	--	--	--	--	--	--	--
4. Language	-.21 [-.72, .45]	-.87** [-.97, -.56]	.18 [-.47, .70]	--	--	--	--	--	--	--
5. SENAS Word List DR	-.74** [-.93, -.25]	.06 [-.56, .64]	.35 [-.31, .79]	-.03 [-.62, .58]	--	--	--	--	--	--
6. SENAS Object Naming	-.30 [-.76, .37]	.79** [.36, .94]	-.55 [-.86, .07]	-.69* [-.91, -.15]	.17 [-.48, .70]	--	--	--	--	--
7.SENAS Nonverbal DR	-.56 [-.87, .06]	.46 [-.19, .83]	-.14 [-.68, .50]	-.41 [-.81, .25]	.60 [.00, .88]	.58 [-.04, .87]	--	--	--	--
8.SENAS Pattern Recog.	-.30 [-.76, .36]	.78** [.33, .94]	-.49 [-.84, .15]	-.69* [-.91, -.16]	.28 [-.38, .76]	.83** [.46, .95]	.62* [.03, .89]	--	--	--
9.SENAS Spatial Loc.	-.41 [-.81, .25]	.57 [-.05, .87]	-.41 [-.81, .25]	-.51 [-.85, .13]	.40 [-.26, .81]	.62* [.03, .89]	.67* [.12, .91]	.76** [.30, .94]	--	--
10.SENAS Category Fluency	-.60 [-.88, .00]	.36 [-.30, .79]	-.39 [-.80, .28]	-.20 [-.72, .45]	.47 [-.18, .83]	.58 [-.03, .88]	.53 [-.10, .86]	.49 [-.15, .84]	.48 [-.16, .84]	--
11.SENAS Letter Fluency	-.25 [-.74, .41]	.61* [.02, .89]	-.40 [-.81, .26]	-.43 [-.82, .23]	.20 [-.45, .72]	.60* [.01, .88]	.33 [-.34, .78]	.58 [-.03, .88]	.45 [-.20, .83]	.52 [-.11, .86]

Note. Values in square brackets indicate the 95% confidence interval for each correlation. The confidence interval is a plausible range of population correlations that could have caused the sample correlation (Cumming, 2014). DR = Delayed Recall; Loc. = Localization; Recog. = recognition; SENAS = Spanish English Neuropsychological Assessment Scale

^*indicates p < .05.

^**indicates p < .01.

In follow-up exploratory analyses, language preference for speaking was significantly correlated with the following SENAS subtests: Object Naming (r = -.14, p = .006), Pattern Recognition (r = -.18, p = .004), and Spatial Localization (r = -.22, p = .003). Language preference for reading was significantly correlated with the following SENAS subtests: Object Naming (r = -.31, p = .001), Pattern Recognition (r = -.12, p = .006), Spatial Localization (r = -.27, p = .001). Language preference for writing was significantly correlated with the following SENAS subtests: Object Naming (r = -.31, p = .002), Pattern Recognition (r = -.26, p = .003), Spatial Localization (r = -.41, p = .0006). Results revealed that greater Spanish language preference for speaking, reading, and writing was correlated with worse performance on Object Naming, Pattern Recognition, and Spatial Localization.

Study 2

Table 4 presents correlations between SENAS subtests, estimated years from familial age of dementia diagnosis, education, language, and acculturation among ADAD carriers (n = 16). SENAS subtest intercorrelations ranged from 0.05 –0.87. Chronologic age and estimated years from familial age of dementia diagnosis were significantly positively correlated (r = 0.87, p <.01). Estimated years from familial age of dementia diagnosis significantly positively correlated with language (r = .74, p <.01) and Latine Orientation (r = .88, p <.01) but negatively correlated with Total Acculturation (r = -.77, p <.01) and Anglo Orientation (r = -.57, p <.01) and sex (r = -.57, p <.01), such that participants in the sample with greater estimated years from familial age of dementia diagnosis (i.e., closer to familial age of dementia diagnosis) were more likely to be highly acculturated to Latine culture, speak Spanish, and be female.

Table 4.

Pearson Correlations among SENAS Variables and Demographic and Cultural Variables in ADAD Carrier Sample (n = 16).

Variable	M	SD	1	2	3	4	5	6	7	8
1. Age	0.01	0.44
2. Estimated Years from Familial Dementia Diagnosis	−0.04	0.54	.87** [.65, .96]
3. Education	0.17	0.30	−.06 [−.56, .46]	.04 [−.48, .54]
4. Sex	0.14	0.35	−.55* [−.83, −.05]	−.61* [−.86, −.14]	−.11 [−.59, .43]
5. Language	−0.11	0.46	.26 [−.29, .68]	.67** [.25, .88]	.04 [−.48, .54]	−.42 [−.77, .11]
6. ARSMA-II MOS	−0.03	0.53	.57* [.08, .84]	.85** [.61, .95]	.08 [−.45, .57]	−.31 [−.71, .24]	.85** [.59, .95]
7. ARSMA-II AOS	0.12	0.44	−.19 [−.64, .36]	−.54* [−.82, −.03]	.10 [−.44, .58]	.21 [−.34, .65]	−.81** [−.93, −.51]	−.79** [−.93, −.47]
8. ARSMA-II Acc. Level	0.12	0.53	−.38 [−.75, .16]	−.74** [−.91, −.36]	−.01 [−.52, .50]	.35 [−.20, .73]	−.91** [−.97, −.74]	−.93** [−.98, −.80]	.95** [.85, .98]
9. SENAS Word List DR	0.17	0.47	−.74** [−.91, −.37]	−.86** [−.95, −.63]	.37 [−.17, .74]	.36 [−.19, .73]	−.68** [−.88, −.26]	−.83** [−.94, −.54]	.65** [.20, .87]	.76** [.40, .92]
10. SENAS Object Naming	0.13	0.42	−.35 [−.73, .20]	−.72** [−.90, −.34]	−.22 [−.66, .33]	.27 [−.28, .69]	−.87** [−.96, −.65]	−.87** [−.95, −.64]	.68** [.26, .89]	.83** [.55, .94]
11. SENAS Nonverbal DR	0.12	0.53	−.72** [−.90, −.33]	−.89** [−.96, −.71]	.17 [−.38, .63]	.46 [−.07, .79]	−.80** [−.93, −.49]	−.89** [−.96, −.71]	.64* [.19, .87]	.81** [.50, .93]
12. SENAS Pattern Recog.	0.04	0.36	−.51 [−.81, .01]	−.39 [−.75, .15]	−.15 [−.62, .39]	.61* [.14, .85]	.11 [−.43, .59]	.07 [−.46, .56]	−.38 [−.75, .16]	−.19 [−.64, .36]
13. SENAS Spatial Loc.	0.09	0.50	−.78** [−.92, −.44]	−.95** [−.98, −.84]	−.26 [−.68, .29]	.62* [.16, .86]	−.71** [−.90, −.31]	−.87** [−.96, −.65]	.52* [.01, .81]	.74** [.36, .91]
14. SENAS Category Fluency	0.19	0.32	−.36 [−.73, .19]	−.45 [−.78, .09]	−.06 [−.55, .47]	.73** [.35, .90]	−.38 [−.75, .16]	−.15 [−.61, .39]	−.06 [−.56, .46]	.11 [−.42, .59]
15. SENAS Phonemic Fluency	0.09	0.35	.48 [−.04, .80]	.52* [.02, .82]	.54* [.04, .82]	−.60* [−.85, −.13]	.14 [−.40, .61]	.31 [−.24, .71]	−.21 [−.65, .34]	−.30 [−.70, .25]

Note. Values in square brackets indicate the 95% confidence interval for each correlation. The confidence interval is a plausible range of population correlations that could have caused the sample correlation (Cumming, 2014). Acc. = Acculturation; ADAD = Autosomal Dominant Alzheimer’s Disease; AOS = Anglo Orientation Scale; ARSMA-II = Acculturation Rating Scale for Mexican Americans – II; DR = delayed recall; Estimated Years from Familial Dementia Diagnosis = estimated number of years away participant’s age is relative to their familial mean age of dementia diagnosis; Loc. = Localization; MOS= Mexican Orientation Scale; Recog. = Recognition; SENAS = Spanish English Neuropsychological Assessment Scale;

^*indicates p < .05.

^**indicates p < .01.

Regarding SENAS correlations, estimated years from familial age of dementia diagnosis was significantly correlated with Word List Delayed Recall (r = -.84, p < .01), Object Naming (r = -.77, p < .01), Nonverbal Memory Delayed Recall (r = -.88, p <.01), and Spatial Localization (r = -.94, p <.01). Education was only significantly associated with SENAS Letter Fluency subtest (r = .58, p <.01).

Language significantly positively correlated with Latine Orientation (r = .86, p < .01) but negatively correlated with Anglo Orientation (r = -.81, p < .01) and Total Acculturation (r = -.91, p < .01). Language was significantly negatively correlated with the following SENAS subtests: Word List Delayed Recall (r = -.75, p < .01), Object Naming (r = -.87, p < .01), Nonverbal Memory Delayed Recall (r = -.86, p <.01), and Spatial Localization (r = -.75, p <.01), such that Spanish language was related to worse performance on the subtests.

Latine Orientation negatively correlated with Word List Delayed Recall (r = -.85, p < .01), Object Naming (r = -.87, p < .01), Nonverbal Memory Delayed Recall (r = -.91, p <.01), and Spatial Localization (r = -.88, p <.01). Conversely, Anglo Orientation positively correlated with Word List Delayed Recall (r = .69, p < .01), Object Naming (r = .68, p < .01), Nonverbal Memory Delayed Recall (r = .67, p <.01), and Spatial Localization (r = .53, p <.01). Total acculturation level (Anglo Orientation – Latine Orientation) significantly positively correlated with Word List Delayed Recall (r = .79, p < .01), Object Naming (r = .83, p < .01), Nonverbal Memory Delayed Recall (r = .83, p <.01), and Spatial Localization (r = 75, p <.01), such that greater acculturation to Anglo (U.S.) culture was related to better performance on the selected SENAS subtests.

Although not a focus of the present study, sex negatively correlated with SENAS Letter Fluency (r = -.58, p <.01) but positively correlated with Pattern Recognition (r = .57, p <.01), Spatial Localization (r = .58, p <.01), and Category Fluency (r = .72, p <.01), such that women performed better on Pattern Recognition, Spatial Localization, and Category Fluency subtests whereas men performed slightly better on Letter Fluency.

Table 5 presents correlations in the ADAD non-carrier sample (n = 13) for SENAS subtests, chronological age, education, language, and acculturation. SENAS subtest intercorrelations ranged from 0.05 –0.91. Chronological age negatively correlated with education (r = -.58, p <.01), Anglo Orientation (r = -.64, p <.01), and Total Acculturation (r = -.69, p <.01) but positively correlated with Latine Orientation (r = .70, p <.01). Chronological age was significantly negatively associated with the following SENAS subtests: Word List Delayed Recall (r = -.69, p <.01), Object Naming (r = -.84, p <.01), and Nonverbal Delayed Recall (r = -.54, p <.01). Education was significantly associated with SENAS Word List Delayed Recall (r = .78, p <.01) and Object Naming (r = .76, p <.01).

Table 5.

Pearson Correlations among SENAS Variables and Demographic and Cultural Variables in ADAD Non-Carrier Sample (n = 13).

Variable	1	2	3	4	5	6	7
1. Age
2. Education	-.80** [-.93, -.47]
3. Sex	-.16 [-.64, .41]	.20 [-.37, .66]
4. Language	.65* [.18, .88]	-.53 [-.83, .00]	-.04 [-.56, .50]
5. ARSMA-II MOS	.70** [.27, .90]	-.47 [-.80, .08]	-.16 [-.64, .41]	.87** [.62, .96]
6. ARSMA-II AOS	-.64* [-.87, -.16]	.52 [-.02, .82]	.05 [-.49, .56]	-1.00** [-1.00, -.98]	-.86** [-.96, -.61]
7. ARSMA-II Acc. Level	-.69** [-.89, -.25]	.49 [-.06, .81]	.15 [-.41, .63]	-.96** [-.99, -.87]	-.96** [-.99, -.88]	.96** [.87, .99]
8. SENAS Word List DR	-.69** [-.90, -.26]	.78** [.43, .93]	.39 [-.17, .76]	-.50 [-.82, .04]	-.70** [-.90, -.27]	.50 [-.04, .81]	.59* [.08, .85]
9. SENAS Object Naming	-.84** [-.95, -.55]	.76** [.38, .92]	.07 [-.48, .58]	-.44 [-.79, .11]	-.54* [-.83, -.01]	.43 [-.14, .78]	.46 [-.10, .80]
10. SENAS Nonverbal DR	-.54* [-.83, -.02]	.31 [-.26, .72]	.06 [-.49, .57]	-.06 [-.57, .49]	-.45 [-.79, .11]	.04 [-.50, .56]	.24 [-.33, .68]
11. SENAS Pattern Recog.	-.22 [-.67, .35]	.16 [-.41, .63]	.25 [-.32, .69]	.39 [-.18, .76]	-.05 [-.57, .49]	-.39 [-.76, .18]	-.18 [-.65, .39]
12. SENAS Spatial Loc.	-.43 [-.78, .13]	.24 [-.33, .69]	.03 [-.50, .56]	.27 [-.30, .70]	-.07 [-.58, .48]	-.30 [-.72, .27]	-.13 [-.62, .43]
13. SENAS Category Fluency	-.15 [-.63, .42]	.15 [-.41, .63]	.37 [-.20, .75]	.47 [-.07, .80]	.09 [-.46, .59]	-.48 [-.81, .06]	-.30 [-.72, .27]
14. SENAS Phonemic Fluency	.05 [-.49, .57]	.19 [-.38, .66]	.66* [.19, .88]	.39 [-.18, .76]	.43 [-.13, .78]	-.40 [-.77, .17]	-.39 [-.76, .17]

Note. M and SD are used to represent mean and standard deviation, respectively. Values in square brackets indicate the 95% confidence interval for each

correlation. The confidence interval is a plausible range of population correlations that could have caused the sample correlation (Cumming, 2014).

^*indicates p < .05.

^**indicates p < .01.

Language significantly positively correlated with Latine Orientation (r = .87, p <.01) but negatively correlated with Anglo Orientation (r = -1.0, p <.01) and Total Acculturation (r = -.96, p <.01), such that Spanish language was related to greater Latine orientation. In contrast, English language positively correlated with greater Anglo Orientation and Total Acculturation. Greater Latine Orientation negatively correlated with SENAS Word List Delayed Recall (r = -.70, p <.01) and Object Naming (r = -.54, p <.01). Anglo Orientation was not significantly correlated with any SENAS subtest. However, total Acculturation positively correlated with Word List Delayed Recall (r = .59, p <.01), such that greater Anglo orientation relative to Latine orientation was related to better delayed recall performance.

ADAD Subgroup SENAS Performance

Independent samples t-tests showed that the mutation carriers within 10 years of familial age of dementia onset and non-carriers differed significantly on Word List Delayed Recall (t(19) = -2.33, p = 0.03), with a large effect size (d=1.04, 95% CI[0.09, 1.97]), and Object Naming (t(19) = -2.43, p = 0.03), with a large effect size (d=1.09, 95% CI[0.13, 2.02]). Modest, albeit non-statistically significant, differences were observed in Spatial Localization performance (t(19) = -1.65, p = 0.11), with a medium to large effect size d=0.74, 95% CI[-0.18, 1.64]).

Study 3 - Exploratory Analyses

Group Comparisons Between LALES and ADAD Samples

Partial correlations (Table 6) were estimated to examine whether estimated years from familial age of dementia diagnosis, holding constant effects of education and chronological age, more strongly correlated with SENAS performance among the ADAD mutation carrier sample compared to the correlation between chronological age, holding constant effects of education effects, and SENAS performance among the LALES older adult sample. Results revealed stronger correlations among the ADAD sample on Nonverbal Memory Delay (0.45 in ADAD mutation carriers vs. 0.25 in LALES sample), Word List Delay (0.45 in ADAD mutation carriers vs. 0.36 in LALES sample), and Object Naming (0.50 in ADAD mutation carriers vs. 0.24 in LALES sample). One test, Category Fluency, had a stronger correlation in the LALES group (0.35) compared to the ADAD mutation carrier sample (0.19).

Table 6.

Partial correlations for relationships between chronological age (LALES) and estimated years from familial dementia diagnosis (ADAD) on SENAS subtests in LALES (n = 202) ADAD carrier sample (n =16), adjusted for education (both samples) and chronological age (ADAD carrier sample only)

	LALES Sample (n = 202)			ADAD Carrier Sample (n = 16)
	r	95% CI	p	r	95% CI	p
Nonverbal Memory Delay	.25	[0.12, 0.38]	<0.001	.45	[−0.08, 0.78]	0.09
Word List Delay	.36	[0.24, 0.48]	<0.001	.45	[−0.05, 0.77]	0.08
Pattern Recognition	.10	[−0.05, 0.25]	0.20	.20	[−0.31, 0.62]	0.44
Spatial Localization	.06	[−0.09, 0.20]	0.45	.41	[−0.08, 0.75]	0.10
Category Fluency	.35	[0.22, 0.46]	<0.001	.19	[−0.32, 0.62]	0.47
Phonemic Fluency	.05	[−0.09, 0.19]	0.45	.26	[−0.25, 0.66]	0.32
Object Naming	.24	[0.11, 0.37]	<.001	.50	[0.02, 0.80]	0.04

Note. ADAD = Autosomal Dominant Alzheimer’s Disease; CI = confidence interval; Estimated years from familial dementia diagnosis = estimated number of years away participant’s age is relative to their familial mean age of dementia diagnosis; r = Pearson’s r partial correlation coefficient adjusted for education in both examples to control for effects, chronological age adjusted for ADAD sample only to examine effects of estimated years from familial dementia diagnosis above and beyond effects of chronological age; SENAS = Spanish English Neuropsychological Assessment Scale

Discussion

This study was comprised of three components: 1) Characterizing SENAS performance and associations with chronological age, education, and language in a community-based sample of older adult Latines living in the greater Los Angeles area; 2) Exploring properties of the SENAS in adult Latines at risk for ADAD by examining the effects of estimated years from familial age of dementia diagnosis, education, language, and acculturation among carriers, and the effects of chronologic age, education, language, and acculturation among ADAD non-carriers; 3) Conducting exploratory analyses to examine differences in SENAS performance between ADAD carriers and our older adult sample, as well as differences in SENAS performance between carriers within ten years of their estimated years from familial age of dementia diagnosis and their non-carrier counterparts. This study adds a Los Angeles community-based older adult Latine sample to the extant SENAS literature and is the first to report the use of the SENAS among Latines at risk for ADAD.

Study 1 - Older Adult Latine Sample (LALES)

Among the LALES sample, greater chronological age was related to worse verbal delayed recall, findings consistent with prior research on cognitive domains impacted by aging (Murman, 2015). Language fluency and naming subtests were positively associated with education, consistent with the literature highlighting the role of education in lexical access ability, a cognitive function necessary for the successful completion of the task (Shao et al., 2014). Language was correlated with tests of object naming and visuospatial skills, such that participants who were given the Spanish version of the SENAS performed worse than those who were given the English version. Although only available for a partial sample of LALES participants, language preferences for speaking, reading, and writing were each associated with tests of object naming and visuospatial skills. Cognitive tests designed to measure one domain (e.g., visuospatial skills) may recruit multiple cognitive functions required to complete the task. Prior research suggests that processing visuospatial expressions implicates dorsolateral brain regions, which are involved in both visuospatial perception and the language used to refer to the perceived space or object, and may explain, at least in part, the findings of language correlations with tests that involve visual information (Rocca et al., 2020).

Study 2 – ADAD Sample

ADAD Carriers

Among ADAD carriers, estimated years from familial age of dementia diagnosis was significantly associated with verbal and nonverbal delayed recall, object naming, and visuospatial subtests, supporting our hypothesis. This finding is consistent with prior ADAD research using other neuropsychological measures (Medina et al., 2021; Ringman, 2005; Ryman et al., 2014). Episodic memory subtest findings are consistent with prior literature (Caselli & Reiman, 2014; Parnetti et al., 2019; Rabinovici, 2019), but the additional finding of a visuospatial subtest is also important, as atypical presentations affecting nonmemory domains are also reported in about one-third of ADAD or non-familial early onset AD forms (Balasa et al., 2011; Koedam et al., 2010; Mendez, 2012). Notably, estimated years from familial age of dementia diagnosis was quite strongly correlated with chronological age among ADAD carriers. Moreover, the estimate appears to have an incremental increase across all cognitive outcomes above and beyond the effects of chronologic age (see Table 4). Unfortunately, due to our limited sample size, we were insufficiently powered to test incremental increase through partial correlations controlling for chronological age. Nonetheless, this is an important avenue for future research, as this estimate may be a useful indicator in staging preclinical decline among ADAD carriers.

Years of education were only significantly positively correlated with a verbal fluency task, consistent with findings in the older adult sample. Consistent with prior research on the SENAS in older adult samples, the Spanish version of the SENAS was correlated with lower scores on SENAS subtests than the English version (Mungas et al., 2005a). Further, greater Latine orientation, a measure of acculturation, was related to worse verbal and nonverbal memory, object naming, and aspects of visuospatial skills.

Conversely, greater total acculturation to Anglo culture relative to one’s Latine culture was related to better performance on measures of verbal and nonverbal memory, object naming, and visuospatial skills. It is important to note that the acculturation measure used in the present study accounts for a significant portion of language-based acculturative preferences (e.g., language preference for watching TV). Lower performance related to Spanish language and greater endorsement of Latine acculturation, thus, do not reflect innate differences in cognitive capacity. Importantly, variations by test language version may reflect inherent differences in language structure (e.g., syllabic differences for objects and digits; Mungas et al., 2000; Rosenstein et al., 2023) which impact timed tasks.

Overall, these findings suggest that SENAS measures of verbal and nonverbal memory, object naming, and visuospatial skills may be helpful in detecting early impairment among a largely asymptomatic ADAD carrier sample.

ADAD Non-Carriers

Among the ADAD non-carriers, chronological age was significantly negatively associated with verbal and nonverbal recall and object naming. Estimated years from familial age of dementia diagnosis was not used in this sample because non-carriers would not develop dementia at the ages determined by the mutation carriers of their relatives because they do not carry the genetic mutation. Similar to findings in the carrier sample, greater Latine orientation was correlated with worse performance on verbal delayed recall and object naming. In contrast, total acculturation and Anglo orientation were related to better verbal delayed recall performance.

Mutation carriers within ten years of their estimated familial age of dementia diagnosis performed significantly worse on verbal delayed recall and object naming measures. In addition, a trend-level association was observed on a measure of visuospatial skills. Thus, although preliminary and limited by small sample size, these SENAS subtests may help distinguish carriers from non-carriers, though more research is needed.

Exploratory Study 3

A unique contribution of our study is the exploratory comparison between relationships between estimated years from familial age of dementia diagnosis and SENAS subtests among confirmed ADAD carriers and chronologic age and the SENAS subtests in an older adult, community-based Latine sample (LALES). Stronger associations were observed in the ADAD carrier sample compared to the older adult community sample on measures of verbal and nonverbal memory recall and object naming, consistent with the hypothesis that these domains are more specifically affected in AD than with aging in general. Although the correlation between estimated years from familial age of dementia diagnosis and chronologic age among mutation carriers was strong, the association between estimated years from familial age of dementia diagnosis and SENAS, holding constant chronologic age, still correlated with SENAS performance. Estimated years from familial age of dementia diagnosis, thus, may be a useful metric that may be a valuable estimator for staging early cognitive changes.

Limitations

The present study has several limitations. First, ADAD mutations are rare, accounting for less than 1% of AD cases. Therefore, our ADAD sample sizes were small, with a high proportion of carriers of the A431E PSEN1 mutation, potentially limiting generalizability to other ADAD mutations and exploring variation in performance by mutation type. Future research should replicate the current findings in a larger ADAD sample. Additionally, study inclusion criteria limited participants with a CDR score of less than 1 to examine early markers of cognitive changes before receiving a symptomatic dementia rating. Therefore, a larger sample including more symptomatic and cognitively impaired ADAD carriers may demonstrate stronger correlations between estimated years from familial age of dementia diagnosis and other SENAS subtests.

Second, the median educational attainment of the LALES sample was significantly lower than the ADAD mutation carrier sample. While we attempted to control for effects of education in comparative analyses between the ADAD sample and LALES samples, it is important to acknowledge the effects of education on test performance. Lower educational attainment is associated with worse cognitive test performance, partly due to a lack of exposure to writing, reading, and comprehension skills required for optimal standardized test performance typically achieved in formal educational settings. (e.g., lexical ability, categorization, and generalization processes; Ardila, 2005; Brigola et al., 2019). However, the representation of individuals with less education is also a strength of this study, as much of the research to date has been carried out on more highly educated populations with poor Latine representation. Although education has been demonstrated to be protective among non-Latine Whites, disparities in educational access and quality among minoritized communities may reduce cognitive reserve and elevate dementia risk (Arce Rentería et al., 2019; Avila et al., 2021). More research is needed to better characterize dementia courses among Latine patients with lower educational quality and literacy levels.

Third, although our older adult community sample (LALES) performed worse on the SENAS than our ADAD carrier sample, this may be a function of the significant age difference and the effects of age-related comorbidities on cognition (e.g., diabetes, hypertension). As this was a community-based sample that included persons with health conditions, including vascular risk factors, we cannot rule out confounding influences of such factors on cognition in the older adult sample (Mayeda et al., 2013). Further, specific documentation of or a formal diagnosis of dementia or depression was not a component of the LALES ancillary study. Thus, we were unable to consider clinical dementia ratings as part of our inclusion criteria for the LALES sample. Though no LALES participants were known to have overt dementia, it is unknown whether our sample contained individuals with mild degrees of acquired cognitive impairment or current levels of depressive symptomatology that may have impacted performance on the SENAS. However, informant questionnaires were completed by knowledgeable family members (Informant Questionnaire on Cognitive Decline in the Elderly; IQCODE), which revealed a median value of 3.03, indicating no cognitive or functional change observed by the informant. Further, IQCODE scores did not significantly correlate with any of the SENAS subtests. We acknowledge this is not a substitute for a CDR score or formal diagnosis. However, scores provide some insight into the perceived cognitive and functional status of LALES participants according to their respective informants. Nevertheless, we are limited in our exploratory findings and caution against overinterpreting group differences observed in the above exploratory analyses between ADAD and LALES participants.

Finally, while we considered some aspects of cultural influences on test performance, including acculturation and language, comprehensive measures of socioeconomic status and other sociocultural variables (e.g., acculturation) were not collected as a part of the LALES study and only have been more recently introduced to the ADAD cohort. Therefore, we are limited in characterizing other potential cultural influences on SENAS performance in our ADAD and LALES samples. However, prior research has reported that the effects of acculturation on SENAS performance were not statistically significant after controlling for the language of test administration and education (Mungas et al., 2005a). Nevertheless, we note that controlling for broad variables like education and language does not address specific cultural factors such as acculturation that were available in the ADAD sample. A related limitation is that the majority of our participants were of Mexican origin, but our sample also consisted of other diverse Latine heritage groups (El Salvador, Guatemala, Puerto Rico). Thus, it is important to acknowledge the potential for differences in task performance can vary across Latine heritage groups, including language-related tasks. However, the extent to which samples from different subcultures differ in performance has not been well-researched in the literature. Although many participants are from Spanish-speaking backgrounds, there is significant variation in language across Latine heritage groups and cultural differences that may impact performance on the SENAS.

Unfortunately, due to small sample size, particularly for participants from non-Mexican origins, we were unable to meaningfully characterize potential differences in performance on the SENAS by subgroup or account for linguistic and cultural differences across Latine heritage groups. Therefore, our results may not represent a larger sample of other Latine heritage groups. We caution against extending these findings to other populations that could not be as well-characterized as our Mexican-origin majority sample. In addition, limitations inherent to the acculturation measure should be noted, which primarily assessed language preferences (e.g., reading or thinking in Spanish vs. English) or broad identification questions (e.g., “My father identifies as X”), further hindering our ability to contextualize cultural nuances for participants from different Latine heritage groups. Therefore, we would like to emphasize the importance of accounting for heterogeneity in Latine heritage groups in future research, perhaps through more nuanced acculturation measures and/or qualitative methods to provide a more comprehensive picture of participants’ diverse linguistic and cultural perspectives.

Overall, our findings suggest that SENAS subtests of verbal and non-verbal delayed recall, object naming, and visuospatial skills may be useful in identifying and tracking cognitive changes in ADAD and are similar to the subtests most impacted by the effects of aging observed in the older adult Latine population. In addition, the non-verbal memory subtest showed the strongest correlation with estimated years from familial age of dementia diagnosis among ADAD mutation carriers, which may have utility across language versions, though more research is needed. Future directions include longitudinal exploration of SENAS score profiles over time in ADAD carriers and non-carriers, examining SENAS performance with neuroimaging, and comparison with an older adult sample with confirmed AD diagnoses to examine whether patterns of cognitive decline affect similar or distinct domains in ADAD.