Skip to main content
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2017 Nov 1.
Published in final edited form as: Am J Geriatr Psychiatry. 2016 Apr 19;24(11):1095–1104. doi: 10.1016/j.jagp.2016.04.004

Mood changes in cognitively normal older adults are linked to Alzheimer’s disease biomarker levels

Ganesh M Babulal 1,2, Nupur Ghoshal 1,2, Denise Head 2,3,4, Elizabeth K Vernon 1, David M Holtzman 1,2,3, Tammie L S Benzinger 1,5,9, Anne M Fagan 1,2,3, John C Morris 1,2,6,7,8, Catherine M Roe 1,2
PMCID: PMC5069099  NIHMSID: NIHMS779564  PMID: 27426238

Abstract

Objectives

To evaluate whether cerebrospinal fluid (CSF) and PET Pittsburgh Compound B (PiB) biomarkers of underlying Alzheimer disease (AD) pathology (β-amyloid42 [Aβ42], tau, phosphorylated tau181 [ptau181], tau/Aβ42, ptau181/Aβ42 and mean cortical binding potential [MCBP] for PET-PiB) predict changes in mood in cognitively normal older adults.

Setting

Knight Alzheimer’s Disease Research Center (ADRC) at Washington University (WU).

Participants

Participants, 65 year of age or older, were enrolled from longitudinal studies at the WU Knight ADRC.

Measurements

CSF, PET-PiB biomarkers, Clinical Dementia Rating (CDR), Mini-Mental State Examination (MMSE), Profile of Mood States-Short Form (POMS-SF), the Geriatric Depression Scale (GDS) and Neuropsychiatric Inventory Questionnaire (NPI-Q).

Results

Data from 118 participants at baseline and 66 participants at one-year follow-up were analyzed. CSF and PET biomarkers were not associated cross-sectionally with any mood disturbances at baseline (p >0.05). Changes in mood as indicated by the total mood disturbance score on the POMS-SF, selected POMS-SF subscales, GDS, and NPI-Q scores from baseline to one-year follow-up were associated with (p < .05) CSF and PET-PiB biomarkers. There was no statistically significant decline in cognitive functioning

Conclusion

Generally, higher values of CSF and PET-PiB biomarkers are associated with more changes in mood in cognitively normal older adults. Further work is needed to understand the temporal development of mood changes over several years during the phase of preclinical AD. Evaluating mood as a noncognitive outcome may provide further insight into the development of preclinical AD in cognitively normal older adults.

Keywords: Alzheimer disease, Older adults, Mood, Depression, Biomarkers

Introduction

Mood disturbances, alternatively termed behavioral disturbances or neuropsychiatric symptoms, are common concomitant conditions of symptomatic Alzheimer’s disease (AD).1 Depressive symptoms in particular, have been found to predict the development of symptomatic AD for persons who are cognitively normal and persons with mild cognitive impairment (MCI).2, 3 Other mood disturbances such as apathy and anxiety are associated with decreased cognitive functioning, increased functional impairment, poorer prognosis, and faster rate of decline leading to institutionalization.4 Research efforts have recently shifted from symptomatic AD to preclinical AD, which is defined as cognitive normality with biomarker evidence (cerebrospinal fluid (CSF), imaging) that the pathological process of AD has begun.5 Experts recommend examining neurobehavioral outcomes (e.g. mood, social interaction) to link the pathological process to clinical symptoms, delineate a time course and understand the impact of neurobehavioral outcomes on functional decline prior to overt symptom onset.5

Studies using biomarkers suggest that the natural history of changes in mood and depressive symptoms in particular, begin during the preclinical period.68 However, the current literature on mood in preclinical AD is skewed to the evaluation of singular, negative dimensions of mood (e.g. depression, anxiety). There is a critical need to more comprehensively assess mood disturbances in preclinical AD with global measures of mood to further understand its role and progression in AD. We evaluated the relationship between changes in mood states (positive and negative), changes in cognitive functioning and AD biomarker values (CSF, Positron Emission Technology (PET) amyloid imaging) in a sample of cognitively normal older adults.

Methods

Standard protocol approvals, registrations, and patient consents

All study protocols, consent documents and questionnaires were approved by Washington University Human Research Protection Office. Written informed consent was obtained from all participants and their informants. Study data were collected and managed using REDCap electronic data capture tools12 hosted at Washington University.

Participants

We used available data from participants enrolled in a longitudinal study assessing preclinical AD and driving performance (R01 AG043434) at the Knight Alzheimer’s Disease Research Center (ADRC) at Washington University School of Medicine in St. Louis. The study population was a subset of those enrolled in the affiliated R01 AG043434 study. Participants were part of a rolling enrollment process that occurred from years one to three.

Inclusion criteria

Participants were required to be English-speaking adults, age 65 years or older, willing to take part in PET Pittsburgh Compound B (PiB) amyloid imaging and lumbar puncture for collection of CSF, and be cognitively normal (i.e. no dementia symptoms) as defined by a score of 0 on the Clinical Dementia Rating (CDR).9, 10 Experienced clinicians integrate information from a neurological exam and interviews with the participant and an informant (a family member or close acquaintance) to derive a CDR. The participant serves as their own control where intraindividual changes over time indicate whether symptomatic AD encompassing both MCI due to AD, and AD dementia, is present and if so, its severity.11, 12 A standard scoring algorithm is used to generate the overall CDR (0 cognitively normal, 0.5 very mild impairment, 1 mild, 2 moderate, and 3 severe dementia). More information for CDR scoring may be found at (http://www.biostat.wustl.edu/~adrc/cdrpgm/). The CDR-Sum of Boxes is the sum of the scores from the six CDR domains, ranges from 0–18, and provides a more detailed, continuous measure used to interpret progression of dementia.13 At baseline, participants take part in clinical and psychometric assessments including the Mini-Mental State Examination.14 Biomarkers (PET-PiB amyloid imaging and CSF) are also obtained at baseline. At yearly follow-up, participants repeat the clinical and psychometric assessments.

Pittsburgh Compound B uptake measurement

Participants take part in PET-PiB imaging to determine presence and burden of brain amyloid.15 A binding potential (BP) value reflecting the region of interest (ROI) binding value proportional to the number of binding sites for each ROI is calculated. Dynamic PET imaging was conducted with a Siemens Biograph PET-CT scanner in three-dimensional mode after intravenous administration of approximately 12mCi of PiB. The images were reconstructed on a 128×128×63 matrix (2.12×2.12×2.43 mm) using filtered back-projection. Participants also underwent a structural magnetic resonance imaging (MRI) exam at 3 Tesla, which was segmented using FreeSurfer 5.1 (Martinos Center, Boston, MA). Visual inspection of the automated segmentation results was performed for quality assurance purposes in all datasets. Correction was done when necessary according to the FreeSurfer manual (http://surfer.nmr.mgh.harvard.edu/fswiki/). Regions were combined as described by Su et al.16 The mean cortical binding potential (MCBP) is obtained by taking the mean of the binding potentials from brain regions (prefrontal cortex, gyrus rectus, lateral temporal cortex, and precuneus) known to have high uptake of PiB among participants with AD, as previously described.17, 18

CSF measurement

Following an overnight fast, experienced neurologists use a 22-gauge Sprotte spinal needle to collect 20–30 mL of CSF via standard lumbar puncture (LP) into a polypropylene tube at 8:00 AM. After brief centrifugation, CSF samples were aliquoted and stored at −84°C.19 All biomarker assays include a common reference standard, within-plate sample randomization and strict standardized protocol adherence. β-amyloid42 (Aβ42), total tau, and phosphorylated tau181 (ptau181) levels were measured using enzyme linked immunosorbant assay (ELISA) (INNOTEST; Fujiebio [formerly Innogenetics], Ghent, Belgium). Lower levels of Aβ42 are indicative of amyloid plagues while higher levels of tau and ptau181 indicate neuronal degeneration; ratios of tau/Aβ42 and ptau181/Aβ42 predict cognitive decline.20

Outcome measures

Three outcome measures were evaluated. The Profile of Mood States -Short Form (POMS-SF) is a 30-item questionnaire that assesses different mood states (both positive and negative) and total mood disturbance (TMD).21 There are six subscales with individual t-scores (30–80) to detect impairment in anxiety, depression, anger, vigor, fatigue and confusion. Higher scores on the vigor subscale indicate more positive mood, and higher scores on the remaining subscales indicate more negative mood. The TMD score is calculated by summing scores on the anxiety, depression, anger, fatigue, and confusion subscales, and then subtracting the vigor subscale score. The TMD scale ranges from −20 to 100 with higher scores indicating greater mood disturbance. The Geriatric Depression Scale short form (GDS) is a 15-item screen for depression in older adults.22 Respondents answer “yes” or “no” for each item, with higher scores (range: 0–15) indicating greater depression. The Neuropsychiatric Inventory Questionnaire (NPI-Q) consists of 12 items, each assessing 12 behavioral domains.23 The informant indicates whether each symptom is present (“yes” or “no”). For each “yes” response, the informant is then asked to rate the severity of the symptom (mild, moderate, or severe). Higher scores (range: 0–36) on the NPI-Q indicate greater neuropsychiatric symptoms.

Statistical analyses

For cross-sectional analyses, general linear models were used to determine whether MCBP values (treated as dichotomous using a median split) were associated with TMD scores at baseline (dependent variable treated as continuous), after adjustment for age, gender, and education. This analysis was repeated five times, once for each of the CSF variables at baseline (Aβ42, tau, ptau181, tau/Aβ42 and ptau181/Aβ42). Lower values of Aβ42, and higher values of all other biomarkers, are associated with preclinical AD. When appropriate, given the distribution of scores, we then repeated the analyses using the biomarker variables listed above for the POMS-SF subscales, and the GDS and NPI-Q measures. As with the analyses of MCBP, all other analyses treated the biomarker variables as dichotomous, where the variables reflected lower and higher biomarker values constructed using median splits. For the longitudinal analyses, a change score from baseline to one-year follow-up was calculated, separately for TMD scores, POMS-SF subscales, GDS and NPI-Q measures. This score indicates the change from baseline to follow-up and can be either positive or negative. All analyses were repeated using the respective change scores in place of baseline scores. Finally, we examined changes in memory using the CDR, MMSE, Wechsler Memory Scales24 task of Associate Learning and Delayed Logical Memory, and the free-recall portion of the Free and Cued Selective Reminding Test.25

Results

Data were available for 118 participants at baseline and 66 participants at a follow-up that took place approximately one-year later. Due to rolling enrollment into the study, only 66 participants had follow-up data available for longitudinal analyses. Compared to normative scores, baseline scores on all subscales of the POMS-SF27, GDS28 and NPI-Q29 indicate little to no mood disturbances as self-reported by participants or informants (table 1). At baseline, all participants were cognitively normal as indicated by the CDR (all CDR 0) and MMSE scores (mean=29.34, SD=0.95). Additionally, 7.6% of participants had an active mood disorder while 15.3% reported use of an antidepressant. Because there was little variation (scores indicated little to no symptoms) on the GDS, NPI-Q, and depression, anger and fatigue subscales of the POMS-SF at baseline, analyses were not conducted for these measures. The values of each biomarker variable were not associated with the TMD or remaining subscales on the POMS-SF in the cross-sectional analyses (table 2). Additionally, we assessed the correlations between CSF Aβ42 and MCBP for PET PiB at baseline since both biomarkers examine amyloid burden, but in different parts of the central nervous system. When treated as continuous variables, there was no correlation (r = −0.07, p = 0.45), however, there was moderate and statistically significant association when both biomarkers were treated dichotomously (χ2 = 18.32, p = < 0.001).

Table 1.

Demographics.

Total Sample (n = 118)*
Age, y 72.5±4.7
Education, y 16.1±2.6
Women, N 61 (51.7%)
Race, Caucasian, N 107 (90.7%)
Diagnosis of active mood disorder, N 9 (7.6%)
Diagnosis of remote mood disorder, N 5 (4.2%)
Antidepressant use 18 (15.3%)
Biomarkers
CSF Aβ42, pg/mL (N=118) 939.9±1376.7
CSF tau, pg/mL (N=118) 349.6±199.5
CSF ptau181, pg/mL (N=117) 65.8±40.0
CSF tau/Aβ42 (N=118) 0.52±0.47
CSF ptau181/Aβ42 (N=117) 0.09±0.07
MCBP for PET PiB (N=116) 0.17±0.23
Baseline Follow up
POMS-SF Total Mood Disturbance −5.6±7.5 −5.3±7.1
Tension-Anxiety 33.0±4.0 32.7±3.6
Depression 32.3±1.3 32.7±1.8
Anger 36.3±1.1 36.4±1.4
Vigor 60.0±9.8 58.9±8.6
Fatigue 32.2±3.9 32.1±3.4
Confusion 37.6±3.0 37.7±2.8
GDS (N=91|50) 0.99±1.5 0.92±1.1
NPI-Q (N=93|50) 0.74±1.8 1.00±1.7

Abbreviations: POMS-SF = Profile of Mood States – Short Form; Aβ42 = β-amyloid42; ptau181 = phosphorylated tau181; MCBP = Mean Cortical Binding Potential; PiB = Pittsburgh compound B; GDS = Geriatric Depression Scale; NPI-Q = Neuropsychiatric Inventory Questionnaire.

*

Mean±SD or percentage

Table 2.

Cross-sectional baseline differences in mood after adjusting for age, gender and education.

CSF Aβ42
CSF tau
CSF ptau181
Lower* Higher* p Lower Higher p Lower Higher p
POMS TMD −4.9 (−6.9 to −2.9) −6.4 (−8.3 to −4.4) 0.29 −6.1 (−8.0 to −4.1) −5.2 (−7.2 to −3.1) 0.51 −5.9 (−7.9 to −4.0) −5.3 (−7.3 to −3.3) 0.66
 Anxiety 33.1 (32.1 to 34.2) 32.9 (31.9 to 34.0) 0.80 32.0 (31.8 to 33.9) 33.2 (32.2 to 34.3) 0.64 32.6 (31.6 to 33.7) 33.5 (32.4 to 34.5) 0.27
 Vigor 59.7 (57.1 to 62.3) 60.5 (57.9 to 63.0) 0.66 60.3 (57.7 to 62.8) 59.9 (57.3 to 62.5) 0.83 59.9 (57.3 to 62.4) 60.3 (57.7 to 63.0) 0.80
 Confusion 38.2 (37.4 to 40.0) 37.1 (36.3 to 37.9) 0.06 37.8 (37.0 to 38.6) 37.5 (36.7 to 38.6) 0.62 37.8 (37.0 to 38.6) 37.5 (36.7 to 38.3) 0.57
CSF tau/Aβ42
CSF ptau/Aβ42
MCBP for PET PIB
Lower Higher p Lower Higher p Lower Higher p
POMS TMD −5.8 (−7.8 to −3.8) −5.5 (−7.0 to −3.5) 0.86 −5.9 (−7.9 to −4.0) −5.2 (−7.3 to −3.2) 0.63 −6.1 (−8.1 to −4.1) −5.0 (−7.1 to −3.0) 0.45
 Anxiety 33.1 (32.1 to 34.2) 32.8 (31.8 to 33.9) 0.69 32.8 (31.7 to 33.8) 33.4 (32.3 to 34.4) 0.43 32.7 (31.7 to 33.8) 33.5 (32.4 to 34.5) 0.33
 Vigor 59.9 (57.3 to 62.5) 60.4 (57.8 to 63.1) 0.78 60.0 (57.4 to 62.5) 60.3 (57.6 to 63.9) 0.87 60.1 (57.6 to 62.7) 60.2 (57.5 to 62.9) 0.95
 Confusion 37.3 (36.5 to 38.1) 37.9 (37.1 to 38.8) 0.28 37.4 (36.6 to 38.1) 38.0 (37.2 to 38.8) 0.28 37.4 (36.6 to 38.2) 38.1 (37.3 to 38.9) 0.23

Abbreviations: POMS-SF = Profile of Mood States – Short Form; TMD = Total Mood Disturbance; Aβ42 = β-amyloid42; ptau181 = phosphorylated tau181; MCBP = Mean Cortical Binding Potential; PiB = Pittsburgh compound B.

*

Mean (95% Confidence Interval)

In the longitudinal analyses, cognitive decline was not statistically significant as indicated by the CDR-Sum of Boxes (t = 1.09, p = .279) or MMSE (t = −1.78, p = .859) from baseline to follow-up. One participant at follow-up had a CDR 0.5 but an MMSE score of 30. All other participants remained at CDR 0 at follow-up. We further examined change on neuropsychological tasks of memory, which may be more sensitive to early-related AD changes and cognitive decline. There was no statistically significant change on the Wechsler Memory Scales task of Associate Learning (t = −1.81, p = .857) and Delayed Logical Memory (t = −0.05, p = .962) or the free-recall score of the Free and Cued Selective Reminding Test (t = −0.52, p = .606). Additionally, there was no significant correlation between the change scores from the three memory tasks, CDR sum of boxes and MMSE and the TMD, NPIQ or GDS (table 3).

Table 3.

Correlation of change scores between measures of mood and neuropsychological memory tasks.

POMS TMD GDS NPI-Q WMS-AL WMS-DLM FCSRT CDR-SOB MMSE
POMS TMD r 1
GDS r −0.006 1
NPI-Q r −0.038 0.345* 1
WMS-AL r −0.121 −0.139 −0.021 1
WMS-DLM r −0.106 0.11 0.089 0.097 1
FCSRT r 0.039 0.024 0.004 0.117 0.192 1
CDR-SOB r 0.017 0.000 0.000 0.074 0.077 0.003 1
MMSE r 0.083 −0.164 0.027 −0.035 −0.065 −0.432* 0.001 1

Abbreviations: POMS-SF = Profile of Mood States – Short Form; TMD = Total Mood Disturbance; GDS = Geriatric Depression Scale; NPI-Q = Neuropsychiatric Inventory Questionnaire; WMS-AL = Wechsler Memory Scales- Associate Learning; WMS-DLM = Wechsler Memory Scales-Delayed Logical Memory; FCSRT = Free and Cued Selective Reminding Test;

*

= p < .05

In the longitudinal analyses for mood changes, each of the change score outcomes was normally distributed and showed enough variation for statistical analysis. Participants with higher values of CSF tau/Aβ42 had greater increases in total mood disturbance scores across the one-year follow-up period compared to participants with lower values (table 4). Higher values of CSF tau/Aβ42 also predicted greater increases longitudinally in the negative emotions of anxiety, depression and confusion as measured by the POMS-SF subscales. MCBP for PET-PiB had a significant association with change in GDS scores, such that participants with greater increases in depressive affect across the follow-up period were in the MCBP group with higher values. Finally, CSF tau, ptau181, tau/Aβ42 and MCBP biomarkers also predicted change scores on the NPI-Q. Higher values of each of these biomarkers were associated with greater increases in neuropsychiatric symptoms across time.

Table 4.

Differences in mood change scores after adjusting for age, gender and education.

CSF Aβ42
CSF tau
Lower* Higher* p Lower Higher p
POMS TMD 1.26 (−0.85 to 3.38) −1.09 (−3.17 to 0.99) 0.12 −0.89 (−3.04 to 1.25) 1.08 (−1.07 to 3.23) 0.20
 Anxiety 0.83 (−0.14 to 1.81) 0.56 (−0.40 to 1.52) 0.69 0.12 (−0.83 to 1.10) 1.27 (0.31 to 2.23) 0.10
 Depression −0.51 (−1.44 to 0.13) −0.17 (−0.80 to 0.46) 0.46 −0.76 (−1.39 to −0.13) 0.07 (−0.55 to 0.70) 0.07
 Anger −0.00 (−0.50 to 0.50) −0.36 (−0.85 to 0.13) 0.32 −0.32 (−0.82 to 0.19) −0.06 (−.56 to 0.45) 0.47
 Vigor −1.09 (−4.14 to 1.95) 1.79 (−1.21 to 4.79) 0.18 0.43 (−2.69 to 3.55) 0.29 (−2.87 to 3.41) 0.95
 Fatigue 0.67 (−0.48 to 1.82) −0.83 (−1.96 to 0.30) 0.67 −0.63 (−1.80 to 0.54) 0.51 (−0.66 to 1.68) 0.17
 Confusion 0.44 (−0.55 to 1.43) 0.26 (−0.72 to 1.23) 0.79 0.07 (−0.93 to 1.06) 0.65 (−0.34 to 1.65) 0.41
GDS −0.06 (−0.50 to 0.39) −0.44 (−0.87 to −0.01) 0.22 −0.27 (−0.76 to 0.21) −0.25 (−0.68 to 0.17) 0.95
NPI−Q −0.27 (−0.89 to 0.36) −0.68 (−1.30 to −0.07) 0.34 −1.33 (−1.95 to −0.71) 0.14 (−0.40 to 0.68) 0.001
CSF ptau181
CSF tau/Aβ42
Lower Higher p Lower Higher p
POMS TMD −0.74 (−2.90 to 1.41) 0.81 (−1.35 to 2.96) 0.32 −1.97 (−4.02 to 0.07) 2.21 (0.17 to 4.24) 0.005
 Anxiety 0.20 (−0.77 to 1.17) 1.21 (0.24 to 2.18) 0.15 −0.15 (−1.09 to 0.79) 1.50 (0.56 to 2.43) 0.02
 Depression −0.76 (−1.39 to −0.13) 0.07 (−0.55 to 0.70) 0.07 −0.81 (−1.44 to −0.19) 0.13 (−0.50 to 0.75) 0.04
 Anger −0.32 (−0.83 to 0.18) −0.05 (−0.56 to 0.45) 0.45 −0.44 (−0.95 to 0.06) 0.07 (−0.43 to 0.57) 0.15
 Vigor 0.43 (−2.69 to 3.54) 0.42 (−2.70 to 3.53) 1.00 1.67 (−1.42 to 4.76) −1.00 (−4.08 to 2.08) 0.23
 Fatigue −0.40 (−1.59 to 0.79) 0.21 (−0.97 to 1.40) 0.47 −0.82 (−1.97 to 0.33) 0.75 (−0.40 to 1.90) 0.59
 Confusion 0.13 (−0.86 to 1.13) 0.52 (−0.48 to 1.52) 0.59 −0.33 (−1.28 to 0.63) 1.12 (0.16 to 2.07) 0.04
GDS −0.32 (−0.77 to 0.13) −0.21 (−0.66 to 0.25) 0.73 −0.40 (−0.90 to 0.10) −0.16 (−0.57 to 0.25) 0.47
NPI-Q −0.93 (−1.53 to −0.34) −0.05 (−0.67 to 0.57) 0.05 −1.30 (−1.95 to −0.64) 0.05 (−0.49 to 0.59 0.003
CSF ptau/Aβ42 MCBP for PET PIB

Lower Higher p Lower Higher p
POMS TMD −1.30 (−3.42 to 0.82) 1.36 (−0.75 to 3.48) 0.08 −1.20 (−3.47 to 1.06) 0.84 (−1.44 to 3.13) 0.21
 Anxiety −0.22 (−1.15 to 0.71) 1.63 (0.70 to 2.55) 0.007 0.23 (−0.81 to 1.27) 1.23 (0.18 to 2.28) 0.19
 Depression −0.70 (−1.34 to −0.07) 0.02 (−0.62 to 0.65) 0.12 −0.45 (−1.11 to 0.22) −0.16 (−0.83 to 0.51) 0.54
 Anger −0.31 (−0.82 to 0.20) −0.06 (−0.57 to 0.44) 0.50 −0.56 (−1.08 to −0.04) 0.23 (−0.29 to 0.75) 0.04
 Vigor 0.87 (−2.25 to 3.98) − 0.02 (−3.13 to 3.09) 0.69 1.13 (−2.08 to 4.33) 0.66 (−2.57 to 3.89) 0.84
 Fatigue −0.55 (−1.73 to 0.63) 0.37 (−0.82 to 1.55) 0.28 −0.72 (−1.96 to 0.52) 0.26 (−0.99 to 1.51) 0.27
 Confusion −0.01 (−0.10 to 0.99) 0.66 (−0.33 to 1.65) 0.35 0.12 (−0.92 to 1.16) 0.73 (−0.32 to 1.79) 0.41
GDS −0.40 (−0.88 to 0.07) −0.16 (−0.59 to 0.28) 0.44 −0.73 (−1.20 to −0.26) 0.08 (−0.33 to 0.49) 0.01
NPI−Q −0.94 (−1.60 to −0.29) −0.18 (−0.78 to 0.42) 0.09 −1.07 (−1.74 to −0.39) −0.13 (−0.70 to 0.45) 0.04

Abbreviations: POMS-SF = Profile of Mood States – Short Form; TMD = Total Mood Disturbance; Aβ42 = β-amyloid42; ptau181 = phosphorylated tau181; MCBP = Mean Cortical Binding Potential; PiB = Pittsburgh compound B; GDS = Geriatric Depression Scale; NPI-Q = Neuropsychiatric Inventory Questionnaire.

*

Mean (95% Confidence Interval)

Also significant using the Benjamini-Hochberg procedure

Discussion

Our findings suggest that CSF and PET-PiB biomarkers are able to predict increased mood disturbances longitudinally over one-year in a sample of cognitively normal older adults (CDR 0), but are unrelated to mood at baseline when assessed cross-sectionally. The lack of a relationship between measures of mood and the biomarkers at baseline together with the finding that participants with more abnormal biomarker values at baseline had higher change scores on mood measures (i.e., increasing mood disturbance) across a short follow-up period (one-year) suggests that the presence of abnormal AD biomarker levels are associated with a measurable impact on mood as the disease processes, although these changes are subtle and may not be recognized without formal testing. These findings extend our previous work26 on the development of depressive and neuropsychiatric symptoms in participants with CDR > 0.

Traditional measures of mood in the AD literature have been useful in establishing the prevalence of mood disturbances and elucidating its relationship to cognitive decline. However, these scales are limited since assessment of mood is based upon a single mood state (e.g. GDS) or behavioral symptoms (NPI-Q). Mood states are commonly misinterpreted as being polar opposites (e.g. happy vs. sad, fatigue vs. energetic). However, this unilateral relationship is limited because mood states are multidimensional where an individual can experience multiple mood states at the same time. The POMS-SF has been used in samples of older adults to assess mood states; it provides a more comprehensive perspective of mood since it assesses negative mood states like anxiety and depression but also a positive mood state like vigor. All subscales are incorporated to give a total mood disturbance score. Changes in the POMS-SF TMD and anxiety, depression and confusion subscales were all associated with AD biomarker values. Vigor was not statistically different between the biomarker groups, suggesting that negative moods drive the shift in total disturbance score.

A recent study using a PET imaging FDDNP tracer found the POMS-SF vigor subscale to be associated with lower binding of amyloid and tau in participants with mild cognitive impairment, but not controls.27 The vigor subscale was inversely correlated with FDDNP-PET binding in the posterior cingulate region. Since FDDNP labels amyloid plaques and neurofibrillary tangles, changes in vigor may be driven by tau, which is best known to correlate with measures of cognitive decline.28 Although that cross-sectional study had a modest sample size and only examined the POMS-SF vigor subscale, it suggests that positive moods like vigor may be related to biomarker values in early AD. Speculatively, changes in vigor may occur later in preclinical AD, whereas negative mood changes may occur earlier.

Previous work has examined the relationship between amyloid and either depressive symptomology or depressive disorder. One study found strong associations between CSF Aβ42,40 and their ratios in cognitively normal populations with depressive symptoms and major depressive disorder.29 Another study showed a relationship between abnormal CSF Aβ42 and anxiety and agitation, but not depression and apathy in persons with MCI.30 Others have found no relationship between depression scores and CSF Aβ42, tau and ptau181 for persons with AD and cognitively normal persons.31 Studies using PET-PiB imaging have found increased PiB retention in persons with depression with and without cognitive impairment suggesting a shared biological mechanism between AD and depression.32 See table 5 for other studies examining amyloid-depression relationships, including those that have examined plasma measures of amyloid. Many of these studies are cross-sectional analyses and may not have collected longitudinal data to examine change over time.

Table 5.

Representative studies examining relationships between amyloid and depression using CSF, imaging and plasma

Biomarker type Study Sample(s) Results Biomarker relationship to depression found?
CSF Pomara, N et al. 2012 American Journal of Psychiatry. 169(5), 523–530. 28 with MDD

19 no depression
Elderly individuals with depression had lower levels of CSF Aβ42 compared to those without depression Yes
Diniz, BS et al. 2014. The Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 69(6), 845–851. 25 antidepressant-free patients with late life depression.

25 HC
No significant differences in AD related biomarkers between groups No
Amyloid Imaging Tateno, A et al. 2015. International journal of geriatric psychiatry, 30(7), 720–728. 33 MCI patients with a history of GD

22 HC
In depression and MCI groups, amyloid uptake was significantly associated with depression onset. Yes
Wu, KY et al. 2014. European journal of nuclear medicine and molecular imaging, 41(4), 714–722. 25 depressed patients

11 non-depressed patients
Depressed patients had greater amyloid uptake in parietal, precuneus, frontal, temporal, and occipital regions. Yes
Yasuno, F et al. (2016). International journal of geriatric psychiatry. 29 cognitively normal older participants More depressive symptoms were associated with greater amyloid uptake Yes
Kumar A et al. (2011). Archives of general psychiatry, 68(11), 1143–1150. 20 patients with MDD

19 healthy controls
[18F]FDDNP uptake was higher in posterior cingulate and lateral temporal regions in MDD group compared to healthy controls Yes
Plasma Pomara N et al. (2006). Neurochem Res, 31(3), 341–349. 47 with late life MDD

35 controls
Higher Aβ42 and Aβ42/Aβ40 were found in depressed individuals compared to healthy controls Yes
Kita Y et al. (2009). Psychogeriatrics, 9(4), 180–185. 30 patients with MDD

30 controls
40 and Aβ42 alone did not differ by groups. The Aβ40/Aβ42 ratio was higher in MDD group compared to controls Mixed
Blasko, I et al. (2010). Am. J. Geriat. Psychiatry, 18(11), 973–982. 331 HC 42predicted future first depressive episode Yes

Abbreviations: CSF = Cerebrospinal Fluid; MDD = Major Depressive Disorder; GD = Geriatric Depression; Aβ42 = β-amyloid42; Aβ40 = β-amyloid40; MCI = Mild Cognitive Impairment; HC = Healthy Controls;

In our study, the relationship between the POMS and the biomarkers seemed to be stronger with tau compared to amyloid pathology. Changes in amyloid tend to occur first and before changes in tau within AD. However, neuropsychological measures of cognition have always been more closely correlated with tau pathology (burden and distribution) compared to amyloid pathology.28, 33 The POMS assessment of mood seems to follow this paradigm. Additionally, higher values of CSF tau/Aβ42 and MCBP for PET-PiB had more associations with changes in mood compared to other biomarkers. This relationship suggests that the ratio may be more sensitive to mood changes compared to either CSF Aβ42 or tau alone.19 Strikingly, studies looking at decline in cognition over a relatively short time period (a few years) among persons cognitively normal at baseline have indicated that the ratio of CSF tau/Aβ42 is a better predictor of decline than either Aβ42 or tau by themselves.8, 34 Thus, like cognition, variables reflecting combinations of biomarkers may be better predictors of decline in noncognitive outcomes such as mood, compared to individual biomarkers.

GDS scores were associated with MCBP for PET-PiB but not CSF Aβ42. Although both measures reflect amyloid levels, imaging of amyloid using PET-PiB assesses fibrillar amyloid in the form of plaques, whereas CSF reflects amyloid in a soluble state. The CSF and imaging amyloid measures are related to each other, but there is not a one-to-one correspondence.35

There are several limitations of the study. We found no cognitive decline over one-year on the MMSE and the CDR Sum of Boxes, which may have been due at least in part, to ceiling effects for the MMSE and floor effects for the CDR Sum of Boxes. However, this remained true even when we specifically examined measures of memory without floor and ceiling effects: delayed episodic memory, delayed free recall, and paired associates.

Due to rolling enrollment and testing in this study, data for change score analyses were only available for 66 participants at year one follow up. When comparing groups, statistical power is greatest when the groups have equal numbers of members. To maximize the statistical power of our analyses, we used a median split to divide the groups into those with “higher” and “lower” values of each biomarker. Thus, the groups were not representative of preclinical AD vs. no preclinical AD. In most studies, the cutoffs used for defining preclinical AD result in approximately 30% of cognitively normal participants being labeled as having preclinical AD. Indeed, in previous studies with larger samples, we used cutoffs of 459 pg/mL for Aβ42, 339 pg/mL for tau, and 67 pg/mL for p-tau181 for CSF biomarkers36 and a cutoff of .18 for MCBP for PiB37 to define preclinical AD. As expected, assigning participants to groups based on the higher 50% and lower 50% of the specific biomarker distribution resulted in lower cutoffs. Specifically, cutoffs used here were 459 pg/mL for Aβ42, 303 pg/mL for tau, and 54 pg/mL for p-tau181 for CSF biomarkers and .066 for MCBP for PiB.

Using an alpha of .05, as we did in this study, one would expect one out of 20 statistical tests to be significant by chance. Since we conducted N=54 tests in our change analyses (table 4), we would expect 54/20 = 2.7 of those tests might be significant by chance alone. However, we found N=11 (i.e., more than 4 times the expected number if due to chance) significant test results, and all change scores were in the expected direction (i.e., the direction of declining performance). The ultimate test of which of the results in this study, if any, are due to chance will be determined by future attempts at replication. However, in the meantime, we have indicated in Table 4 which 6 of the 11 tests are significant using the Benjamini-Hochberg procedure with a false discovery rate of 20%.38, 39

In this initial study that comprehensively assessed mood (positive and negative) in preclinical AD, we found that CSF and PET-PiB biomarkers predicted changes in mood over a one-year follow-up in a sample of adults ≥ 65 years and cognitively normal at baseline. Greater mood disturbances, including negative symptoms such as anxiety, depression and confusion were associated with biomarker values, despite no cognitive decline. These findings support the use of biomarkers in identifying changes in noncognitive outcomes such as mood in preclinical AD. Further work is needed to delineate the natural course of mood disturbances in preclinical AD and the resulting impact on functional activity.

Acknowledgments

Funding for this study was provided by the National Institute on Aging [R01-AG043434, P50-AG05681, P01-AG03991, P01-AG026276, and K12 HD001459]; Fred Simmons and Olga Mohan, and the Charles and Joanne Knight Alzheimer’s Research Initiative of the Washington University Knight Alzheimer’s Disease Research Center (ADRC). The authors thank the participants, investigators, and staff of the Knight ADRC Clinical Core (participant assessments), the investigators and staff of the Driving Performance In Preclinical Alzheimer’s Disease study (R01-AG043434.), the investigators and staff of the Biomarker Core for the Adult Children Study (P01-AG026276) for cerebrospinal fluid analysis, and the investigators and staff of the Imaging Core of the Healthy Aging and Senile Dementia study (P01-AG03991) for amyloid imaging. Imaging facilities were supported by the Washington University Institute of Clinical and Translational Sciences grant UL1-TR000448 from the National Center for Advancing Translational Sciences (NCATS) of the National Institutes of Health (NIH). Imaging analyses used the services of the Neuroimaging Informatics and Analysis Center, supported by NIH grant 5P30NS048056.

Disclosures

Dr. Babulal reports no disclosures.

Dr. Ghoshal has participated in clinical trials of antidementia drugs sponsored by: Bristol Myers Squibb, Eli Lilly/Avid Radiopharmaceuticals, Janssen Immunotherapy, Novartis, Pfizer, Wyeth, SNIFF (The Study of Nasal Insulin to Fight Forgetfullness) study, and A4 (The Anti-Amyloid Treatment in Asymptomatic Alzheimer’s Disease) trial.

Dr. Head reports no disclosures.

Ms. Vernon reports no disclosures.

Dr. Holtzman reports receiving research support from Eli Lilly and C2N Diagnostics smf consulting for AstraZeneca, Genentech, and Eli Lilly. He is on the Scientific Advisory Board of C2N Diagnostics and Neurophage. He co-founded and has ownership interests in C2N Diagnostics.

Dr. Benzinger reports support from research grants from Avid Radiopharmaceuticals (a wholly owned subsidiary of Eli Lilly) and clinical trials involving Eli Lilly and Hoffman LaRoche

Dr. Fagan reports being on the scientific advisory boards of IBL International and Roche and is a consultant for AbbVie and Novartis.

Dr. John C. Morris reports that he owns stock or has equity interest (outside of mutual funds or other externally directed accounts) in any pharmaceutical or biotechnology company. Dr. Morris has participated or is currently participating in clinical trials of antidementia drugs sponsored by the following companies: Janssen Immunotherapy, Pfizer, Eli Lilly/Avid Radiopharmaceuticals, SNIFF (The Study of Nasal Insulin to Fight Forgetfullness) study, and A4 (The Anti-Amyloid Treatment in Asymptomatic Alzheimer’s Disease) trial. Dr. Morris has served as a consultant for Lilly USA, ISIS Pharmaceuticals, and Charles Dana Foundation. He receives research support from Eli Lilly/Avid Radiopharmaceuticals and is funded by NIH grants # P50AG005681; P01AG003991; P01AG026276 and U19AG032438.

Dr. Roe reports no disclosures.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Author Contributions:

GMB: Study concept and design, data analysis and interpretation, drafting and critical revision of manuscript.

NG: Data interpretation, drafting and critical revision of manuscript.

DH: Data interpretation, critical revision of manuscript.

EKV: Data interpretation, critical revision of manuscript.

DMH: Data acquisition and interpretation, critical revision of manuscript.

TLSB: Data acquisition, and interpretation, critical revision of manuscript.

AMF: Data acquisition, and interpretation, critical revision of manuscript.

JCM: Data acquisition, and interpretation, critical revision of manuscript.

CMR: Study concept and design, data analysis and interpretation, drafting and critical revision of manuscript.

References

  • 1.Berger A-K, Fratiglioni L, Forsell Y, et al. The occurrence of depressive symptoms in the preclinical phase of AD A population-based study. Neurology. 1999;53:1998–1998. doi: 10.1212/wnl.53.9.1998. [DOI] [PubMed] [Google Scholar]
  • 2.Schmid NS, Taylor KI, Foldi NS, et al. Neuropsychological signs of Alzheimer’s disease 8 years prior to diagnosis. Journal of Alzheimer’s Disease. 2013;34:537–546. doi: 10.3233/JAD-121234. [DOI] [PubMed] [Google Scholar]
  • 3.Bartolini M, Coccia M, Luzzi S, et al. Motivational symptoms of depression mask preclinical Alzheimer’s disease in elderly subjects. Dementia and geriatric cognitive disorders. 2004;19:31–36. doi: 10.1159/000080968. [DOI] [PubMed] [Google Scholar]
  • 4.Bidzan M, Bidzan L. Neurobehavioral manifestation in early period of Alzheimer disease and vascular dementia. Psychiatria polska. 2013;48:319–330. [PubMed] [Google Scholar]
  • 5.Sperling RA, Aisen PS, Beckett LA, et al. Toward defining the preclinical stages of Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimer’s & Dementia. 2011;7:280–292. doi: 10.1016/j.jalz.2011.03.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Lavretsky H, Siddarth P, Kepe V, et al. Depression and anxiety symptoms are associated with cerebral FDDNP-PET binding in middle-aged and older nondemented adults. The American Journal of Geriatric Psychiatry. 2009;17:493–502. doi: 10.1097/jgp.0b013e3181953b82. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Gustafson DR, Skoog I, Rosengren L, et al. Cerebrospinal fluid beta-amyloid 1–42 concentration may predict cognitive decline in older women. Journal of neurology, neurosurgery, and psychiatry. 2007;78:461–464. doi: 10.1136/jnnp.2006.100529. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Roe CM, Fagan AM, Grant EA, et al. CSF biomarkers of Alzheimer disease “Noncognitive” outcomes. Neurology. 2013;81:2028–2031. doi: 10.1212/01.wnl.0000436940.78152.05. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Burke WJ, Miller JP, Rubin EH, et al. Reliability of the Washington University clinical dementia rating. Archives of neurology. 1988;45:31–32. doi: 10.1001/archneur.1988.00520250037015. [DOI] [PubMed] [Google Scholar]
  • 10.Morris JC. The Clinical Dementia Rating (CDR): current version and scoring rules. Neurology. 1993 doi: 10.1212/wnl.43.11.2412-a. [DOI] [PubMed] [Google Scholar]
  • 11.Albert MS, DeKosky ST, Dickson D, et al. The diagnosis of mild cognitive impairment due to Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimer’s & dementia. 2011;7:270–279. doi: 10.1016/j.jalz.2011.03.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.McKhann GM, Knopman DS, Chertkow H, et al. The diagnosis of dementia due to Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimer’s & dementia. 2011;7:263–269. doi: 10.1016/j.jalz.2011.03.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.O’Bryant SE, Waring SC, Cullum CM, et al. Staging dementia using Clinical Dementia Rating Scale Sum of Boxes scores: a Texas Alzheimer’s research consortium study. Archives of neurology. 2008;65:1091–1095. doi: 10.1001/archneur.65.8.1091. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Folstein MF, Folstein SE, McHugh PR. “Mini-mental state”: a practical method for grading the cognitive state of patients for the clinician. Journal of psychiatric research. 1975;12:189–198. doi: 10.1016/0022-3956(75)90026-6. [DOI] [PubMed] [Google Scholar]
  • 15.Klunk WE, Engler H, Nordberg A, et al. Imaging brain amyloid in Alzheimer’s disease with Pittsburgh Compound-B. Annals of neurology. 2004;55:306–319. doi: 10.1002/ana.20009. [DOI] [PubMed] [Google Scholar]
  • 16.Su Y, D’Angelo GM, Vlassenko AG, et al. Quantitative analysis of PiB-PET with FreeSurfer ROIs. PloS one. 2013;8:e73377. doi: 10.1371/journal.pone.0073377. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Morris JC, Roe CM, Grant EA, et al. Pittsburgh compound B imaging and prediction of progression from cognitive normality to symptomatic Alzheimer disease. Archives of neurology. 2009;66:1469–1475. doi: 10.1001/archneurol.2009.269. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Storandt M, Mintun MA, Head D, et al. Cognitive decline and brain volume loss as signatures of cerebral amyloid-β peptide deposition identified with Pittsburgh compound B: cognitive decline associated with Aβ deposition. Archives of neurology. 2009;66:1476–1481. doi: 10.1001/archneurol.2009.272. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Fagan AM, Roe CM, Xiong C, et al. Cerebrospinal fluid tau/β-amyloid42 ratio as a prediction of cognitive decline in nondemented older adults. Archives of neurology. 2007;64:343–349. doi: 10.1001/archneur.64.3.noc60123. [DOI] [PubMed] [Google Scholar]
  • 20.Blennow K, Zetterberg H, Fagan AM. Fluid biomarkers in Alzheimer disease. Cold Spring Harbor perspectives in medicine. 2012;2:a006221. doi: 10.1101/cshperspect.a006221. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Curran SL, Andrykowski MA, Studts JL. Short form of the Profile of Mood States (POMS-SF): Psychometric information. Psychological Assessment. 1995;7:80. [Google Scholar]
  • 22.Montorio I, Izal M. The Geriatric Depression Scale: a review of its development and utility. International Psychogeriatrics. 1996;8:103–112. doi: 10.1017/s1041610296002505. [DOI] [PubMed] [Google Scholar]
  • 23.Kaufer DI, Cummings JL, Ketchel P, et al. Validation of the NPI-Q, a brief clinical form of the Neuropsychiatric Inventory. The Journal of neuropsychiatry and clinical neurosciences. 2000 doi: 10.1176/jnp.12.2.233. [DOI] [PubMed] [Google Scholar]
  • 24.Wechsler D, Stone CP. Wechsler memory scale manual. Psychological Corporation; 1973. [Google Scholar]
  • 25.Grober E, Buschke H, Crystal H, et al. Screening for dementia by memory testing. Neurology. 1988;38:900–900. doi: 10.1212/wnl.38.6.900. [DOI] [PubMed] [Google Scholar]
  • 26.Masters MC, Morris JC, Roe CM. “Noncognitive” symptoms of early Alzheimer disease A longitudinal analysis. Neurology. 2015;84:617–622. doi: 10.1212/WNL.0000000000001238. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Chen ST, Siddarth P, Saito NY, et al. Psychological well-being and regional brain amyloid and tau in mild cognitive impairment. The American Journal of Geriatric Psychiatry. 2014;22:362–369. doi: 10.1016/j.jagp.2012.09.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Amariglio RE, Mormino EC, Pietras AC, et al. Subjective cognitive concerns, amyloid-beta, and neurodegeneration in clinically normal elderly. Neurology. 2015;85:56–62. doi: 10.1212/WNL.0000000000001712. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Gudmundsson P, Skoog I, Waern M, et al. The relationship between cerebrospinal fluid biomarkers and depression in elderly women. The American journal of geriatric psychiatry. 2007;15:832–838. doi: 10.1097/JGP.0b013e3180547091. [DOI] [PubMed] [Google Scholar]
  • 30.Ramakers IHGB, Verhey FRJ, Scheltens P, et al. Anxiety is related to Alzheimer cerebrospinal fluid markers in subjects with mild cognitive impairment. Psychological medicine. 2013;43:911–920. doi: 10.1017/S0033291712001870. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Kramberger MG, Jelic V, Kåreholt I, et al. Cerebrospinal Fluid Alzheimer Markers in Depressed Elderly Subjects with and without Alzheimer’s Disease. Dementia and Geriatric Cognitive Disorders EXTRA. 2012;2:48–56. doi: 10.1159/000334644. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Butters MA, Klunk WE, Mathis CA, et al. Imaging Alzheimer pathology in late-life depression with PET and Pittsburgh Compound-B. Alzheimer disease and associated disorders. 2008;22:261. doi: 10.1097/WAD.0b013e31816c92bf. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Ghoshal N, Garc F, Wuu J, et al. Tau conformational changes correspond to impairments of episodic memory in mild cognitive impairment and Alzheimer’s disease. Experimental neurology. 2002;177:475–493. doi: 10.1006/exnr.2002.8014. [DOI] [PubMed] [Google Scholar]
  • 34.Li G, Sokal I, Quinn J, et al. CSF tau/Aβ42 ratio for increased risk of mild cognitive impairment A follow-up study. Neurology. 2007;69:631–639. doi: 10.1212/01.wnl.0000267428.62582.aa. [DOI] [PubMed] [Google Scholar]
  • 35.Fagan AM, Mintun MA, Shah AR, et al. Cerebrospinal fluid tau and ptau181 increase with cortical amyloid deposition in cognitively normal individuals: implications for future clinical trials of Alzheimer’s disease. EMBO molecular medicine. 2009;1:371–380. doi: 10.1002/emmm.200900048. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Vos SJB, Xiong C, Visser PJ, et al. Preclinical Alzheimer’s disease and its outcome: a longitudinal cohort study. The Lancet Neurology. 2013;12:957–965. doi: 10.1016/S1474-4422(13)70194-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Roe CM, Mintun MA, D’Angelo G, et al. Alzheimer disease and cognitive reserve: variation of education effect with carbon 11–labeled Pittsburgh Compound B uptake. Archives of Neurology. 2008;65:1467–1471. doi: 10.1001/archneur.65.11.1467. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. Journal of the Royal Statistical Society. Series B (Methodological) 1995:289–300. [Google Scholar]
  • 39.McDonald JH. Handbook of biological statistics. Sparky House Publishing; Baltimore, MD: 2009. [Google Scholar]

RESOURCES