Stability of Subjective Executive Functioning in Older Adults with aMCI and Subjective Cognitive Decline

Abstract

Objective

Subjective memory concerns are characteristic of individuals with amnestic mild cognitive impairment (aMCI) and subjective cognitive decline (SCD), though subjective changes in executive functions have also been reported. In a cohort study, we examined the temporal stability of subjective report of executive functioning in a high education (mean = 16.8 years) sample of cognitively normal (CN) older adults and those with aMCI or SCD.

Method

Participants (CN, n = 22; aMCI, n = 21; SCD, n = 24) and their informants completed the BRIEF-A and neuropsychological tests at two time points separated by approximately 1 year.

Results

Analyses focused on those with diagnostic stability (95.7%). Participants with aMCI and SCD, and their informants, endorsed worse executive functions relative to CN at both time points. No group by time interaction was observed for subjective or objective measures of executive function.

Conclusions

Diagnostically stable CN older adults, and those with prodromal dementia conditions, report stable executive functioning at 1-year follow-up.

Introduction

Amnestic mild cognitive impairment (aMCI) is typically characterized by objective memory deficit and subjective memory concerns in the context of generally independent daily functioning (Jack et al., 2018). Subjective concerns are the focus of considerable attention as a possible predictor of conversion to MCI and dementia (Jessen et al., 2011; Reisberg, Shulman, Torossian, Leng, & Zhu, 2010; Mitchell, Beaumont, Ferguson, Yadegarfar, & Stubbs, 2014). Findings are inconsistent, however, regarding the temporal stability of subjective memory concerns in aMCI and those with subjective cognitive decline (SCD; i.e., no impairment on neuropsychological tests despite self-report of cognitive worsening). Some evidence suggests that memory concerns become less stable as individuals transition from aMCI to Alzheimer’s disease, although studies have often been cross-sectional with comparisons of aMCI to other cognitive groups or to informant ratings at a single time point (Eschen, Martin, Gasser, & Kliegel, 2009; Buckley et al., 2015). Other evidence suggests that stable SCD (as compared to unstable SCD) increases the risk of converting to MCI (subtype unspecified; Roehr, Villringer, Angermeyer, Luck, & Riedel-Heller, 2016). Longitudinal data over approximately 5 years suggested that stability of memory concerns among individuals classified as SCD at baseline was associated with increased risk for transition to aMCI or dementia as compared to those with unstable memory concerns across time points (Wolfsgruber et al., 2016).

Not surprisingly, research on subjective cognitive concerns in older adults has largely focused on the memory domain (Rabin et al., 2015). Individuals with aMCI and SCD, however, also present with concerns about other aspects of cognition (Saunders & Summers, 2010). For example, older adults with aMCI or SCD have reported greater problems with executive functioning than cognitively normal (CN) older adults, especially regarding working memory (Rabin et al., 2006). A similar pattern of group differences was seen on informant report in the study, although those with aMCI or SCD reported greater difficulty than observed by informants. Furthermore, there were no significant relationships between Behavior Rating Inventory of Executive Function for Adults (BRIEF-A) ratings and objective measures of executive functioning. Longitudinal data on the stability of cognitive concerns in aMCI and SCD is limited, particularly for non-memory domains such as executive functioning. Such information may be clinically relevant given that prior research has found that greater executive function concerns by older adults and their informants at baseline predicted conversion to MCI (subtype unspecified) and greater functional disability (Farias et al., 2017).

Although subjective executive functioning concerns frequently appear alongside memory concerns and may help predict decline, it remains unknown whether the executive function concerns remain stable when there is no significant change in objective cognitive status. Examining the consistency of executive functioning concerns may allow for more accurate predictions of decline, which could enhance long-term clinical planning and design of research studies. In the present study, we evaluated the temporal stability of subjective executive functioning in CN older adults and those with aMCI and SCD, over a period of approximately 1 year. We hypothesized that both participants and informants would report similar executive functioning at both time points.

Method

Participants

Participants initially included 21 individuals with aMCI, 27 with SCD, and 22 CN enrolled in the Dartmouth Memory and Aging study, a longitudinal investigation of preclinical and early dementia. Participants were recruited from flyers, newspaper ads, public lectures, and referrals from medical center clinics. Participants completed a comprehensive screening process: medical chart review, neuropsychological evaluation, structural MRI, and standardized phone interview with memory screen (Saykin et al., 2006). Inclusion criteria were at least 60 years old, right-handed, fluent in English, and a minimum of 12 years of education. Exclusion criteria comprised any uncontrolled or confounding medical, psychiatric, or neurological condition (other than aMCI), history of head trauma with more than a five-minute loss of consciousness, history of substance use disorder, or current clinical depression as determined by a board-certified geriatric psychiatrist. Data from a subset of participants were included as part of a prior paper on subjective executive functioning in aMCI and SCD (Rabin et al., 2006), with baseline visits completed from 2004 to 2006. For the present study, baseline visits were completed from 2004 to 2007 and follow-up visits from 2005 to 2009. As detailed in this initial study, all participants had a knowledgeable informant to answer questions about their cognition and general health. After providing a complete description of the study, written informed consent was obtained following a protocol approved by the Dartmouth College Committee for the Protection of Human Subjects.

Measures

The BRIEF-A includes 75 non-overlapping items and yields 9 theoretically and empirically derived and well-validated clinical scales that measure domains of executive functioning: Inhibit, Shift, Emotional Control, Self-Monitor, Initiate, Working Memory, Plan/Organize, Task-Monitor, and Organization of Materials. The clinical scales compose two indices, Behavior Regulation Index (BRI) and Metacognition Index (MI), as well as an overall summary score, the Global Executive Composite (GEC). Higher T-scores indicate greater difficulty. A T score ≥ 65 is considered clinically meaningful (Roth, Isquith, & Gioia, 2005). The BRIEF-A was not used to inform diagnostic classification. Mood was assessed using an adjusted 26-item Geriatric Depression Scale (GDS) with cognitive items removed (excluded because they were utilized in an index score to classify SCD). Premorbid IQ was assessed via the American National Adult Reading Test (ANART).

Performance-based neuropsychological measures were also completed at both time points, including: the Mini-Mental State Examination (MMSE); maximal spans for Digit Span Forward and Backward from the Wechsler Adult Intelligence Scale, Third Edition (WAIS-III); raw score for total trials 1–5, short-delay (SD) and long-delay (LD) recall, as well as recognition discriminability from the California Verbal Learning Test, Second Edition (CVLT-II); time to complete condition 4 of Trail Making from the Delis–Kaplan Executive Function System (D-KEFS); categories completed and number of perseverative errors on the Wisconsin Card Sorting Test—64 (WCST); and total raw score for the Boston Naming Test (BNT).

Statistical Analyses

Data were analyzed using analysis of variance (ANOVA). Mixed-model repeated measures ANOVA comparing groups at baseline (Time 1) and 1-year follow-up (Time 2) were conducted with the nine scales, the two indices (BRI, MI), and the summary score (GEC) of the BRIEF-A, as well as with the performance-based neuropsychological measures. Post-hoc analyses were conducted using Tukey’s least significant difference (LSD) statistic. The Benjamini-Hochberg procedure (Benjamini & Hochberg, 1995) was used to decreases the false discovery rate (FDR) for the neuropsychological tests and the BRIEF-A. This procedure helps maintain a low probability of Type I error while also identifying as many significant comparisons as possible. Calculations were done using the R function p.adjust (with method = “BH”) in R version 3.6.0 (R Development Core Team, 2017), with the FDR set to allow for no more than 5% false positives. For the neuropsychological tests, 33 statistical comparisons were conducted using ANOVA, with an additional 42 post-hoc analyses, for a total of 75 tests. For the BRIEF-A, 72 statistical comparisons were conducted using ANOVA, with an additional 111 post-hoc analyses, for a total of 183 tests. Alpha level was set at p = .05.

Results

Participant Characteristic Data

Table 1 presents the participant characteristic data. Between the first and second time points, two SCDs participants transitioned to aMCI, while a third SCD transitioned to an atypical presentation (unable to be classified). In order to focus on the majority with stable diagnoses, these three participants were removed from the dataset, leaving 24 SCD older adults. There were no significant group differences for age, gender, years of education, adjusted GDS scores, premorbid IQ, or time elapsed between assessments. The mean interval between assessments was 420 days (range 237–828). Consistent with the demographics of northern New England at the time of data collection, the overall sample was predominantly European American, with one Hispanic and one Asian American participant.

Performance-Based Neuropsychological Measures

Table 1 presents the neuropsychological test data with p-values adjusted by FDR (i.e., q-values) for comparisons across groups at Time 1, Time 2, and the change from Time 1 to Time 2. Post-hoc analyses indicated that, at Time 1, the aMCI group performed worse than both the SCD and CN groups on the MMSE (q = 0.002) and CVLT-II (Total, SD, LD, and discriminability; all q = 0.002), and worse than the CN group on Trail Making (q = 0.01). These same group differences were observed at Time 2 for the MMSE (q = 0.002) and CVLT-II (Total, SD, LD, and discriminability; q = 0.002). At Time 2, the aMCI group was slower on Trail Making than both the CN (q = 0.002) and SCD (q = 0.005) groups. Furthermore, the aMCI group performed worse relative to the SCD group for Digit Span Backwards (q = 0.015) and BNT (q = 0.015). No group by time interaction was observed for any of the performance-based measures.

Subjective Executive Function

Table 2 shows BRIEF-A self-report data with q-values for comparisons across the three groups at Time 1, Time 2, and the change from Time 1 to Time 2. Post-hoc analyses indicated that, at Time 1, worse executive function was reported by the aMCI than CN group for the Inhibit (q = 0.035), Shift (q = 0.005), Emotional Control (q = 0.021), Initiate (q = 0.028), Working Memory (q = 0.005), Plan/Organize (q = 0.005), and Task Monitor (q = 0.016) scales, as well as the MI (q = 0.005), BRI (q = 0.005), and GEC (q = 0.005). Poorer executive function was also reported by the SCD than CN group on the Shift (q = 0.018), Initiate (q = 0.031) Working Memory (q = 0.005), and Plan/Organize (q = 0.011) scales, as well as the MI (q = 0.011) and GEC (q = 0.013). No significant differences emerged between aMCI and SCD groups.

At Time 2, the aMCI group continued to endorse more difficulty than the CN group on the Shift, Emotional Control, Initiate, Working Memory, Plan/Organize, and Task Monitor scales, as well as the MI, BRI, and GEC (all q = 0.005). Similarly, the SCD group continued to endorse worse executive functioning than the CN group for Shift (q = 0.028), Initiate (q = 0.008), Working Memory (q = 0.005), and Plan/Organize (q = 0.008) scales, as well as the MI (q = 0.005) and GEC (q = 0.005). Furthermore, at Time 2 Task Monitor was also more elevated for the SCD than CN group (q = 0.008). No differences were observed between aMCI and SCD groups.

Table 3 shows BRIEF-A informant report data with q-values for comparisons across groups at Time 1, Time 2, and the change from Time 1 to Time 2. At Time 1, more elevated BRIEF-A scores in the aMCI than CN group was seen for the Inhibit (q = 0.04), Shift (q = 0.011), Working Memory (q = 0.005), and Organization of Materials (q = 0.016) scales, as well as the MI (q = 0.013), BRI (q = 0.04), and GEC (q = 0.011). In addition, the aMCI group endorsed having worse Working Memory than the SCD group (q = 0.048). No differences were seen between SCD and CN groups.

At Time 2 for the informant report, more elevated BRIEF-A scores in the aMCI than CN group were observed for Shift (p = .005), Emotional Control (q = 0.011), Initiate (q = 0.031), Working Memory (q = 0.005), Plan/Organize (q = 0.005), and Organization of Material (q = 0.028) scales, as well as the BRI (q = 0.008), MI (q = 0.005), and GEC (q = 0.005). In addition, the aMCI group endorsed greater difficulty than the SCD group on the Plan/Organize scale (q = 0.031) and for the GEC (q = 0.044). No differences were seen between the SCD and CN groups. No group by time interaction was observed for any index or scale score.

Discussion

Older adults with aMCI and those with SCD, whose diagnosis remained stable over approximately 1 year, reported significant problems with executive functioning that also remained stable over this period. Similarly, knowledgeable informants rated those with aMCI and SCD as having greater difficulty with executive functions than the informants of CN participants, although informant scores indicated less difficulty than participant scores for both the aMCI and SCD groups. These findings at the 1-year follow-up are generally consistent with a previous study of many of these same participants that reported greater subjective problems with executive functioning at baseline in aMCI and with SCD (Rabin et al., 2006) compared to CN, including prominent working memory concerns by study participants and their informants.

Our finding that unchanged diagnosis was associated with stability of both subjective and objective executive functioning is consistent with previous studies in which relative stability on objective measures of memory and language predicted a stable MCI diagnosis (Manly et al., 2008; Ellendt et al., 2016). Past literature, including baseline data of these participants, has generally shown a minimal or small relationship between the BRIEF-A and standardized neuropsychological tests of executive function (Rabin et al., 2006; Combs, Garcia-Willingham, Berry, van Horne, & Segerstrom, 2018). Future studies may investigate if objective executive function test performance adds predictive validity for conversion. It has been reported, however, that subjective cognitive concerns may remain stable in SCD for two or more years even as objective executive functioning subtly declines (Jeong, Park, Song, Chung, & Rhie, 2017). It remains unclear whether older adults who convert to a more severe diagnostic state (i.e., CN to SCD, SCD to MCI, or MCI to dementia) show worsening of subjective executive function over time, or whether such deterioration, if present, is helpful in predicting conversion.

The present findings should be interpreted in the context of the study limitations. Participants were relatively homogenous in terms of ethnicity and on average had a high level of education, thus limiting generalizability. This is particularly salient in light of research suggesting that older adults with fewer years of education may endorse greater levels of complaints, yet those who are highly educated may have complaints that are more prognostic of cognitive decline (Molinuevo et al., 2017). Furthermore, the follow-up period of one-year may have limited our ability to uncover changes in subjective or objective executive functioning. For example, memory concerns among cognitively intact older adults may precede objective cognitive dysfunction by 10 or more years (Kryscio et al., 2014; Kaup, Nettiksimmons, LeBlanc, & Yaffe, 2015). A limitation of the present study is the relatively small sample size to the total number of statistical tests. Ultimately, these results should be replicated in a larger independent sample. Finally, our MCI sample was exclusively of the amnestic subtype. Thus, further research will be needed to determine whether there are differences in subjective executive functioning between aMCI and other MCI subtypes, and whether such subjective ratings (self and/or informant) are differentially predictive of diagnostic conversion.

Despite these limitations, to our knowledge, the current study is the first to examine the stability of subjective executive functioning concerns in older adults with stable diagnoses and their informants, and to compare those concerns to both baseline and follow-up objective cognitive performance. Further research, with a larger sample and longer period of follow-up with additional time points, would strengthen the understanding of subjective executive functioning concerns as a predictor. For example, increased concerns among both older adults and informants could precede diagnostic change. Alternatively, increased informant concerns in the context of stable or reduced self-reported concerns could indicate a loss of personal insight into cognitive decline (Morris & Mograbi, 2013; Buckley et al., 2015). This multifaceted exploration may enhance the ability to predict which older adults will convert to a more severe diagnosis and in what timeframe.

Acknowledgements

We thank James Ford, Ph.D. for his assistance with statistical analyses.

Funding

This work was supported in part by National Institutes of Health [R01 AG019771 and P30 AG010133 to A.J.S.].

Conflict of Interest

Robert M. Roth is an author of the BRIEF-A and receives royalties from the publisher. There are no other known conflicts of interest.

References