Skip to main content
PMC home page
Psychological Science logoLink to Psychological Science
. 2021 May 27;32(6):849–860. doi: 10.1177/0956797620985518

Undetected Neurodegenerative Disease Biases Estimates of Cognitive Change in Older Adults

Karra D Harrington 1,2,3, Andrew J Aschenbrenner 4,5, Paul Maruff 1,6, Colin L Masters 1, Anne M Fagan 4,5,7, Tammie L S Benzinger 4,8,9, Brian A Gordon 4,7,8, Carlos Cruchaga 10,11, John C Morris 4,5,12,13,14, Jason Hassenstab 4,5,15,✉
PMCID: PMC8726587  PMID: 34043464

Abstract

Neurodegenerative disease is highly prevalent among older adults and, if undetected, may obscure estimates of cognitive change among aging samples. Our aim in this study was to determine the nature and magnitude of cognitive change in the absence of common neuropathologic markers of neurodegenerative disease. Cognitively normal older adults (ages 65–89 years, N = 199) were classified as normal or abnormal using neuroimaging and cerebrospinal-fluid biomarkers of β-amyloid, tau, and neurodegeneration. When cognitive change was modeled without accounting for biomarker status, significant decline was evident for semantic memory, processing speed, and working memory. However, after adjusting for biomarker status, we found that the rate of change was attenuated and that the biomarker-normal group demonstrated no decline for any cognitive domain. These results indicate that estimates of cognitive change in otherwise healthy older adults will be biased toward decline when the presence of early neurodegenerative disease is not accounted for.

Keywords: cognitive aging, β-amyloid, tau, neurodegeneration, memory, processing speed

Current psychological models of aging propose that cognitive decline is an inevitable consequence of increasing age (Anderson & Craik, 2017). Such expectations may create negative societal and individual attitudes toward older adults, particularly in relation to employment, independent living, and management of health and well-being (Blazer & Wallace, 2016). Global population trends show that the proportion of individuals over the age of 65 years will increase substantially in the next decades, with greater numbers of older adults expected to remain employed longer, stay involved in community and social endeavors, and live independently (Harper, 2014). Consequently, models of cognitive aging should be continually refined to ensure their accuracy is optimal for input on policies and expectations about the aging population.

Recent meta-analytic estimates of cognitive change in late life indicate that decline occurs across global cognition, memory, processing speed, language, and executive function, and such decline ranges between −0.12 to −0.26 SD units per decade of increasing age (Lipnicki et al., 2017). There is, however, growing support for an alternate hypothesis that estimates of cognitive change in older adults may have been biased negatively by the influence of diseases that are prevalent in late life, particularly neurodegenerative disease, such as Alzheimer’s disease (Spiro & Brady, 2011).

With advances in neuroimaging and biomarker technology, it is now possible to investigate the neurobiological mechanisms of cognitive change in late life (Hedden & Gabrieli, 2004; Jagust, 2013). There is much evidence that what is considered to be age-related cognitive decline is mediated by underlying neurobiological changes, such as vascular changes or an accumulation of neuropathology (Boyle et al., 2013; Hassenstab et al., 2016; Hedden et al., 2016). However, many of these investigations employed a cross-sectional design or relied on postmortem indices of neuropathology. Consequently, debate continues about the strength of the evidence for neurobiological mechanisms of age-related cognitive change and the extent to which the inadvertent inclusion of adults with neurodegenerative disease in healthy aging samples biases estimates of cognitive decline (Salthouse, 2011). One approach is to reevaluate estimates of cognitive change prospectively in samples known to be free of the most common neuropathological markers of neurodegenerative disease in later life.

The most prevalent neurodegenerative disease among older adults is Alzheimer’s disease, which is characterized by an insidious onset with neuropathology accumulating decades prior to the expression of dementia (Holtzman et al., 2011). Biomarker methods using analyses of cerebrospinal fluid (CSF) and neuroimaging techniques—positron emission tomography (PET) and MRI—have been validated for the detection of Alzheimer’s disease–related neuropathology, including accumulation of β-amyloid (Aβ) and hyperphosphorylated tau protein, as well as neurodegeneration. Together, these three neuropathologic markers are informative about the presence and severity of Alzheimer’s disease (Jack et al., 2017). When considered in combination, these biomarkers are also useful in identifying individuals who most likely do not have Alzheimer’s disease. In cognitively normal older adults, biomarker studies indicate that the prevalence of at least one of these neuropathologic markers increases from 44% at age 65 to 86% at age 80 (Jack et al., 2017). Furthermore, prospective studies indicate that cognitively normal older adults with biomarker evidence of Alzheimer’s disease neuropathology show subtle progressive decline across episodic memory, semantic memory, visuospatial function, and global cognition (Baker et al., 2016; Boyle et al., 2013). Therefore, the potential for undetected Alzheimer’s disease to influence estimates of cognitive change increases with age and is likely to impact a broad range of cognitive processes.

Statement of Relevance.

The results of the present research have important implications for expectations of cognition with aging as well as social and individual expectations of older adults. In this study, we demonstrated that undetected neurodegenerative disease negatively biases models of cognitive change in aging samples, leading to an overestimation of the rate of decline in multiple cognitive abilities. This information should be used to inform expectations of older adults at the broader societal level, to enhance self-care practices for older adults, and to encourage clinicians to seek explanations for failing cognition in older adults. Given the relevance of the findings from this study to individual and societal attitudes about aging and the ramifications of those attitudes for the health and well-being of older adults, it is vital that models of cognitive aging be reconsidered from the context of preclinical neurodegenerative disease.

Our aim in this study was to examine whether undetected Alzheimer’s disease influenced estimates of cognitive change in a large sample of clinically verified, cognitively normal older adults and to extend our prior work on the impact of preclinical Alzheimer’s disease on cross-sectional normative data (Hassenstab et al., 2016). The influence of each of the three biomarkers and their interaction is important for developing models of the genesis of Alzheimer’s disease, and we and others have examined this in detail in prior publications (Aschenbrenner et al., 2018). However, the focus of this study was to understand trajectories of cognitive change in the absence of Alzheimer’s disease rather than to focus on relationships between specific indicators of neuropathology and cognition. Our first hypothesis was that, without accounting for biomarker status, there would be significant decline on all cognitive measures. The second hypothesis was that estimates of cognitive decline would be reduced once biomarker status was accounted for.

Method

Participants

Older adults (N = 199) enrolled in ongoing studies of aging and dementia at the Knight Alzheimer Disease Research Center (ADRC) were used in this study. Inclusion and exclusion criteria for the Knight ADRC studies have previously been described in detail (Berg et al., 1998; Coats & Morris, 2005). Briefly, all participants were living independently in the community at study entry and had no severe or uncontrolled medical illnesses that would prevent ongoing engagement in the study. All Knight ADRC studies are approved by the Human Research Protection Office at Washington University School of Medicine, and all participants provided written informed consent at the time of enrollment in the study.

Selection criteria for the present study also required that participants were between 65 and 90 years old and had a Clinical Dementia Rating (CDR; Morris, 1993) of 0 at their baseline assessment. Additionally, participants were required to have undergone at least one lumbar puncture for CSF sampling as well as a structural MRI and an Aβ PET scan during the study to determine levels of Aβ, tau, and neurodegeneration. When biomarkers were collected at multiple visits, data from the most recent visit were used for biomarker classification. Given that biomarker levels can change over time, the most recent biomarker classification was selected to capture the measurement closest to participants’ current state during the study period. There were 20 participants included in the study who progressed from biomarker normal to abnormal within the study period. The biomarker classification of all other participants remained stable.

The sample size for the present study was determined by the maximum number of participants from the existing Knight ADRC studies for whom biomarker and cognitive data were available and who met the study’s inclusion criteria.

Our prior publication (Hassenstab et al., 2016), which examined the impact of Alzheimer’s disease biomarkers on cross-sectional normative data, included 264 Knight ADRC participants. The present study included 145 of these participants and 53 participants who were not in the prior study. As in most cohort studies, there are many reasons why subsets of data drawn at different times do not overlap completely. These reasons include attrition, availability of all modalities of biomarker data, and sufficient follow-up intervals for longitudinal analysis.

Measures

Demographic and clinical

Age, sex, and education were self-reported by participants. Presence and severity of dementia was assessed with the CDR. Cognitive normality is indicated by a CDR of 0, whereas 0.5, 1, 2, and 3 indicate very mild, mild, moderate, and severe dementia, respectively (Berg et al., 1992). Apolipoprotein E (APOE) genotype was determined from a blood sample (as described in Pizzie et al., 2014).

Cognitive

The neuropsychological batteries used in the Knight ADRC studies, their processes for administration and scoring, and their data-quality assurance have been described in detail previously (Johnson et al., 2008; Pizzie et al., 2014). The core of the Knight ADRC batteries consists of measures drawn from the Uniform Data Set that are collected across all National Institute on Aging–funded Alzheimer’s Disease Research Centers in the United States (Weintraub et al., 2009).

The present study used a subset of the Knight ADRC neuropsychological batteries. These measures were selected because they were optimal for studying cognition in healthy older adults in that they provided normal data distributions without floor or ceiling effects (Harrington et al., 2017). Table 1 summarizes the individual test measures used in the present study according to theoretical cognitive domain. Domain-specific cognitive-composite scores were derived from the included tests on the basis of consensus agreement of a panel of cognitive psychologists and neuropsychologists.

Table 1.

Summary of the Neuropsychological Tests Comprising Each Cognitive-Composite Score

Cognitive domain and neuropsychological test Reference
Episodic memory
 Logical Memory: Delayed Recall Wechsler (1997)
 Associate Learning Wechsler (1997)
 Free and Cued Selective Reminding Test: Free Recall Grober et al. (1988)
Working memory
 Digit Span Backwards Wechsler (1981)
 Block Design Wechsler (1981)
 Letter-Number Sequencing Wechsler (1981)
Semantic memory
 Category Fluency for Animals and Vegetables Goodglass & Kaplan (1983)
 Word Fluency for letters /s/ and /p/ Thurstone & Thurstone (1949)
Processing speed
 Trail Making Test Parts A and B Armitage (1946)
 Digit Symbol Coding Wechsler (1981)
Open in a new tab

β-amyloid

PET with Carbon-11-labeled Pittsburgh Compound B (PIB) was used as the biomarker for Aβ. Data were processed using regions of interest derived from FreeSurfer (Version 5.3; Desikan et al., 2006). Participants were injected with approximately 10 mCi PIB prior to undergoing a 60-min dynamic PET scan (Su et al., 2013). Data from the 30- to 60-min postinjection window were converted to standardized-uptake-value ratios (SUVRs) relative to the cerebellar cortex. Data were partial-volume corrected using a regional-spread-function approach (Rousset et al., 1998; Su et al., 2015). A mean cortical SUVR was calculated from the prefrontal, parietal, and temporal regions of interest (Su et al., 2013). Aβ was classified as abnormal when SUVRs were greater than 1.42 (Hassenstab et al., 2016). Although CSF markers of Aβ were also available, they were not used in the present study because of previously established issues with drift in Aβ values from CSF analysis (Schindler et al., 2017).

Tau

CSF analysis for phosphorylated tau (ptau181) was used as the biomarker for tau. Methodologies for CSF collection and analysis have been described in detail previously (Vos et al., 2013). Abnormal ptau181 was defined as levels greater than 67 pg/mL, according to cutoffs derived in prior studies (Vos et al., 2013).

Neurodegeneration

Structural MRI was used as the biomarker for neurodegeneration. MRI data were acquired on a Siemens TIM Trio 3T scanner. T1-weighted images were acquired using a magnetization-prepared rapid acquisition with gradient echo (MPRAGE) sequence with the following parameters: repetition time = 2,400 ms, echo time = 3.16 ms, flip angle = 8°, field of view = 256 mm, in-plane resolution = 176 × 256, slice thickness = 1 mm acquired in sagittal orientation. Images had a 1-mm isotropic resolution. The FreeSurfer imaging-analysis suite (Version 5.3) was used to obtain regional cortical thicknesses. This method has been shown to generate volumes and thicknesses with high correspondence to manually generated data (Desikan et al., 2006). A temporal-lobe cortical-thickness composite was calculated from the entorhinal, inferior temporal, middle temporal, and fusiform regions of interest. These regions have previously been shown to accurately discriminate between Aβ– cognitively normal and Aβ+ cognitively impaired individuals (Jack et al., 2017). To establish a cutoff score that differentiated clinically recognizable disease from healthy aging, we used a receiver-operating-curve analysis to determine the threshold for temporal-lobe cortical thickness that maximized the accuracy in discriminating between Aβ– cognitively normal participants and individuals who were Aβ+ and met criteria for mild cognitive impairment or dementia. On the basis of this analysis, we classified abnormal neurodegeneration for temporal-lobe cortical-thickness as values less than 2.60 mm.

Procedure

Clinical and cognitive assessments, including CDR and neuropsychological testing, were conducted annually. Participants completed a mean of 6.77 (SD = 2.57) follow-up visits to the Knight ADRC with a close friend or family member. At each assessment, trained clinicians interviewed both the participant and their friend or family member about the participant’s cognitive and functional abilities to determine CDR score and diagnosis. Participants then underwent neuropsychological assessment and provided a blood sample for APOE genotyping. CSF collection and neuroimaging were conducted at a separate visit, typically within 2 weeks of the clinical and cognitive assessments. Participants underwent a mean of 1.75 (SD = 0.85) PET scans, 1.75 (SD = 0.89) CSF collections, and 2.37 (SD = 1.16) MRI scans.

Data coding

Sex, years of education, biomarker status, progression status, and APOE genotype were each coded as dichotomous categorical variables. Education was classified using 12 years as a cutoff (i.e., > 12 or ≤ 12). Biomarker status was classified as abnormal if any of the three biomarkers were identified as abnormal and as normal if all three biomarkers were in the normal range. Participants were classified as progressors if their CDR was greater than 0 at their most recent follow-up assessment or as stable if their CDR remained at 0 at their most recent follow-up assessment. APOE genotype was classified according to whether individuals carried an ε4 allele. Baseline age was centered using the mean age of the sample (71.7 years).

All neuropsychological test scores were converted to z scores using the mean and standard deviation of the entire sample at baseline. The relevant z scores for each cognitive domain were then averaged to create composite scores (see Table 1).

Data analysis

All analyses were completed using the R statistical computing environment (Version 3.4.0; R Core Team, 2017) and the lme4 package (Bates et al., 2015). Data-frequency distributions were generated for each outcome measure and visually inspected for diversion from normality. The biomarker-normal and biomarker-abnormal groups were compared on demographic and clinical characteristics using independent-samples t tests for continuous measures and χ2 analyses for categorical measures.

To address the first hypothesis (that there would be significant decline on all cognitive measures), we used linear mixed models (LMMs) to assess change in cognition. Baseline age, sex, education, APOE status, and years in the study were entered as fixed factors. Random slopes for years in the study and intercepts for each participant were also included. Each cognitive-composite score was entered as a dependent variable.

To address the second hypothesis (that estimates of cognitive decline would be reduced when biomarker status was accounted for), we repeated the LMM analyses with biomarker status (normal or abnormal) and the interaction between biomarker status and years in the study.

A post hoc analysis was then conducted to determine whether progression to a CDR greater than 0 contributed to cognitive decline in addition to biomarker status. The LMM analyses were repeated with progression status (progressor or stable) and the interaction between progression status and years in the study added as covariates in addition to biomarker classification.

To further consider whether disease progression might obscure our results, we conducted a sensitivity analysis in which we ran all of the mixed models with the data for participants whose biomarkers transitioned from normal to abnormal, censored to only those time points at which their biomarkers remained normal.

Results

Sample biomarker classifications and characteristics

Figure 1 shows the prevalence of each of the three biomarkers within the sample and their overlap. To answer our specific hypotheses, we split the sample according to whether individuals were positive on any of the three biomarkers. A Pearson product-moment correlation coefficient was computed to assess the relationship between baseline age and each of the three biomarkers. There was no correlation between baseline age and Aβ, r = .10, 95% confidence interval (CI) = [−.04, .24], p = .14, or baseline age and tau, r = −.01, 95% CI = [−.15, .12], p = .84. There was a significant negative correlation between baseline age and neurodegeneration, r = −.35, 95% CI = [−.47, −.23], p < .01.

Fig. 1.

Fig. 1.

Open in a new tab

Venn diagram showing the prevalence of positive classifications for each of the three biomarkers in the sample. Of the 133 participants who were classified as biomarker abnormal, 72 were amyloid positive, 68 were tau positive, and 60 were neurodegeneration positive. Additionally, 66 participants were classified as negative on all three biomarkers.

Table 2 summarizes the demographic and clinical characteristics of the sample and each of the biomarker groups. Comparison showed that the biomarker-abnormal group was older at baseline and was more likely to have progressed to a CDR greater than 0 at follow-up than the biomarker-normal group. There were no other significant differences between the two groups.

Table 2.

Demographic and Clinical Characteristics of the Sample

Variable Full sample
(N = 199)		 Biomarker-normal group
(n = 66)		 Biomarker-abnormal group
(n = 133)		 p
Baseline age (years) M = 71.69 (SD = 5.36) M = 69.51 (SD = 4.20) M = 72.78 (SD = 5.55) < .001
Years of education M = 15.77 (SD = 2.82) M = 15.47 (SD = 2.58) M = 15.92 (SD = 2.94) .27
Number of visits M = 6.77 (SD = 2.57) M = 6.82 (SD = 2.55) M = 6.75 (SD = 2.57) .86
Female n = 110 (55.27%) n = 39 (59.09%) n = 71 (53.38%) .54
White n = 190 (95.48%) n = 64 (96.97%) n = 126 (94.74%) .68
APOE ε4 carrier n = 68 (34.17%) n = 20 (30.30%) n = 48 (36.09%) .51
CDR > 0 at last visit n = 30 (15.08%) n = 2 (3.03%) n = 28 (21.05%) .01
Amyloid positive n = 72 (36.18%)
Tau positive n = 68 (34.17%)
Neurodegeneration positive n = 60 (30.15%)
Open in a new tab

Note: Biomarker assessment included positron emission tomography with Carbon-11-labeled Pittsburgh Compound B for β-amyloid, cerebrospinal-fluid analysis for phosphorylated tau (ptau181), and temporal-lobe cortical thickness on anatomical MRI for neurodegeneration. The biomarker-negative group included only individuals who were negative on all three biomarkers. The biomarker-positive group included all individuals who were positive on at least one of the three biomarkers. APOE = apolipoprotein E; CDR = Clinical Dementia Rating.

Unadjusted cognitive change

Table 3 summarizes the results of the LMM analyses. Baseline age was significantly associated with baseline performance on each cognitive composite. Associations were consistently negative, ranging from −0.04 to −0.07 SD units for each additional year of age (see Table 3, Model 1). There was a significant main effect of years in the study on all cognitive composites, except for episodic memory. The rate of decline ranged from −0.02 to −0.03 SD units per year in the study (see Table 3, Model 1). Figure 2 shows the unadjusted slopes for each cognitive-composite score.

Table 3.

Parameter Estimates From Linear Mixed Models Examining Change in Cognitive-Composite Scores

Domain and predictor Model 1 Model 2 Model 3
b SE p b SE p b SE p
Episodic memory
 Intercept −0.010 0.083 .910 −0.161 0.110 .146 −0.141 0.110 .200
 Sex (female) 0.540 0.097 < .001 0.546 0.098 < .001 0.531 0.097 < .001
 Education (< 12 years) −0.368 0.120 < .001 −0.369 0.120 .002 −0.374 0.119 .002
 APOE (ε4 carrier) −0.148 0.103 .152 −0.158 0.103 .127 −0.149 0.102 .146
 Baseline age (centered) −0.044 0.009 < .001 −0.047 0.010 < .001 −0.040 0.010 < .001
 Years in the study 0.010 0.009 .277 0.051 0.015 .001 0.058 0.012 < .001
 Biomarker status (abnormal) 0.227 0.111 .041 0.221 0.110 .047
 Clinical progression (progressor) −0.051 0.151 .737
 Biomarker Status (Abnormal) × Years in the Study −0.061 0.019 .001 −0.034 0.015 .030
 Clinical Progression (Progressor) × Years in the Study −0.182 0.021 < .001
Working memory
 Intercept 0.057 0.079 .470 0.014 0.108 .893 0.030 0.109 .785
 Sex (female) 0.080 0.093 .390 0.080 0.093 .392 0.073 0.094 .440
 Education (< 12 years) −0.333 0.115 .004 −0.333 0.115 .004 −0.336 0.116 .004
 APOE (ε4 carrier) 0.010 0.098 .920 0.009 0.098 .929 0.015 0.099 .881
 Baseline age (centered) −0.057 0.009 < .001 −0.057 0.009 < .001 −0.054 0.010 < .001
 Years in the study −0.031 0.006 < .001 −0.021 0.011 .056 −0.020 0.011 .062
 Biomarker status (abnormal) 0.066 0.111 .551 0.044 0.113 .695
 Clinical progression (progressor) 0.165 0.172 .340
 Biomarker Status (Abnormal) × Years in the Study −0.014 0.014 .285 −0.002 0.014 .886
 Clinical Progression (Progressor) × Years in the Study −0.085 0.022 < .001
Semantic memory
 Intercept −0.041 0.083 .622 −0.106 0.112 .346 −0.117 0.113 .300
 Sex (female) 0.277 0.098 .005 0.277 0.098 .005 0.289 0.099 .004
 Education (< 12 years) −0.378 0.121 .002 −0.378 0.121 .002 −0.382 0.122 .002
 APOE (ε4 carrier) 0.077 0.103 .456 0.075 0.104 .471 0.065 0.105 .534
 Baseline age (centered) −0.049 0.009 < .001 −0.049 0.010 < .001 −0.050 0.010 < .001
 Years in the study −0.027 0.007 < .001 −0.007 0.011 .544 −0.004 0.010 .654
 Biomarker status (abnormal) 0.098 0.113 .385 0.056 0.113 .625
 Clinical progression (progressor) 0.284 0.154 .067
 Biomarker Status (Abnormal) × Years in the Study −0.030 0.014 .030 −0.014 0.013 .271
 Clinical Progression (Progressor) × Years in the Study −0.099 0.017 < .001
Processing speed
 Intercept 0.026 0.083 .750 −0.031 0.110 .779 −0.036 0.110 .743
 Sex (female) 0.180 0.098 .067 0.183 0.098 .064 0.184 0.099 .064
 Education (< 12 years) −0.365 0.121 .003 −0.366 0.121 .003 −0.364 0.121 .003
 APOE (ε4 carrier) 0.081 0.103 .435 0.076 0.104 .468 0.080 0.104 .445
 Baseline age (centered) −0.068 0.009 < .001 −0.069 0.010 < .001 −0.068 0.010 < .001
 Years in the study −0.022 0.008 .004 0.004 0.013 .776 0.008 0.011 .449
 Biomarker status (abnormal) 0.086 0.109 .432 0.053 0.109 .626
 Clinical progression (progressor) 0.163 0.146 .266
 Biomarker Status (Abnormal) × Years in the Study −0.039 0.016 .015 −0.017 0.014 .222
 Clinical Progression (Progressor) × Years in the Study −0.137 0.018 < .001
Open in a new tab

Note: APOE = apolipoprotein E.

Fig. 2.

Fig. 2.

Open in a new tab

Rate of change in each cognitive-composite score, separately for the full sample (black line), biomarker-abnormal group (red line), and biomarker-normal group (blue line). Error bands represent 95% confidence intervals.

Cognitive change adjusted for biomarker status

Next, biomarker status and the interaction between biomarker status and years in the study were included in the LMM analyses. There was a significant interaction between biomarker status and years in the study for episodic memory, semantic memory, and processing speed, but not for working memory (see Table 3, Model 2). The biomarker-normal group showed significant improvement on the episodic-memory composite (0.05 SD units per each year in the study) and did not show significant change on the semantic-memory or processing-speed composites. In contrast, the biomarker-abnormal group showed significant decline in episodic memory, semantic memory, and processing speed. Figure 2 shows the slopes for each cognitive-composite score in each of the biomarker groups.

Cognitive change adjusted for progression status

Of 199 participants, 30 (15.1%) progressed to a CDR greater than 0 at their most recent clinical assessment, indicating clinical onset of dementia. Of those who progressed to a CDR greater than 0, 28 were classified as biomarker abnormal. There was a significant interaction between progression status and years in the study for all four cognitive composites, with progressors showing significantly greater decline than nonprogressors (see Table 3, Model 3). When progression status was added to the models, the interaction between biomarker status and years in the study remained significant for only episodic memory. For semantic memory and processing speed, the difference in slopes between the biomarker-normal and biomarker-abnormal groups was not significant after taking progression status into account.

Of 199 participants, 20 (10.1%) transitioned from biomarker negative to biomarker positive during the study period. For these participants, we censored the data to only those time points at which they remained biomarker normal, and then we reran each of the mixed-effects models. After we censored the data, the overall pattern of results was unchanged, although there was a slight increase in the rate of decline for the biomarker-abnormal group for the episodic-memory, semantic-memory, and processing-speed composites.

Discussion

The first hypothesis—that without accounting for biomarker status there would be significant decline on all cognitive measures—was supported partially. Decline was observed for working memory, semantic memory, and processing speed but not for episodic memory (Fig. 2). The magnitude of decline ranged from −0.22 SD units per decade for processing speed to −0.31 SD units per decade for working memory (Table 3, Model 1). These estimates are consistent with those from a recent large meta-analysis of age-related decline in executive function (e.g., M = −0.23 SD units per decade), language (e.g., M = −0.16 SD units per decade), and processing speed (e.g., M = −0.26 SD units per decade; Lipnicki et al., 2017). Thus, in the current prospective examination of a large sample of clinically verified cognitively normal older adults, the expected decline in multiple domains of cognition was evident when biomarkers of neurodegenerative disease were not considered.

The absence of decline in episodic memory was unexpected and inconsistent with findings from previous studies (Lipnicki et al., 2017). When change in performance was considered for each of the biomarker groups separately, there was a significant trend toward improvement in the biomarker-normal group and for decline in the biomarker-abnormal group. This is consistent with other prospective studies utilizing the same or similar episodic-memory assessments, which showed improved test scores in their cognitively normal samples (Goldberg et al., 2015) but not for individuals with preclinical Alzheimer’s disease (Gavett et al., 2016; Hassenstab et al., 2015). This improvement in performance is often described as a practice or retest effect, which appears to be most common on episodic-memory measures (Calamia et al., 2012), particularly in studies including biomarker-normal participants that use the same test stimuli at each assessment (Hassenstab et al., 2015), as is the case in the Knight ADRC cohort. This finding highlights a further issue for estimation of cognitive change—that is, the potential for practice effects to obscure decline in longitudinal studies. Nevertheless, our finding demonstrating a clear interaction between biomarker status and time supports hypotheses that preclinical neurodegenerative disease biases estimates of cognitive change in aging samples.

The second hypothesis, that estimates of cognitive decline would be reduced once biomarker status was accounted for, was supported for all aspects of cognition. For semantic memory and processing speed, biomarker-normal participants demonstrated essentially no change over time, whereas for episodic memory, they showed substantial improvement over time. For working memory, the slope estimate for decline was reduced from −0.031 SD units per year across all participants to −0.021 SD units per year in the biomarker-normal participants (Table 3, Model 2). This is consistent with the results of prior studies, which have shown that residual cognitive decline remains in similar measures of working memory after accounting for neuropathology (Hohman et al., 2017; Yu et al., 2014). Taken together, these findings accord with those of other recent studies that have shown that cognitive decline is modified by the presence of neuropathologic lesions associated with Alzheimer’s disease (Hohman et al., 2017; Yu et al., 2014). These results suggest that, in the absence of preclinical neurodegenerative disease, episodic memory, semantic memory, and processing speed show only very subtle or no decline in cognitively normal older adults.

When progression to a CDR greater than 0 was included in the post hoc analyses, there was no longer a difference between biomarker-normal and biomarker-abnormal groups for change in semantic memory and processing speed. One explanation is that the progressors may have had more severe or diffuse neuropathologic lesions, suggesting that their disease was more advanced compared with that of participants from the biomarker-abnormal group who remained stable during the study period. Notably, prior studies have found that decline in processing speed and working memory occurs closer to the time of clinical diagnosis than changes in episodic memory, which occur earlier in the disease process (Mormino et al., 2017). In neurobiological models of Alzheimer’s disease, accumulation of neuropathologic lesions initially occurs primarily in the medial temporal lobes that subserve episodic-memory function (Braak & Braak, 1991). In line with this, even after accounting for clinical progression, results showed that episodic memory declined significantly more in individuals with abnormal biomarkers relative to the biomarker-normal group.

This study should be considered in the context of several limitations. First, the sample was composed of individuals volunteering for studies of aging and dementia and primarily consisted of individuals who were White, highly educated, and motivated to participate in research studies. Therefore, the results may not generalize to the broader population, particularly to individuals with lower levels of education or from diverse racial and ethnic backgrounds. A further consideration is the degree to which the current sample represents typical aging: It is important to highlight that the cohort study that participants were drawn from for the present study was designed to investigate risk of dementia, and consequently having a relatively high proportion of the sample with one or more abnormal biomarkers was expected. Nevertheless, the results of the present study demonstrate the potential for undetected neurodegenerative disease to influence trajectories of cognitive change in otherwise healthy older adults.

Second, the use of cutoff scores on the biomarkers to categorize individuals as normal or abnormal may be suboptimal for identifying neurodegenerative disease, as they are not able to account for individual differences or intraindividual changes. To address this limitation, we used the most recent biomarker-assessment results to classify individuals as normal or abnormal, which ensured that any individuals who transitioned from normal to abnormal (n = 20, or 10.1%) on any of the biomarkers were accurately classified as abnormal. Given the relatively slow accumulation of neuropathology in Alzheimer’s disease, which occurs over decades (Villemagne et al., 2013), these individuals were likely already on a pathological trajectory at the start of the study period despite not yet meeting criteria for abnormality. Notably, when data for these participants were censored to only the time points at which they were classified as biomarker normal, the overall outcome of the study did not change.

Third, although this study considered the influence of three of the most common markers of neuropathology in older adults, there are many other factors that are prevalent in older adults that may also influence cognitive change. For example, vascular disease is highly prevalent among older adults and is associated with cognitive decline (Spiro & Brady, 2011). However, it is notable that, despite this, cognitive decline was substantially attenuated after we accounted for the three biomarkers included in this study.

Despite these limitations, the results of the study did demonstrate that in this well-characterized sample of older adults, estimates of cognitive change were influenced by markers of neurodegenerative disease. Notably, failure to account for the influence of neurodegenerative disease led to an overestimation of the rate of cognitive decline across multiple domains of cognition. The results of the present study demonstrate that in the absence of neuropathological changes, cognitive decline associated with aging is attenuated.

Footnotes

ORCID iD: Jason Hassenstab Inline graphic https://orcid.org/0000-0002-7802-3371

Transparency

Action Editor: John Jonides

Editor: D. Stephen Lindsay

Author Contributions

K. D. Harrington, A. J. Aschenbrenner, P. Maruff, and J. Hassenstab developed the study concept and design. Data were collected by A. M. Fagan, T. L. S. Benzinger, B. A. Gordon, C. Cruchaga, J. C. Morris, and J. Hassenstab. K. D. Harrington, A. J. Aschenbrenner, P. Maruff, B. A. Gordon, and J. Hassenstab analyzed and interpreted the data. K. D. Harrington drafted the manuscript under the supervision of P. Maruff, C. L. Masters, and J. Hassenstab. All authors provided critical revisions and approved the final manuscript for submission.

Declaration of Conflicting Interests: P. Maruff is the Chief Innovation Officer of CogState. The authors declared that there were no other potential conflicts of interest with respect to the authorship or the publication of this article.

Funding: K. D. Harrington received PhD scholarships cofunded by the Dementia Australia Research Foundation, the Florey Institute of Neuroscience and Mental Health, the Yulgilbar Alzheimer’s Research Program, and the Cooperative Research Centre (CRC) for Mental Health. The CRC program is an Australian government initiative. Funding for the Knight Alzheimer Disease Research Center is provided by National Institutes of Health Grants P50AG005681, P01AG003991, and P01AG026276 to J. C. Morris.

Open Practices: Data and materials for this study have not been made publicly available, and the design and analysis plans were not preregistered.

References

  1. Anderson N. D., Craik F. I. M. (2017). 50 years of cognitive aging theory. Journals of Gerontology Series B: Psychological Sciences & Social Sciences, 72(1), 1–6. 10.1093/geronb/gbw108 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Armitage S. G. (1946). An analysis of certain psychological tests used for the evaluation of brain injury. Psychological Monographs, 60(1), i–48. [Google Scholar]
  3. Aschenbrenner A. J., Gordon B. A., Benzinger T. L. S., Morris J. C., Hassenstab J. J. (2018). Influence of tau PET, amyloid PET, and hippocampal volume on cognition in Alzheimer disease. Neurology, 91(9), Article e859. 10.1212/WNL.0000000000006075 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Baker J. E., Lim Y. Y., Pietrzak R. H., Hassenstab J., Snyder P. J., Masters C. L., Maruff P. (2016). Cognitive impairment and decline in cognitively normal older adults with high amyloid-β: A meta-analysis. Alzheimer’s & Dementia: Diagnosis, Assessment & Disease Monitoring, 6(1), 108–121. 10.1016/j.dadm.2016.09.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bates D., Mächler M., Bolker B., Walker S. (2015). Fitting linear mixed-effects models using lme4. Journal of Statistical Software, 67(1). 10.18637/jss.v067.i01 [DOI] [Google Scholar]
  6. Berg L., McKeel D. W., Jr., Miller J. P., Storandt M., Rubin E. H., Morris J. C., Baty J., Coats M., Norton J., Goate A. M., Price J. L., Gearing M., Mirra S. S., Saunders A. M. (1998). Clinicopathologic studies in cognitively healthy aging and Alzheimer disease: Relation of histologic markers to dementia severity, age, sex, and apolipoprotein E genotype. Archives of Neurology, 55(3), 326–335. 10.1001/archneur.55.3.326 [DOI] [PubMed] [Google Scholar]
  7. Berg L., Miller J. P., Baty J., Rubin E. H., Morris J. C., Figiel G. (1992). Mild senile dementia of the Alzheimer type. 4. Evaluation of intervention. Annals of Neurology, 31(3), 242–249. 10.1002/ana.410310303 [DOI] [PubMed] [Google Scholar]
  8. Blazer D. G., Wallace R. B. (2016). Cognitive aging: What every geriatric psychiatrist should know. The American Journal of Geriatric Psychiatry, 24, 776–781. 10.1016/j.jagp.2016.06.013 [DOI] [PubMed] [Google Scholar]
  9. Boyle P. A., Yu L., Wilson R. S., Schneider J. A., Bennett D. A. (2013). Relation of neuropathology with cognitive decline among older persons without dementia. Frontiers in Aging Neuroscience, 5, Article 50. 10.3389/fnagi.2013.00050 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Braak H., Braak E. (1991). Neuropathological stageing of Alzheimer-related changes. Acta Neuropathologica, 82(4), 239–259. 10.1007/BF00308809 [DOI] [PubMed] [Google Scholar]
  11. Calamia M., Markon K., Tranel D. (2012). Scoring higher the second time around: Meta-analyses of practice effects in neuropsychological assessment. The Clinical Neuropsychologist, 26(4), 543–570. 10.1080/13854046.2012.680913 [DOI] [PubMed] [Google Scholar]
  12. Coats M., Morris J. C. (2005). Antecedent biomarkers of Alzheimer’s disease: The adult children study. Journal of Geriatric Psychiatry and Neurology, 18(4), 242–244. 10.1177/0891988705281881 [DOI] [PubMed] [Google Scholar]
  13. Desikan R. S., Ségonne F., Fischl B., Quinn B. T., Dickerson B. C., Blacker D., Buckner R. L., Dale A. M., Maguire R. P., Hyman B. T., Albert M. S., Killiany R. J. (2006). An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest. NeuroImage, 31(3), 968–980. 10.1016/j.neuroimage.2006.01.021 [DOI] [PubMed] [Google Scholar]
  14. Gavett B. E., Gurnani A. S., Saurman J. L., Chapman K. R., Steinberg E. G., Martin B., Chaisson C. E., Mez J., Tripodis Y., Stern R. A. (2016). Practice effects on story memory and list learning tests in the neuropsychological assessment of older adults. PLOS ONE, 11(10), Article 0164492. 10.1371/journal.pone.0164492 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Goldberg T. E., Harvey P. D., Wesnes K. A., Snyder P. J., Schneider L. S. (2015). Practice effects due to serial cognitive assessment: Implications for preclinical Alzheimer’s disease randomized controlled trials. Alzheimer’s & Dementia: Diagnosis, Assessment & Disease Monitoring, 1(1), 103–111. 10.1016/j.dadm.2014.11.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Goodglass H., Kaplan E. (1983). Boston Diagnostic Aphasia Examination Booklet, III, ORAL EXPRESSION, J. Animal Naming (Fluency in Controlled Association). Lea & Febiger. [Google Scholar]
  17. Grober E., Buschke H., Crystal H., Bang S., Dresner R. (1988). Screening for dementia by memory testing. Neurology, 38(6), 900–903. [DOI] [PubMed] [Google Scholar]
  18. Harper S. (2014). Economic and social implications of aging societies. Science, 346(6209), 587–591. 10.1126/science.1254405 [DOI] [PubMed] [Google Scholar]
  19. Harrington K. D., Lim Y. Y., Ames D., Hassenstab J., Rainey-Smith S., Robertson J., Salvado O., Masters C. L., & Maruff, P., for the AIBL Research Group. (2017). Using robust normative data to investigate the neuropsychology of cognitive aging. Archives of Clinical Neuropsychology, 32(2), 142–154. 10.1093/arclin/acw106 [DOI] [PubMed] [Google Scholar]
  20. Hassenstab J., Chasse R., Grabow P., Benzinger T. L. S., Fagan A. M., Xiong C., Jasielec M., Grant E., Morris J. C. (2016). Certified normal: Alzheimer’s disease biomarkers and normative estimates of cognitive functioning. Neurobiology of Aging, 43, 23–33. 10.1016/j.neurobiolaging.2016.03.014 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Hassenstab J., Ruvolo D., Jasielec M., Xiong C., Grant E., Morris J. C. (2015). Absence of practice effects in preclinical Alzheimer’s disease. Neuropsychology, 29(6), 940–948. 10.1037/neu0000208 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Hedden T., Gabrieli J. D. E. (2004). Insights into the ageing mind: A view from cognitive neuroscience. Nature Reviews Neuroscience, 5(2), 87–96. 10.1038/nrn1323 [DOI] [PubMed] [Google Scholar]
  23. Hedden T., Schultz A. P., Rieckmann A., Mormino E. C., Johnson K. A., Sperling R. A., Buckner R. L. (2016). Multiple brain markers are linked to age-related variation in cognition. Cerebral Cortex, 26(4), 1388–1400. 10.1093/cercor/bhu238 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Hohman T. J., Tommet D., Marks S., Contreras J., Jones R., Mungas D., & Alzheimer’s Neuroimaging Initiative. (2017). Evaluating Alzheimer’s disease biomarkers as mediators of age-related cognitive decline. Neurobiology of Aging, 58, 120–128. 10.1016/j.neurobiolaging.2017.06.022 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Holtzman D. M., Morris J. C., Goate A. M. (2011). Alzheimer’s disease: The challenge of the second century. Science Translational Medicine, 3(77), Article 77SR1. 10.1126/scitranslmed.3002369 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Jack C. R., Jr., Wiste H. J., Weigand S. D., Therneau T. M., Knopman D. S., Lowe V., Vemuri P., Mielke M. M., Roberts R. O., Machulda M. M., Senjem M. L., Gunter J. L., Rocca W. A., Petersen R. C. (2017). Age-specific and sex-specific prevalence of cerebral β-amyloidosis, tauopathy, and neurodegeneration in cognitively unimpaired individuals aged 50–95 years: A cross-sectional study. The Lancet Neurology, 16(6), 435–444. 10.1016/S1474-4422(17)30077-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Jagust W. (2013). Vulnerable neural systems and the borderland of brain aging and neurodegeneration. Neuron, 77(2), 219–234. 10.1016/j.neuron.2013.01.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Johnson D. K., Storandt M., Morris J. C., Langford Z. D., Galvin J. E. (2008). Cognitive profiles in dementia: Alzheimer disease vs healthy brain aging. Neurology, 71(22), 1783–1789. 10.1212/01.wnl.0000335972.35970.70 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Lipnicki D. M., Crawford J. D., Dutta R., Thalamuthu A., Kochan N. A., Andrews G., Lima-Costa M. F., Castro-Costa E., Brayne C., Matthews F. E., Stephan B. C. M., Lipton R. B., Katz M. J., Ritchie K., Scali J., Ancelin M.-L., Scarmeas N., Yannakoulia M., Dardiotis E., . . . Cohort Studies of Memory in an International Consortium (COSMIC). (2017). Age-related cognitive decline and associations with sex, education and apolipoprotein E genotype across ethnocultural groups and geographic regions: A collaborative cohort study. PLOS Medicine, 14(3), Article e1002261. 10.1371/journal.pmed.1002261 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Mormino E. C., Papp K. V., Rentz D. M., Donohue M. C., Amariglio R., Quiroz Y. T., Chhatwal J., Marshall G. A., Donovan N., Jackson J., Gatchel J. R., Hanseeuw B. J., Schultz A. P., Aisen P. S., Johnson K. A., Sperling R. A. (2017). Early and late change on the preclinical Alzheimer’s cognitive composite in clinically normal older individuals with elevated amyloid β. Alzheimer’s & Dementia: The Journal of the Alzheimer’s Association, 13(9), 1004–1012. 10.1016/j.jalz.2017.01.018 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Morris J. C. (1993). The Clinical Dementia Rating (CDR): Current version and scoring rules. Neurology, 43, 2412–2414. 10.1212/WNL.43.11.2412-a [DOI] [PubMed] [Google Scholar]
  32. Pizzie R., Hindman H., Roe C., Head D., Grant E., Morris J. C., Hassenstab J. J. (2014). Physical activity and cognitive trajectories in cognitively normal adults: The Adult Children Study. Alzheimer Disease and Associated Disorders, 28(1), 50–57. 10.1097/WAD.0b013e31829628d4 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. R Core Team. (2017). R: A language and environment for statistical computing (Version 3.4.0) [Computer software]. R Foundation for Statistical Computing. https://www.R-project.org/ [Google Scholar]
  34. Rousset O. G., Ma Y., Evans A. C. (1998). Correction for partial volume effects in PET: Principle and validation. Journal of Nuclear Medicine, 39(5), 904–911. [PubMed] [Google Scholar]
  35. Salthouse T. A. (2011). Neuroanatomical substrates of age-related cognitive decline. Psychological Bulletin, 137(5), 753–784. 10.1037/a0023262 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Schindler S. E., Sutphen C. L., Teunissen C., McCue L. M., Morris J. C., Holtzman D. M., Mulder S. D., Scheltens P., Xiong C., Fagan A. M. (2017). Upward drift in cerebrospinal fluid amyloid β 42 assay values for more than 10 years. Alzheimer’s & Dementia: The Journal of the Alzheimer’s Association, 14(1), 62–70. 10.1016/j.jalz.2017.06.2264 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Spiro A., Brady C. B. (2011). Integrating health into cognitive aging: Toward a preventive cognitive neuroscience of aging. The Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 66B(Suppl. 1), i17–i25. 10.1093/geronb/gbr018 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Su Y., Blazey T. M., Snyder A. Z., Raichle M. E., Marcus D. S., Ances B. M., Bateman R. J., Cairns N. J., Aldea P., Cash L., Christensen J. J., Friedrichsen K., Hornbeck R. C., Farrar A. M., Owen C. J., Mayeux R., Brickman A. M., Klunk W., Price J. C., . . . the Dominantly Inherited Alzheimer Network. (2015). Partial volume correction in quantitative amyloid imaging. NeuroImage, 107, 55–64. 10.1016/j.neuroimage.2014.11.058 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Su Y., D’Angelo G. M., Vlassenko A. G., Zhou G., Snyder A. Z., Marcus D. S., Blazey T. M., Christensen J. J., Vora S., Morris J. C., Mintun M. A., Benzinger T. L. S. (2013). Quantitative analysis of PiB-PET with FreeSurfer ROIs. PLOS ONE, 8(11), Article e73377. 10.1371/journal.pone.0073377 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Thurstone L. E., Thurstone T. G. (1949). Examiner manual for the SRA Primary Mental Abilities Test. Science Research Associates. [Google Scholar]
  41. Villemagne V. L., Burnham S., Bourgeat P., Brown B., Ellis K. A., Salvado O., Szoeke C., Macaulay S. L., Martins R., Maruff P., Ames D., Rowe C. C., Masters C. L., for the Australian Imaging Biomarkers and Lifestyle (AIBL) Research Group (2013). Amyloid β deposition, neurodegeneration, and cognitive decline in sporadic Alzheimer’s disease: A prospective cohort study. The Lancet Neurology, 12(4), 357–367. 10.1016/S1474-4422(13)70044-9 [DOI] [PubMed] [Google Scholar]
  42. Vos S. J., Xiong C., Visser P. J., Jasielec M. S., Hassenstab J., Grant E. A., Cairns N. J., Morris J. C., Holtzman D. M., Fagan A. M. (2013). Preclinical Alzheimer’s disease and its outcome: A longitudinal cohort study. The Lancet Neurology, 12(10), 957–965. 10.1016/S1474-4422(13)70194-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Wechsler D. (1981). Manual for the Wechsler Adult Intelligence Scale, Revised. Psychological Corp. [Google Scholar]
  44. Wechsler D. (1997). Wechsler Memory Scale (3rd ed.): Administration and scoring manual. Psychological Corp. [Google Scholar]
  45. Weintraub S., Salmon D., Mercaldo N., Ferris S., Graff-Radford N. R., Chui H., Cummings J., DeCarli C., Foster N. L., Galasko D., Peskind E., Dietrich W., Beekly D. L., Kukull W. A., Morris J. C. (2009). The Alzheimer’s Disease Centers’ Uniform Data Set (UDS): The Neuropsychological Test Battery. Alzheimer Disease and Associated Disorders, 23(2), 91–101. 10.1097/WAD.0b013e318191c7dd [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Yu L., Boyle P. A., Leurgans S., Schneider J. A., Bennett D. A. (2014). Disentangling the effects of age and APOE on neuropathology and late life cognitive decline. Neurobiology of Aging, 35(4), 819–826. 10.1016/j.neurobiolaging.2013.10.074 [DOI] [PMC free article] [PubMed] [Google Scholar]

Articles from Psychological Science are provided here courtesy of SAGE Publications

RESOURCES