HHS Public Access Author manuscript Author Manuscript
J Alzheimers Dis. Author manuscript; available in PMC 2016 August 06. Published in final edited form as: J Alzheimers Dis. 2016 June 18; 53(3): 1097–1105. doi:10.3233/JAD-150987.
Lower late-life body-mass index is associated with higher cortical amyloid burden in clinically normal elderly David C. Hsu, MDa,d,f, Elizabeth C. Mormino, PhDb, Aaron P. Schultz, PhDb, Rebecca E. Amariglio, PhDb,d, Nancy J. Donovan, MDa,d,e, Dorene M. Rentz, PsyDb,d,e, Keith A. Johnson, MDb,c,d, Reisa A. Sperling, MDb,d, and Gad A. Marshall, MDb,d,* for the Harvard Aging Brain Study
Author Manuscript
aDepartment
of Psychiatry, Massachusetts General Hospital, Harvard Medical School, Boston, MA 02114, USA
bDepartment
of Neurology, Massachusetts General Hospital, Harvard Medical School, Boston, MA 02114, USA
cDepartment
of Radiology, Massachusetts General Hospital, Harvard Medical School, Boston, MA
02114, USA dCenter
for Alzheimer Research and Treatment, Department of Neurology, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA 02115, USA
eDepartment
of Psychiatry, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA
02115, USA
Author Manuscript
fDepartment
of Psychiatry, Mercy Medical Group, Sacramento, CA 95815
Abstract Background—Lower body-mass index (BMI) in late life has been associated with an increased risk of dementia, and weight loss has been associated with more rapid decline in Alzheimer disease (AD) dementia. Objective—To explore the association between BMI and cortical amyloid burden in clinically normal (CN) elderly at risk for AD dementia.
Author Manuscript
Methods—Cross-sectional analyses were completed using baseline data from the Harvard Aging Brain Study, consisting of 280 community-dwelling CN older adults aged 62–90. Assessments included medical histories and physical exam, Pittsburgh compound B (PiB) positron emission tomography (PET) amyloid imaging, and apolipoprotein E ɛ4 (APOE4) genotyping. For the primary analysis, a general linear regression model was used to evaluate the association of BMI with PiB retention. Covariates included age, sex, years of education, and APOE4 carrier status. Secondary analyses were performed for BMI subdivisions (normal, overweight, obese), APOE4 carriers, and BMI × APOE4 interaction.
*
Correspondence to: Gad A. Marshall, MD, Center for Alzheimer Research and Treatment, Brigham and Women’s Hospital, 221 Longwood Avenue, BL-104H, Boston, MA 02115, P: 617-732-8085, F: 617-264-5212, [email protected]. Results of this manuscript were presented at the 9th Human Amyloid Imaging meeting on January 15, 2015. The authors have no relevant conflicts of interest to disclose.
Hsu et al.
Page 2
Author Manuscript
Results—In the primary analysis, greater PiB retention was associated with lower BMI (β= −0.14, p=0.02). In the secondary analyses, APOE4 carrier status (β=−0.27, p=0.02) and normal BMI (β=−0.25, p=0.01), as opposed to overweight or obese BMI, were associated with greater PiB retention. The BMI × APOE4 interaction was also significant (β=−0.14, p=0.04). Conclusions—This finding offers new insight into the role of BMI at the preclinical stage of AD, wherein lower BMI late in life is associated with greater cortical amyloid burden. Future studies are needed to elucidate the mechanism behind this association, especially in those with lower BMI who are APOE4 carriers. Keywords Alzheimer’s disease; amyloid; apolipoprotein E; body-mass index; clinically normal elderly; Pittsburgh compound B; positron emission tomography
Author Manuscript
Introduction An obesity paradox [1] exists in Alzheimer disease (AD) and dementia. Midlife obesity can increase the risk of late-life dementia [2, 3], but lower late-life body-mass index (BMI) is also related to worse AD outcomes. Higher late-life BMI is associated with decreased risk of dementia [4, 5], better cognition [6], slower progression of AD [7], and decreased mortality [8]. Weight loss over the clinical course of AD has been shown in several large cohort studies [9–13]. Early reports focused on those only in the dementia phase of AD, but later studies have explored earlier stages of the disease [7]. Patients with AD enter the dementia phase at a lower weight, and this weight loss increases mortality [13].
Author Manuscript
A 2007 study has suggested that weight loss for all causes of dementia may start 10–20 years prior to diagnosis [11]. Autopsy studies have reported that AD pathology has been associated with low body weight [14, 15]. An inverse relationship was reported by in studies using the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database for BMI and AD biomarkers, including cortical amyloid as measured by Pittsburgh compound B positron emission tomography (PiB-PET) and cerebrospinal fluid (CSF) amyloid-beta and tau markers [16–18]. Apolipoprotein ɛ4 (APOE4) may also be associated with weight loss in AD [19, 20].
Author Manuscript
In recent years, the concept of preclinical AD has been gaining traction leading to the first proposed research criteria for preclinical AD [21]. Based on these criteria, elderly individuals who are cognitively normal and have elevated amyloid as visualized by amyloid PET or measured in the CSF are hypothesized to be in the preclinical stage of AD. Using that definition, multiple recent studies have demonstrated greater rate of cognitive decline over time in cognitively normal individuals with elevated amyloid at baseline, especially in the presence of an APOE4 allele or proxies of neurodegeneration, when compared to individuals without elevated amyloid [22–25]. In the current study, we explored the relationship between BMI and cortical amyloid burden visualized by PiB-PET in clinically normal (CN) elderly participating in the Harvard Aging J Alzheimers Dis. Author manuscript; available in PMC 2016 August 06.
Hsu et al.
Page 3
Author Manuscript
Brain Study (HABS), which leverages multi-modality imaging biomarkers and extensive clinical assessments in older adults at risk for preclinical AD. We aimed to investigate whether the obesity paradox is already evident in the preclinical phase of AD. Specifically, we hypothesized that in CN elderly, lower BMI will be associated with greater cortical amyloid burden, supporting the late-life component of the obesity paradox described above; this association will also be evident in the group of individuals with elevated amyloid burden above a specific threshold, consistent with the presence of preclinical AD; and this association will be modified by the presence of APOE4, such that APOE4 carriers with lower BMI will have greater amyloid burden.
Methods Participants
Author Manuscript
Baseline data from 280 participants in HABS, a longitudinal observational study of normal cognitive aging and preclinical AD at the Massachusetts General Hospital (MGH) were included in this study. Inclusion criteria for HABS were as follows: English-speaking, community dwelling, age 62–90, good general medical health as determined by a study physician after extensive review of medical history, and normal cognition based on a Clinical Dementia Rating (CDR) [26] global score of 0, education-adjusted Mini-Mental State Exam (MMSE) [27] score of 27–30, and normal education-adjusted memory performance on the Logical Memory delayed paragraph recall of the Wechsler Memory Scale-Revised [28]. All participants underwent informed consent approved by the Partners Human Research Committee prior to undergoing study procedures.
Author Manuscript
The baseline visit involved a comprehensive review of participants’ medical, neurological, and psychiatric health to confirm their status as clinically normal. A survey of the medical history, a general physical and neurological exam performed by a study physician, vital signs, laboratory tests, and neuroimaging studies were completed for each participant. Exclusion criteria included serious and unstable medical, neurological, or psychiatric illness, substance use, history of traumatic brain injury, extensive cerebrovascular disease visualized on magnetic resonance imaging (MRI), and a Modified Hachinski Ischemic Scale score of ≥ 4. However, participants with stable, treated vascular risk factors, such as hyperlipidemia, hypertension, and diabetes, or cardiovascular disease were not excluded. Body-Mass Index data
Author Manuscript
Each participant had their height measured in inches and weight measured in pounds. These values were then converted to metric units in order to calculate BMI, which is kilograms divided by meters squared. The study used the standard formula for calculating BMI as outlined by the National Institute of Health (NIH) [29]. BMI subdivisions for secondary analysis were also taken from NIH guidelines for application to participants’ baseline data. They were characterized as follows: underweight (BMI < 18.5), normal (BMI 18.5–24.9), overweight (25–29.9), and obese (BMI > 30). Analyses using weight only and BMI-squared were also performed, and results were nearly identical to those using BMI (data not shown).
J Alzheimers Dis. Author manuscript; available in PMC 2016 August 06.
Hsu et al.
Page 4
APOE data
Author Manuscript
APOE4 genotyping was performed for each participant, and APOE4 carrier status was determined by having at least one APOE4 allele. PET data
Author Manuscript
PiB has been used in the past decade to quantify fibrillar cortical amyloid burden. Participants underwent PiB-PET imaging at baseline at the MGH PET facility. C11-PiB synthesis and imaging, using a Siemens/CTI ECAT EXACT HR+ PET scanner (3D mode; 63 image planes; 15.2 cm axial field of view; 5.6 mm transaxial resolution and 2.4 mm slice interval; 39 frames: 8 × 15 s,4 × 60 s,27 × 120 s), were performed as previously reported [30,[31,[32,[33]. Briefly, 8.5 to 15 mCi of C11-PiB was injected intravenously as a bolus with dynamic acquisition occurring for 60 min immediately after. PET data were reconstructed and attenuation corrected. PiB distribution volume ratio (DVR) using the Logan graphical analysis technique (cerebellar grey reference) was calculated for an aggregate of cortical regions that typically have elevated PiB retention in AD dementia, including frontal, lateral parietal and temporal, and retrosplenial cortices. PiB retention was used as a continuous variable and was positively skewed. Therefore, higher PiB values may have given more weight to the analyses proposed. Log transformation improved the skewness, but results were essentially the same (data not shown). Statistical analyses
Author Manuscript
All statistical analyses were performed using IBM SPSS version 22. We employed general linear regression models incorporating BMI and PiB retention. The primary model evaluated the association between BMI (predictor of interest) and PiB retention (dependent variable) after adjusting for covariates. PiB was chosen as the dependent variable for ease of interpretation of BMI among standard controls for PiB. Three secondary models were evaluated: 1) the association of continuous measures of BMI and PiB within BMI categories (normal, overweight, obese), 2) the association of continuous BMI and PiB within groups defined by APOE4 carrier status, and 3) the association of continuous BMI and PiB including an interaction term for BMI and APOE4 status. Demographic covariates included age, sex, and years of education. All analyses were also adjusted for APOE4 carrier status, except for secondary model 2 in which groups were divided based on APOE4 carrier status.
Results Author Manuscript
Baseline characteristics All 280 subjects were clinically normal at baseline by CDR, MMSE, and logical memory score (see Table 1). They were on average college-educated, 59% female, and 73.7 ± 6.2 years old. Twenty-nine percent were APOE4 carriers. Subjects were roughly average for the American population in terms of prevalence for hypertension (59%) and statin use (43%), but had less diabetes mellitus type 2 (9%) and active smoking (4%). BMI was normally
J Alzheimers Dis. Author manuscript; available in PMC 2016 August 06.
Hsu et al.
Page 5
Author Manuscript
distributed, with a mean of 26.9 (range 16.3–40.7). Mean cortical PiB retention was 1.17 (range 0.95–1.82). BMI subdivisions Two-thirds of the sample was overweight or obese (Table 1). Only 2.5% (n=7) were underweight. These underweight participants were excluded from further BMI subdivision analyses given lack of power. The BMI distribution revealed 31.8% were normal weight, 41.4% were overweight, and 24.3% were obese. APOE4 carrier status was nearly identical in all BMI subdivisions, about 30%. There was no significant difference in other baseline characteristics between BMI subdivisions except for age (overweight participants were older than normal participants who were older than obese participants, p=0.001). Primary analysis
Author Manuscript
Lower BMI scores were associated with greater PiB retention (β=−0.14, p=0.02, model R2=0.19) in the linear regression model adjusted for age, sex, years of education, and APOE4 carrier status (see Figure 1). APOE4 carrier status was also significant in the model (β=0.38, p
J Alzheimers Dis. Author manuscript; available in PMC 2016 August 06. Published in final edited form as: J Alzheimers Dis. 2016 June 18; 53(3): 1097–1105. doi:10.3233/JAD-150987.
Lower late-life body-mass index is associated with higher cortical amyloid burden in clinically normal elderly David C. Hsu, MDa,d,f, Elizabeth C. Mormino, PhDb, Aaron P. Schultz, PhDb, Rebecca E. Amariglio, PhDb,d, Nancy J. Donovan, MDa,d,e, Dorene M. Rentz, PsyDb,d,e, Keith A. Johnson, MDb,c,d, Reisa A. Sperling, MDb,d, and Gad A. Marshall, MDb,d,* for the Harvard Aging Brain Study
Author Manuscript
aDepartment
of Psychiatry, Massachusetts General Hospital, Harvard Medical School, Boston, MA 02114, USA
bDepartment
of Neurology, Massachusetts General Hospital, Harvard Medical School, Boston, MA 02114, USA
cDepartment
of Radiology, Massachusetts General Hospital, Harvard Medical School, Boston, MA
02114, USA dCenter
for Alzheimer Research and Treatment, Department of Neurology, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA 02115, USA
eDepartment
of Psychiatry, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA
02115, USA
Author Manuscript
fDepartment
of Psychiatry, Mercy Medical Group, Sacramento, CA 95815
Abstract Background—Lower body-mass index (BMI) in late life has been associated with an increased risk of dementia, and weight loss has been associated with more rapid decline in Alzheimer disease (AD) dementia. Objective—To explore the association between BMI and cortical amyloid burden in clinically normal (CN) elderly at risk for AD dementia.
Author Manuscript
Methods—Cross-sectional analyses were completed using baseline data from the Harvard Aging Brain Study, consisting of 280 community-dwelling CN older adults aged 62–90. Assessments included medical histories and physical exam, Pittsburgh compound B (PiB) positron emission tomography (PET) amyloid imaging, and apolipoprotein E ɛ4 (APOE4) genotyping. For the primary analysis, a general linear regression model was used to evaluate the association of BMI with PiB retention. Covariates included age, sex, years of education, and APOE4 carrier status. Secondary analyses were performed for BMI subdivisions (normal, overweight, obese), APOE4 carriers, and BMI × APOE4 interaction.
*
Correspondence to: Gad A. Marshall, MD, Center for Alzheimer Research and Treatment, Brigham and Women’s Hospital, 221 Longwood Avenue, BL-104H, Boston, MA 02115, P: 617-732-8085, F: 617-264-5212, [email protected]. Results of this manuscript were presented at the 9th Human Amyloid Imaging meeting on January 15, 2015. The authors have no relevant conflicts of interest to disclose.
Hsu et al.
Page 2
Author Manuscript
Results—In the primary analysis, greater PiB retention was associated with lower BMI (β= −0.14, p=0.02). In the secondary analyses, APOE4 carrier status (β=−0.27, p=0.02) and normal BMI (β=−0.25, p=0.01), as opposed to overweight or obese BMI, were associated with greater PiB retention. The BMI × APOE4 interaction was also significant (β=−0.14, p=0.04). Conclusions—This finding offers new insight into the role of BMI at the preclinical stage of AD, wherein lower BMI late in life is associated with greater cortical amyloid burden. Future studies are needed to elucidate the mechanism behind this association, especially in those with lower BMI who are APOE4 carriers. Keywords Alzheimer’s disease; amyloid; apolipoprotein E; body-mass index; clinically normal elderly; Pittsburgh compound B; positron emission tomography
Author Manuscript
Introduction An obesity paradox [1] exists in Alzheimer disease (AD) and dementia. Midlife obesity can increase the risk of late-life dementia [2, 3], but lower late-life body-mass index (BMI) is also related to worse AD outcomes. Higher late-life BMI is associated with decreased risk of dementia [4, 5], better cognition [6], slower progression of AD [7], and decreased mortality [8]. Weight loss over the clinical course of AD has been shown in several large cohort studies [9–13]. Early reports focused on those only in the dementia phase of AD, but later studies have explored earlier stages of the disease [7]. Patients with AD enter the dementia phase at a lower weight, and this weight loss increases mortality [13].
Author Manuscript
A 2007 study has suggested that weight loss for all causes of dementia may start 10–20 years prior to diagnosis [11]. Autopsy studies have reported that AD pathology has been associated with low body weight [14, 15]. An inverse relationship was reported by in studies using the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database for BMI and AD biomarkers, including cortical amyloid as measured by Pittsburgh compound B positron emission tomography (PiB-PET) and cerebrospinal fluid (CSF) amyloid-beta and tau markers [16–18]. Apolipoprotein ɛ4 (APOE4) may also be associated with weight loss in AD [19, 20].
Author Manuscript
In recent years, the concept of preclinical AD has been gaining traction leading to the first proposed research criteria for preclinical AD [21]. Based on these criteria, elderly individuals who are cognitively normal and have elevated amyloid as visualized by amyloid PET or measured in the CSF are hypothesized to be in the preclinical stage of AD. Using that definition, multiple recent studies have demonstrated greater rate of cognitive decline over time in cognitively normal individuals with elevated amyloid at baseline, especially in the presence of an APOE4 allele or proxies of neurodegeneration, when compared to individuals without elevated amyloid [22–25]. In the current study, we explored the relationship between BMI and cortical amyloid burden visualized by PiB-PET in clinically normal (CN) elderly participating in the Harvard Aging J Alzheimers Dis. Author manuscript; available in PMC 2016 August 06.
Hsu et al.
Page 3
Author Manuscript
Brain Study (HABS), which leverages multi-modality imaging biomarkers and extensive clinical assessments in older adults at risk for preclinical AD. We aimed to investigate whether the obesity paradox is already evident in the preclinical phase of AD. Specifically, we hypothesized that in CN elderly, lower BMI will be associated with greater cortical amyloid burden, supporting the late-life component of the obesity paradox described above; this association will also be evident in the group of individuals with elevated amyloid burden above a specific threshold, consistent with the presence of preclinical AD; and this association will be modified by the presence of APOE4, such that APOE4 carriers with lower BMI will have greater amyloid burden.
Methods Participants
Author Manuscript
Baseline data from 280 participants in HABS, a longitudinal observational study of normal cognitive aging and preclinical AD at the Massachusetts General Hospital (MGH) were included in this study. Inclusion criteria for HABS were as follows: English-speaking, community dwelling, age 62–90, good general medical health as determined by a study physician after extensive review of medical history, and normal cognition based on a Clinical Dementia Rating (CDR) [26] global score of 0, education-adjusted Mini-Mental State Exam (MMSE) [27] score of 27–30, and normal education-adjusted memory performance on the Logical Memory delayed paragraph recall of the Wechsler Memory Scale-Revised [28]. All participants underwent informed consent approved by the Partners Human Research Committee prior to undergoing study procedures.
Author Manuscript
The baseline visit involved a comprehensive review of participants’ medical, neurological, and psychiatric health to confirm their status as clinically normal. A survey of the medical history, a general physical and neurological exam performed by a study physician, vital signs, laboratory tests, and neuroimaging studies were completed for each participant. Exclusion criteria included serious and unstable medical, neurological, or psychiatric illness, substance use, history of traumatic brain injury, extensive cerebrovascular disease visualized on magnetic resonance imaging (MRI), and a Modified Hachinski Ischemic Scale score of ≥ 4. However, participants with stable, treated vascular risk factors, such as hyperlipidemia, hypertension, and diabetes, or cardiovascular disease were not excluded. Body-Mass Index data
Author Manuscript
Each participant had their height measured in inches and weight measured in pounds. These values were then converted to metric units in order to calculate BMI, which is kilograms divided by meters squared. The study used the standard formula for calculating BMI as outlined by the National Institute of Health (NIH) [29]. BMI subdivisions for secondary analysis were also taken from NIH guidelines for application to participants’ baseline data. They were characterized as follows: underweight (BMI < 18.5), normal (BMI 18.5–24.9), overweight (25–29.9), and obese (BMI > 30). Analyses using weight only and BMI-squared were also performed, and results were nearly identical to those using BMI (data not shown).
J Alzheimers Dis. Author manuscript; available in PMC 2016 August 06.
Hsu et al.
Page 4
APOE data
Author Manuscript
APOE4 genotyping was performed for each participant, and APOE4 carrier status was determined by having at least one APOE4 allele. PET data
Author Manuscript
PiB has been used in the past decade to quantify fibrillar cortical amyloid burden. Participants underwent PiB-PET imaging at baseline at the MGH PET facility. C11-PiB synthesis and imaging, using a Siemens/CTI ECAT EXACT HR+ PET scanner (3D mode; 63 image planes; 15.2 cm axial field of view; 5.6 mm transaxial resolution and 2.4 mm slice interval; 39 frames: 8 × 15 s,4 × 60 s,27 × 120 s), were performed as previously reported [30,[31,[32,[33]. Briefly, 8.5 to 15 mCi of C11-PiB was injected intravenously as a bolus with dynamic acquisition occurring for 60 min immediately after. PET data were reconstructed and attenuation corrected. PiB distribution volume ratio (DVR) using the Logan graphical analysis technique (cerebellar grey reference) was calculated for an aggregate of cortical regions that typically have elevated PiB retention in AD dementia, including frontal, lateral parietal and temporal, and retrosplenial cortices. PiB retention was used as a continuous variable and was positively skewed. Therefore, higher PiB values may have given more weight to the analyses proposed. Log transformation improved the skewness, but results were essentially the same (data not shown). Statistical analyses
Author Manuscript
All statistical analyses were performed using IBM SPSS version 22. We employed general linear regression models incorporating BMI and PiB retention. The primary model evaluated the association between BMI (predictor of interest) and PiB retention (dependent variable) after adjusting for covariates. PiB was chosen as the dependent variable for ease of interpretation of BMI among standard controls for PiB. Three secondary models were evaluated: 1) the association of continuous measures of BMI and PiB within BMI categories (normal, overweight, obese), 2) the association of continuous BMI and PiB within groups defined by APOE4 carrier status, and 3) the association of continuous BMI and PiB including an interaction term for BMI and APOE4 status. Demographic covariates included age, sex, and years of education. All analyses were also adjusted for APOE4 carrier status, except for secondary model 2 in which groups were divided based on APOE4 carrier status.
Results Author Manuscript
Baseline characteristics All 280 subjects were clinically normal at baseline by CDR, MMSE, and logical memory score (see Table 1). They were on average college-educated, 59% female, and 73.7 ± 6.2 years old. Twenty-nine percent were APOE4 carriers. Subjects were roughly average for the American population in terms of prevalence for hypertension (59%) and statin use (43%), but had less diabetes mellitus type 2 (9%) and active smoking (4%). BMI was normally
J Alzheimers Dis. Author manuscript; available in PMC 2016 August 06.
Hsu et al.
Page 5
Author Manuscript
distributed, with a mean of 26.9 (range 16.3–40.7). Mean cortical PiB retention was 1.17 (range 0.95–1.82). BMI subdivisions Two-thirds of the sample was overweight or obese (Table 1). Only 2.5% (n=7) were underweight. These underweight participants were excluded from further BMI subdivision analyses given lack of power. The BMI distribution revealed 31.8% were normal weight, 41.4% were overweight, and 24.3% were obese. APOE4 carrier status was nearly identical in all BMI subdivisions, about 30%. There was no significant difference in other baseline characteristics between BMI subdivisions except for age (overweight participants were older than normal participants who were older than obese participants, p=0.001). Primary analysis
Author Manuscript
Lower BMI scores were associated with greater PiB retention (β=−0.14, p=0.02, model R2=0.19) in the linear regression model adjusted for age, sex, years of education, and APOE4 carrier status (see Figure 1). APOE4 carrier status was also significant in the model (β=0.38, p
