What this is
- This research examines the relationship between () changes and Alzheimer's disease (AD) pathology in adults with Down syndrome (DS).
- The study analyzes data from 467 adults with DS, focusing on changes in , (AĪ²) levels, and cognitive decline over time.
- Findings suggest that declines in begin in the early 40s and are associated with increased AĪ² levels and cognitive impairment.
Essence
- declines in adults with Down syndrome starting in their early 40s, correlating with accumulation and cognitive decline.
Key takeaways
- decline begins around age 43 in adults with Down syndrome and is linked to cognitive decline. By age 53, an adult with DS could experience a 1.5 kg/m decrease in .
- Higher baseline PET AĪ² levels correlate with greater reductions over time. Specifically, a PET AĪ² value of 18 Centiloids is associated with a 0.6 kg/m decrease in .
- Cognitive decline is associated with decline; a 5% decrease in the modified Cued Recall Test score corresponds to a 0.1 kg/m decrease in .
Caveats
- The study relies on , which does not specify the nature of weight loss, limiting understanding of underlying mechanisms. Future research should incorporate body composition analysis.
- The sample lacks diversity, with 92% identifying as White, non-Hispanic, which may affect the generalizability of the findings.
Definitions
- body mass index (BMI): A measure calculated from height and weight, used to categorize individuals as underweight, normal weight, overweight, or obese.
- amyloid beta (AĪ²): A protein that accumulates in the brain and is associated with Alzheimer's disease pathology.
AI simplified
BACKGROUND
Weight loss is linked to preclinical stage of Alzheimer's disease (AD), when early pathological changes are present but cognition remains intact, in the general population.1, 2 In studies of lateāonset AD (LOAD), weight loss is documented in the years immediately preceding AD dementia onset.3, 4, 5 For example, Buchman and others3 found that a body mass index (BMI) decrease of 1 kg/m2 per year (vs no change) was associated with a 35% greater risk of developing AD within 5 years. Weight loss is associated with elevated cerebrospinal fluid (CSF) and positron emission tomography (PET) biomarkers of amyloid beta (AĪ²) plaques and neurofibrillary tau tangles,6, 7 two hallmark pathologic features of AD.8, 9 However, the biological mechanisms linking weight loss and AD remain unclear. One possibility is that AĪ² accumulation and/or downstream neurodegeneration may disrupt hormone regulation, including hormones like leptin10 and adiponectin.11 Indeed, there are altered plasma and CSF adiponectin levels in individuals with mild cognitive impairment (MCI) and AD, suggesting metabolic dysfunction.11 Neurodegeneration in brain regions related to memory and sensory processing may also impair weight regulation and appetite control.12, 13 As such, unintentional weight loss can serve as a meaningful clinical sign of impending AD dementia in the general population. It is unclear, however, how weight loss may be related to the development of AD in adults with Down syndrome (DS), a group genetically atārisk for AD (referred to as Down Syndrome assocaited Alzheimer's disease [DSAD]).
Adults with DS, or trisomy 21, have a 90% lifetime risk for developing DSAD due to the triplication of the amyloid precursor protein (APP) gene on chromosome 21.14 Having three copies of the APP gene results in a 1.5Ć greater production of AĪ² across the lifespan.15, 16 In DS, brain AĪ² deposition is evident in the 30s17, 18 and is followed by intracellular neurofibrillary tau tangles in the 40s and 50s.14, 19, 20 The median age of symptomatic DSAD is 53 years.21 It is estimated that 60 to 90% of adults with DS are overweight or obese.22, 23, 24 Furthermore, BMI is at its highest during the third decade of life.25, 26 In a crossāsectional study, Agiovlasitis and others25 found that the BMI of adults with DS is likely to increase through the 30s, with significant decrease in BMI thereafter. Only a handful of studies examined the association between weight loss and DSAD.27, 28, 29 Bayen and others27 reported that adults with DS and AD were twice as likely to lose weight as adults with DS who were cognitively stable. In a large crossāsectional study of adults with DS, using ageātrajectory estimates, BMI decreased by ā0.23 kg/m2 per year starting in the late 30s, which aligns with AĪ² deposition.28
The current study builds on these prior studies to longitudinally assess the association between baseline PET AĪ² and tau burden, change in memory and dementia symptoms (e.g., mental status), and change in BMI across time. Analyses drew on the longitudinal cohort study of adults with DS enrolled in the Alzheimer's Biomarkers ConsortiumāDown Syndrome (ABCāDS30). Analyses leveraged up to four time points (followāup to ā7 years) of data collection. The study aims included: (1) evaluate change in BMI across time and by age in adults with DS; (2) determine the effect of baseline PET AĪ² and tau on BMI change; and (3) evaluate the association between change in BMI and change in memory and dementia symptoms. Based on prior crossāsectional research,25, 28 BMI was hypothesized to increase or remain stable through the late 30s and then decline after age 40 years. Based on research on LOAD,6, 7 higher baseline PET AĪ² and tau were expected to be associated with decreases in BMI overtime. Declines in BMI were expected to be associated with declines in memory and increases in dementia symptoms across time. Moreover, the above associations between BMI decline and AD pathology and cognitive decline were expected to be observed even in adults with DS who did not have clinical dementia, and thus occur early on in the unfolding of DSAD. Understanding of the connection between weight loss and DSAD has important implications for AD screening and for managing the clinical expression of DSAD.
METHODS
Participants
Analyses included 467 adults with DS enrolled in one of nine ABCāDS, a large multiāsite longitudinal study focused on identifying early biomarkers of AD in DS.30 Inclusion criteria at baseline included age ā„25 years, āmental ageā ā„30 months, and chromosomal analysis confirming trisomy 21; exclusion criteria included contraindications for brain imaging or the presence of any untreated medical or mental health conditions that could impact cognitive functioning.
Procedures
Participants completed a multiāday study visit that involved a battery of directly administered cognitive measures, blood draw, neurophysical exam that included assessment of height and weight, and magnetic resonance imaging (MRI) and PET scans.Ā A study partner attended the visit and reported on the participant's sociodemographics, medical history, activities of daily living, and screens for symptoms of cognitive decline or dementia. Visits were repeated every 16 months (Ā±3 months). Analyses include up to four data collection cycles.
Measures
Sociodemographics
The participant's age was reported by the study partner at baseline. The study partner also reported biological sex at birth (male = 1; female = 2) and race/ethnicity (1 = White, 2 = Black; 3 = American Indian; 4 = Asian; 5 = Hispanic). Premorbid intellectual disability (ID) level was based on standardized intelligence quotient (IQ) scores recorded in medical records or obtained in ABCāDS using the StanfordāBinet Intelligence Scales, Fifth Edition31 abbreviated IQ and/or the Kaufman Brief Intelligence Test, Second Edition32 prior to any concerns about cognitive decline. Premorbid ID was coded as: mild (āmental ageā: 9 to 14 years; coded = 1), moderate (āmentalā age: 4 to 8 years; coded = 2), or severe/profound (āmental ageā: <4 years; coded = 3). Genotyping was conducted to determine the presence of apolipoprotein E (APOE) Īµ4 carrier status. Time between data collection cycles in years from baseline was calculated.
BMI
At baseline, height was measured using a tape measure or stadiometer, whereas at each study visit, weight was measured using either a digital or mechanical scale. Both were assessed with shoes off for better accuracy of height and weight and using the same equipment at each visit. BMI was calculated as weight in kilograms divided by height in meters squared. Three categories were defined: Normal + Underweight (<25 kg/m2); Overweight (25 to <30 kg/m2); and Obese (ā„30 kg/m2).33
Clinical AD status
Clinical AD dementia status was based on a case review consensus process that involved expert DS clinicians, study coordinators, and highly trained and experienced staff who were familiar with each participant. Consensus teams were blind to imaging, āomics, and genetic data. All available informantāreported and direct measures of cognition, adaptive functioning, and behavior, along with medical histories, premorbid ID, neurophysical exam findings, and clinical lab results (i.e., complete blood count and metabolic comprehensive panel.) were reviewed. Using this information, participants were given a status of cognitively stable (defined as no evidence of cognitive or functional decline beyond normal aging; codedĀ =Ā 0), MCI (defined as subtle and/or limited decline in cognition and/or adaptive behavior; codedĀ =Ā 1), AD dementia (defined as significant and sustained declines in cognition or functioning; codedĀ =Ā 2), or unable to determine (meaning that changes in cognition and/or functioning were observed but could have been caused by significant life events or medical history changesĀ =Ā 3).
Cognitive functioning
The modified Cued Recall Test (mCRT)34 was used to assess memory. It involves requesting that participants learn and then freely recall a list of 12 pictures. Following three free recall trials, there were cued trials in which a category prompt (e.g., āpiece of fruitā for pictures of grapes) was provided verbally. The mCRT Total score is the sum of free and cued recall across three trials, with a higher mCRT Total score indicating better memory performance. The mCRT Total score has been shown to differentiate adults with DS who are cognitively stable from those with MCI or dementia,35 and is associated with early AĪ² accumulation in DSAD.36, 37
The Down Syndrome Mental Status Examination (DSMSE)38 assesses mental status with items involving recall of personal information, orientation to the day and season, immediate and delayed recall, language, and visuospatial function and praxis. The DSMSE can also distinguish adults with DS with MCI or dementia from those who are cognitively stable.37
Finally, the National Task GroupāEarly Detection Screen for Dementia (NTGāEDSD)39 is an established informantāreported measure of symptoms of dementia. The 6ādomain total score ranges from 0 to 51, with higher scores indicating more dementia symptoms and is sensitive to MCI and dementia in DS.40
AD pathology
MR scans were acquired on a 3T GE Discovery MR750, Siemens Trio, Siemens Prisma, or GE Signa PET/MR depending on the imaging site in a subsample of participants (n = 245). AĪ² was assessed using a tracer: [11C]āPittsburgh compound B ([11C] PiB) or florbetapir (AVā45), whereas tau was assessed using the tracer 18Fāflortaucipir ([18F] AVā1451). Images were taken 50ā70 min postāinjection for [11C] PiB and 80ā100 min for [18F] AVā1451. Data were reconstructed using iterative methods and corrected for movement, deadtime, and radioactive decay. Images were captured in 5āmin frames and corrected for motion on a frameābyāframe basis. Highāresolution T1āweighted images were acquired using a threeādimensional (3D) fast spoiled gradient echo (FSPGR) or magnetization prepared rapid acquisition gradient echo (MPRAGE) sequence and used for anatomic reference in PET processing. Through the use of FreeSurfer 5.3, T1āweighted images were segmented into regions defined by DesikanāKilliany atlas regions.41 Results were inspected and 12 highāquality parcellations were selected as templates. These templates were warped into each participant's native MR space using the Advanced Neuroimaging Tools (ANTs) software package, with a final native space atlas being created by determining the maximum overlap of each parceled region from the 12 templates. Results were accepted or rejected on a visual rating of the final atlasā adherence to the participant's MR anatomy.
The Centiloid method42 was used to calculate amyloid burden from amyloid PET scans. An advantage to the method is that it provides a standardized scale for the two tracers. Briefly, PET scans were registered to the corresponding T1 MRI. The MRI was then warped to the Montreal Neurological Institute (MNI152) template using Statistical Parametric Mapping, version 8, (SPM8). The wrap parameters were then used to cowarp the MRI and PET images. Radioactivity concentration was extracted for the Centiloid standard global region and whole cerebellum using regions of interest (ROIs) predefined on the MNI152 template.43 Global SUVr was taken to be the ratio of tracer concentration in the global region to that of the cerebellum. Using linear+constant transformations specified separately for [11C]PiB37 and [18F] AVā45,43 tissue ratios were converted to centiloid values.
For tau, scans were registered to T1 MRI using PMOD software. Through the use of FreeSurfer 5.3, images were segmented into regions defined by DesikanāKilliany atlas regions.41 Reported SUVrs were calculated by normalizing concentrations to the cerebellar cortex concentration.
Data analysis
Variable distributions and potential outliers were examined using descriptive statistics, box plots, and histograms. Analysis of variance (ANOVA) and chiāsquare tests were run to assess differences in sociodemographic data by BMI categories. Mixed linear models were conducted using the lmer function from the lme4 R package.44 Time, age, and their interaction, as well as biological sex, premorbid ID, and APOE Īµ4 status, were included as fixed effects in the models with random intercepts varying among participants and the data collection site. Models were originally conducted including the full sample but were reārun to include only participants deemed to be cognitively stable. The significance level was set to p ā¤ 0.05 for the data analysis.
RESULTS
Preliminary analyses
At baseline, average participant age was 43.67 years (SD = 10.06), ranging from 25 to 81 and having a normal distribution (skewness: 0.14). The average time between study visits was 15.12 (SD = 8.04) months. The average baseline BMI of the participants was 31.45 (SD = 7.34), with 17.13% of participants being underweight or having normal BMI (BMI <25 kg/m2), 30.84% being overweight (BMI 25 to <30 kg/m2), and 52.04% being obese (BMI ā„30 kg/m2). About half of participants were female (46.25%), with a majority were of White, nonāHispanic (91.01%) race/ethnicity, and 23.98% were APOE Īµ4 carriers. The majority of participants had mild (45.61%) or moderate premorbid ID levels (45.82%), but a subset (8.14%) had severe/profound ID. The majority of participants (73.23%) were cognitively stable at baseline, whereas 57 (12.21%) were deemed to have MCI and 49 (10.49%) had a clinical AD status of dementia. Table 1 provides information about participant characteristics and provides the means and SDs for the main study variables.
Analyses included participants with at least one time point (n = 467) of BMI data. The majority of participants provided multiple timepoints: two (n = 177, 37.90%), three (n = 144, 30.84%), and four (n = 128, 27.41%). An ANOVA and followāup Tukey tests indicated that participants with only one time point of BMI data were younger than those with multiple time points of data (F(4, 464) = 18.789, p < 0.001). There were no statistically significant differences between participants with one versus multiple time points of data in BMI (F(4, 464) = 1.347, p = 0.251), biological sex (Ļ2[4,465] = 2.007, p = 0.735), race Ļ2[4,465] = 4.922, p = 1.00), or premorbid ID (Ļ2[4,465] = 4.080, p = 0.982).
Of the 467 participants, PET AĪ² was available for 245 (52.46%) and tau PET was available for 134 (28.69%). Neuroimaging data were either not collected or not yet processed in a harmonized way for the remaining participants. Analyses compared the sociodemographics of participants with versus without neuroimaging data. Compared with those for whom neuroimaging data was not available, the participants with neuroimaging data were younger (t(435) = 6.792, p < 0.001) but did not differ in biological sex (Ļ2[1,465] = 1.024, p = 0.311), APOE Īµ4 status (Ļ2[1,465] = 2.06, p = 0.357), BMI (Ļ2[1,465] = ā2.04, p = 0.061), race (Ļ2[1,465] = 2.018, p = 0.569), or premorbid ID level (Ļ2[1,465] = 1.394, p = 0.498). The participants with imaging data also scored higher on the mCRT (t(422) = ā2.173, p < 0.001) and DSMSE (t(460) = ā3.730, p < 0.001), but did not differ on the NTG (t(466) = 1.385, p = 0.167). This pattern likely reflects recruitment practices, as sites focused on imaging recruited through legacy studies that involved younger participants than the sites not focused on imaging. However, it is also likely that younger adults with DS and those with fewer coāoccurring medical conditions may be more willing or able to undergo neuroimaging protocols.
The results of an ANOVA and followāup Tukey tests indicated there was a statistically significant difference in age among BMI categories (underweight/normal weight, overweight, and obese) (F(2, 464) = 9.615, p < 0.001). Participants who were obese or overweight were younger than those with BMI in the normal or underweight categories (Table 1). Pearson chiāsquare statistics indicated that female participants were more likely to be in the overweight and obese group compared to male participants (Ļ2[2,467] = 6.75, p = 0.034). Participants who were cognitively stable were more likely to be in the normal or overweight groups than were participants with MCI or dementia (Ļ2[6,467] = 26.92, p < 0.001). There was no statistically significant differences between BMI category by race (Ļ2[10,467] = 10.65, p = 0.385), premorbid ID (Ļ2[4,465] = 7.48, p = 0.113), or APOE Īµ4 status (Ļ2[4,467] = 3.01, p = 0.556).
| Overall= 467N | Normal + Underweight= 80N | Overweight= 144N | Obese= 243N | āvaluep | |
|---|---|---|---|---|---|
| Age, MĀ Ā±Ā SD | 43.67Ā Ā± 10.06 | 47.5Ā Ā± 10.5 | 44.3Ā Ā± 10.5 | 42.1Ā Ā± 9.2 | < 0.001 |
| BMI, MĀ Ā±Ā SD | 31.46Ā Ā± 7.34 | 22.6Ā Ā± 1.9 | 27.4Ā Ā± 1.4 | 36.8Ā Ā± 6.1 | |
| Biological sex, no. (%) | 0.034 | ||||
| Female | 216 (46.25%) | 30 (37.50%) | 60 (41.67%) | 126 (51.85%) | |
| Male | 251 (54.75%) | 50 (62.50%) | 84 (58.33%) | 117 (48.15%) | |
| Race/ethnicity, no. (%) | 0.385 | ||||
| White, nonāHispanic | 425 (91.01%) | 71 (88.75%) | 127 (88.19%) | 227 (93.42%) | |
| Black | 7 (1.50%) | 2 (2.5%) | 2 (1.39%) | 3 (1.24%) | |
| American Indian | 1 (0.21%) | 0 (0.0%) | 0 (0%) | 1 (0.41%) | |
| Asian | 6 (1.28%) | 2 (2.5%) | 4 (2.78%) | 0 (0.0%) | |
| Hispanic | 23 (4.93%) | 4 (5.0%) | 10 (6.94%) | 9 (3.70%) | |
| Diagnosis, no. (%) | < 0.001 | ||||
| No MCI or dementia | 342 (73.23%) | 44 (55.0%) | 105 (72.92%) | 193 (79.42%) | |
| MCI | 57 (12.21%) | 20 (25.0%) | 22 (15.28%) | 15 (6.17%) | |
| AD dementia | 49 (10.49%) | 12 (15.0%) | 11 (7.64%) | 26 (10.70%) | |
| Unable to determine | 19 (4.07%) | 4 (5.0%) | 6 (4.17%) | 9 (3.70%) | |
| Premorbid ID level, no. (%) | 0.113 | ||||
| Mild | 213 (45.61%) | 33 (41.25%) | 59 (40.97%) | 121 (49.79%) | |
| Moderate | 214 (45.82%) | 37 (46.25%) | 76 (52.78%) | 101 (41.56%) | |
| Severe/profound | 38 (8.14%) | 10 (12.50%) | 8 (5.56%) | 20 (8.23%) | |
| Unknown | 2 (0.43%) | 0 (0.0%) | 1 (0.69%) | 1 (0.41%) | |
| Īµ4, no. (%)APOE | 0.556 | ||||
| Present | 112 (23.98%) | 23 (28.75%) | 35 (24.31%) | 54 (22.22%) | |
| Absent | 340 (72.81%) | 53 (66.25%) | 104 (72.22%) | 183 (75.31%) | |
| Unknown | 15 (3.21%) | 4 (5.0%) | 5 (3.47%) | 6 (2.47%) |
Change in BMI over time
Table 2 presents results from the linear mixed model analyzing change in BMI across time and time by age, with sociodemographic controls also entered into the model. Premorbid ID and APOE Īµ4 status were not significantly associated with BMI (Ī² = ā0.861, p = 0.092 and Ī² = 0.0311, p = 0.964). Female participants tended to have a higher BMI than male participants across time (Ī² = 2.07, p = 0.002). There was a significant positive association between time and change in BMI (Ī² = 0.532, p = 0.01), indicating an overall increase in BMI over time. Indicated across the entire sample, there was an increase in BMI over time. However, the model shows the effects of time, time by age, with sociodemographic controls effects on change in BMI across time (Ī² = ā0.011, p = 0.002) and time Ć age (Ī² = ā0.015, p = 0.002) were both significantly negative. Therefore, on average younger individuals gained weight over time, whereas older participants lost weight over time. Model estimates indicate that, on average, weight loss begins at age 43 years, and by age 53 years, an adult with DS would on average experience a 1.5 kg/m2 decrease in BMI (from BMI in their early 40s). By age 64 years, adults with DS on average experience a 3 kg/m2 decrease in BMI (from BMI prior in their early 40s). Figure 1 displays BMI change over time by age.
The model was reārun only including the 342 participants who were cognitively stable at baseline. Significant associations remained; time (Ī² = 0.468, p = 0.034) and time Ć age (Ī² = ā0.013, p = 0.012) predicted decline in BMI over time. To further to explore the impact of weight changes, the model was reārun excluding participants who experienced a BMI changes of ā„10 (n = 10). The results remained consistent: both age (Ī² = ā0.115, p = 0.001) and time Ć age (Ī² = ā0.008, p = 0.042) remained statistically significant.
Association between age and body mass index. A LOESS regression was used to model the association between age and BMI (=Ā 467), capturing the nonālinear trend across the age span. Individual lines are colorācoded to distinguish participants and aid in visualizing individual BMI trajectories across the sample. The green line is the age when AĪ² (CLĀ =Ā 18) reaches positivity (ageĀ =Ā 41 years). The first red line indicates when BMI starts to significantly decline (ageĀ =Ā 43 years), the second red line indicates a 1.5 unit decline (ageĀ =Ā 53), and the third red line indicates a 3.0 unit decline (ageĀ =Ā 64). The purple line indicates the average age at AD onset (ageĀ =Ā 53.98). AD, Alzheimer's disease; AĪ², amyloid beta; BMI, body mass index; CL, Centiloid; kg, kilograms; m, meters. n
| Estimates | |||||
|---|---|---|---|---|---|
| Fixed effects | Ī² | Ī²SE | t | āvaluep | |
| BMI | Intercept | 34.47 | 2.056 | 16.77 | <0.001 70387 |
| Time | 0.532 | 0.208 | 2.56 | 0.011 70387 | |
| Age | ā0.107 | 0.034 | ā3.11 | 0.002 70387 | |
| Biological sex | 2.071 | 0.647 | 3.2 | 0.002 70387 | |
| Premorbid intellectual disability | ā0.861 | 0.51 | ā1.689 | 0.092 | |
| Īµ4APOE | 0.031 | 0.68 | 0.046 | 0.964 | |
| Time Ć age | ā0.015 | 0.005 | ā3.201 | 0.002 70387 | |
BMI and baseline PET AĪ² and tau
Table 3 presents linear mixed models of the effect of baseline PET AĪ² (Model 1) and also baseline PET AĪ² and tau (Model 2) on BMI, with time, time Ć age, and sociodemographic controls also included. Interactions of time Ć PET AĪ² and time Ć PET tau were also included as predictors. Biological sex had a significant positive effect on BMI (Ī² = 2.787, p = 0.002); as in earlier models, female participants had a higher average BMI than male participants. There was also a significant effect of time Ć PET AĪ² on BMI (Ī² = ā0.005, p = 0.003), even when adjusting for age, such that higher baseline PET AĪ² predicted greater BMI decreases across time. This finding suggests that PET AĪ² contributes to BMI decline above and beyond ageārelated effects.
Figure 2 shows the relation between PET AĪ² and BMI. In models that included tau PET, there was no significant effect of tau or tau Ć time on change in BMI over time. In these models, a PET AĪ² value of 18 Centiloids (i.e., threshold for AĪ² positivity18) occurs at age 41 years, 2 years prior to the start of a decline in BMI (age 43 years). Following AĪ² positivity, BMI was estimated to decrease by 0.6 kg/m2 by the time an individual reaches PET AĪ² value of 60 Centiloids (which in prior studies was associated with having MCI45), and BMI is estimated to decrease by 1.2 kg/m2 by the time an individual reaches PET AĪ² value of 100 (which in prior studies was associated with having AD clinical dementia45).
The model was reārun including only the 191 participants who were cognitively stable at baseline to determine if associations remained prior to onset of clinical dementia. The negative association between time Ć PET AĪ² remained significant (Ī² = ā0.013, p = 0.012).
Association between age, amyloidāĪ², and body mass index. A LOESS regression was used to model the association between age and BMI (left yāaxis), capturing the nonālinear trend across the age span and the association between age and AĪ² (right yāaxis) (=Ā 245). AĪ² is measured in CL. AĪ², amyloid beta; BMI, body mass index; CL, Centiloid. n
| Estimates | |||||
|---|---|---|---|---|---|
| Fixed effects | Ī² | SE Ī² | t | āvaluep | |
| Model 1 | Intercept | 31.63 | 3.345 | 9.454 | < 0.001 70387 |
| Time | 0.029 | 0.074 | 0.39 | 0.697 | |
| Age | ā0.080 | 0.068 | ā1.173 | 0.242 | |
| Biological sex | 2.787 | 0.089 | 2.125 | 0.002 70387 | |
| Premorbid intellectual disability | ā0.453 | 0.68 | ā0.667 | 0.506 | |
| Īµ4APOE | 0.857 | 0.99 | 0.858 | 0.391 | |
| Amyloid beta (AĪ²) | 0.0005 | 0.017 | 0.032 | 0.974 | |
| Time Ć AĪ² | ā0.005 | 0.001 | ā3.058 | 0.003 70387 | |
| Model 2 | Intercept | 0.302 | 7.811 | 3.86 | < 0.001 70387 |
| Time | ā0.537 | 0.728 | ā0.737 | 0.463 | |
| Age | ā0.042 | ā0.120 | ā0.393 | 0.695 | |
| Biological sex | 3.884 | 1.315 | 2.954 | 0.003 70387 | |
| Premorbid intellectual disability | ā0.716 | 0.982 | ā0.729 | 0.468 | |
| Īµ4APOE | ā0.274 | 1.533 | ā0.179 | 0.858 | |
| AĪ² | 0 | 0.052 | 0.009 | 0.993 | |
| Tau | ā0.416 | 4.934 | ā0.084 | 0.933 | |
| Time Ć AĪ² | ā0.013 | 0.005 | ā2.304 | 0.023 70387 | |
| Time Ć tau | 0.056 | 0.666 | 0.001 | 0.405 | |
BMI and cognitive functioning
Table 4 presents the linear mixed model results for associations between cognitive performance and BMI across data collection cycles. In models including mCRT total score, time (Ī² = ā0.509, p = 0.004), age (Ī² = ā0.082, p = 0.034), biological sex (Ī² = 2.184, p = 0.002), and the interaction of time Ć mCRT (Ī² = 0.014, p = 0.011) were significant predictors of BMI. Across time, a positive association between mCRT and BMI emerges (i.e., a higher mCRT score is associated with higher BMI). Based on model estimates, a 5% decrease on the mCRT is associated with a 0.1 kg/m2 decrease in BMI. Figure 3 shows the association between mCRT across time and BMI. For models including DSMSE, time (Ī² = ā0.391, p = 0.03), age (Ī² = ā0.094, p = 0.010), and biological sex (Ī² = 2.008, p = 0.002) were significant predictors of BMI. Similar to the model for mCRT, time and age had a negative effect on BMI. There was also a trendālevel interaction of time Ć DSMSE scores (Ī² = 0.005, p = 0.076) on BMI; across time, a positive association between DSMSE scores and BMI was seen. Based on model estimates, a 5% decrease on the DSMSE was associated with a 0.3 kg/m2 decrease in BMI. Finally, in models including NTGāEDSD, age (Ī² = ā0.121, p = 0.001), biological sex (Ī² = 2.078, p = 0.002), and the interaction of time Ć NTGāEDSD (Ī² = ā0.011, p = 0.039) were significant predictors of BMI. Both age and the interaction between time and NTGāEDSD scores were negatively associated with BMI. This finidng means that as age and NTGāEDSD scores increase, BMI decreases. Based on model estimates, a 21% increase on the NTGāEDSD was associated with a 0.1 kg/m2 decrease in BMI.
Models were reārun to include only the 342 participants deemed to be cognitively stable at baseline. There continued to be a positive time Ć mCRT interaction on BMI change (Ī² = 0.016, p = 0.021). There also continued to be a trendālevel time Ć DSMSE effect on BMI change (Ī² = 0.006, p = 0.078). However, there was no longer a significant time Ć NTGāEDSD effect on BMI change (Ī² = ā0.011, p = 0.178), potentially reflecting that the NTGāEDSD is dependent on observed changes reported by study partners rather than direct measures of cognitive functioning.
Associations between cognitive decline and body mass index. (A) Linear regression modeling of the association between age and BMI, comparing individuals who showed decline (orange line) versus those with a stable (blue line) performance on the mCRT (=Ā 431). (B) Linear regression modeling the association between age and BMI, comparing individuals who showed decline (orange line) versus those with a stable (blue line) performance on the DSMSE (=Ā 464) (C) Linear regression modeling of the association between age and BMI, comparing individuals who showed decline (orange line) versus those with a stable (blue line) performance on the between the NTGāEDSD (=Ā 467). Visual models appear modest, but statistical models revealed significant associations between changes in BMI and both the mCRT and NTGāEDSD. BMI, body mass index; DSMSE, Down Syndrome Mental Status Examination; mCRT, modified Cued Recall Test; NTGāEDSD, National Task GroupāEarly Detection Screen for Dementia. n n n
| Estimates | |||||
|---|---|---|---|---|---|
| BMI | Fixed effects | Ī² | SE Ī² | T | āvaluep |
| mCRT | Intercept | 31.866 | 2.598 | 12.266 | < 0.001 70387 |
| Time | ā0.509 | 0.175 | ā2.906 | 0.004 70387 | |
| Age | ā0.082 | 0.038 | ā2.134 | 0.034 70387 | |
| Biological sex | 2.184 | 0.683 | 3.199 | 0.002 70387 | |
| Premorbid intellectual disability | ā0.391 | 0.556 | ā0.704 | 0.482 | |
| Īµ4APOE | 0.159 | 0.734 | 0.217 | 0.829 | |
| mCRT | 0.03 | 0.026 | 1.12 | 0.263 | |
| Time Ć mCRT | 0.014 | 0.006 | 2.545 | 0.011 70387 | |
| DSMSE | Intercept | 32.324 | 2.7 | 11.991 | < 0.001 70387 |
| Time | ā0.391 | 0.18 | ā2.167 | 0.031 70387 | |
| Age | ā0.094 | 0.036 | ā2.596 | 0.01 70387 | |
| Biological sex | 2.008 | 0.65 | 3.088 | 0.002 70387 | |
| Premorbid intellectual disability | ā0.541 | 0.546 | ā0.992 | 0.322 | |
| Īµ4APOE | 0.131 | 0.681 | 0.192 | 0.848 | |
| DSMSE | 0.019 | 0.016 | 1.175 | 0.24 | |
| Time Ć DSMSE | 0.005 | 0.003 | 1.78 | 0.076 | |
| NTGāEDSD | Intercept | 34.873 | 2.079 | 16.772 | < 0.001 70387 |
| Time | ā0.048 | 0.056 | ā0.859 | 0.391 | |
| Age | ā0.121 | 0.035 | ā3.454 | 0.001 70387 | |
| Biological sex | 2.078 | 0.65 | 3.197 | 0.002 70387 | |
| Premorbid intellectual disability | ā0.802 | 0.513 | ā1.562 | 0.119 | |
| Īµ4APOE | 0.095 | 0.685 | 0.139 | 0.89 | |
| NTGāEDSD | 0.021 | 0.02 | 1.043 | 0.295 | |
| Time Ć NTGāEDSE | ā0.011 | 0.005 | ā2.070 | 0.039 70387 | |
Baseline BMI status and cognitive decline
Followāup analyses were conducted to understand differences in BMI decline in relation to cognitive decline by baseline BMI status (normal/underweight, overweight, and obese). Compared to participants who were normal or underweight at baseline, those who were obese had a flatter decline in BMI in relationship to decline in mCRT score (Ī² = 0.144, p = 0.002). There was not a significant difference in the association between mCRT decline and BMI decline between participants who were overweight and those who were normal or underweight at baseline (Ī² = 0.047, p = 0.337). Similarly, on the DSMSE, participants who were obese at baseline showed a flatter decline in BMI in relation to decline in DSMSE score (Ī² = 0.086, p < 0.001) than those who were underweight or normal weight at baseline. There was no statistically significant difference in the association between DSMSE decline and BMI decline in participants who were overweight versus underweight or normal weight (Ī² = 0.033, p = 0.198). There was not an association between decline in NTGāEDSE and BMI decline by baseline BMI status (Obese: Ī² = ā0.011, p = 0.749; Overweight: Ī² = ā0.013, p = 0.719). Figure S1 (mCRT) and Figure S2 (DSMSE) depict significant findings.,
DISCUSSION
In the neurotypical population, unintentional weight loss occurs in the years preceding the onset of AD dementia.2, 4 However, little is known about the relationship between weight loss and the onset of AD dementia in DS. Results from the current longitudinal study built on prior crossāsectional research27, 28, 29 that suggested that the onset of decreasing BMI occurs around the time that early AD pathology (i.e., amyloid burden) begins in adults with DS. Findings from the present study show that on average, adults with DS show declines in BMI beginning in their early 40s. This finding is consistent with previous research also showing that adults with DS experience BMI declines after the third decade of life.25 Based on our model estimates, by age 53 years, an adult with DS would be expected to experience a 1.5 kg/m2 decrease in their BMI relative to their BMI at age 42 years. The effect continues with age so that, compared with their BMI at 42, at age 64 years, an adult with DS would be expected to experience a 3.0 kg/m2 decrease in BMI. It is important to note that the age of onset of initial decline in BMI (43 years) in the current study is slightly older than was reported in our prior crossāsectional study.28 This discrepancy may occur because longitudinal data provide a more precise estimate of withināperson BMI decline. Premorbid ID46, 47 level and APOE Īµ4 status28 were not associated with changes in BMI over time. These findings are consistent with previous research in both the general population48 and individuals with DS.28
In the current study, a higher baseline PET AĪ² level was significantly associated with greater decline in BMI over time in adults with DS. Indeed, the BMI of adults with DS was estimated to decrease by 0.6 kg/m2 from the time of AĪ² positivity (18 centiloids (CL))18 to having a PET AĪ² value of 60 CL and by 1.2 kg/m2 at a PET AĪ² value of 100 CL. The association between higher baseline AĪ² and lower BMI over time held true even when only examining the subset (n = 191) of adults with DS who were cognitively stable and in models controlling for age. Thus, the association between AĪ² burden and BMI decline occurs prior to clinical AD dementia and does not appear to be attributable only to ageārelated effects on BMI. The potential biological mechanisms linking weight loss to early AĪ² burden are not fully understood. However, in mouse models, AĪ² deposition disrupts hormone regulation, including leptin and adiponectin signaling,10, 11 and is posited to affect brain regions related to appetite control and energy regulation, such as the entorhinal cortex and hippocampus.12, 13 In our study, there was no statistical association between baseline tau PET and BMI decline. The lack of significant findings may be due to the smaller sample of participants with available tau data (relative to AĪ² data), which limited statistical power. In addition, an effect of tau burden on BMI decline may have been overshadowed by stronger associations with AĪ² deposition. Given the short time lag between AĪ² positivity and tau positivity in adults with DS compared to LOAD,45, 49 it may be challenging to disentangle the individual contributions of tau relative to AĪ² pathology. Alternatively, it is possible that BMI decline is more closely linked to biological mechanisms triggered by AĪ² than tau burden.
Decline in memory and cognitive functioning across time was also associated with BMI decline in adults with DS, even in models controlling for age and other sociodemographic characteristics. For example, a 5% decrease in the mCRT Total score was associated with a 0.1 kg/m2 decrease in BMI. Similarly, a 21% increase on the NTGāEDSD corresponded to a 0.1 kg/m2 decrease in BMI. Results held true when only examining participants deemed to be cognitively stable, suggesting that BMI decline coincides with early and subtle AD symptomatology. Moreover, in models controlling for age, adults with DS who were obese at baseline had a flatter decline in cognition (mCRT and DSMSE) across time than those who started with a lower BMI (underweight or normal weight). This finding may similarly reflect that those with lower BMIs were farther along in the unfolding of AD symptomatology than those with higher BMIs. Together, these findings suggest that BMI decline may be part of early AD expression and observed years prior to dementia onset. However, it is important to note that in later disease stages (e.g., following clinical AD dementia onset), cognitive decline and other AD symptoms may exacerbate weight loss in adults with DS. Indeed, outside of DS, ADārelated impairments in memory and attention, associated with medial temporal lobe reduction,50 may contribute to forgetting to eat, and other dementia symptoms such as difficulty swallowing may result in reduced caloric intake.51
The present study had both strengths and limitations. In terms of strengths, the study leveraged up to four cycles of data (each spaced ā16 months apart) in a large cohort of adults with DS, most of whom did not yet have clinical AD dementia. The study used PET imaging to assess AD pathology with methods harmonized across sites. Cognitive functioning was assessed using a battery of directly administered and informantābased measures. In terms of limitations, the use of BMI, while practical, does not indicate the specific nature of weight loss (e.g., whether it stems from reductions in muscle mass, bone density, or fat mass). Incorporating body composition scanners into future studies would better identify the sources of weight loss and elucidate the underlying mechanisms. This study cannot fully determine if weight loss was intentional or unintentional. None of the participants had weight loss surgery or were reported to be taking glucagonālike peptideā1 weight loss medications. Many participants had hypothyroidism (60%) and obstructive sleep apnea (36%). These conditions were typically longstanding, but it is possible that their severity altered during the course of the study. In addition, the main results were unaltered in followāup analyses when participants with the greatest weight loss (>10 units, N = 10 [2.14%]) were removed. However, separating intentional from unintentional weight loss will be important in future studies. The present study also lacked diversity, with 92% of participants identifying as White, nonāHispanic. To enhance the generalizability of findings, future research should include samples with broader demographic variability, including measures of economic disadvantage as well as race/ethnicity and biological sex.
In summary, weight loss could be an important sign of impending AD dementia in adults with DS. Further research is also need to disentangle the timeāordered nature of BMI change, age, AĪ² deposition, and cognitive decline in DS in order to identify the biological mechanisms that may drive associations. In the meantime, the findings have important clinical implications. Weight change should be included in DSAD screenings and weight should be monitored in annual visits of adults with DS.
CONFLICT OF INTEREST STATEMENT
The authors declare no conflicts of interest. Author disclosures are available in the. Supporting Information
CONSENT STATEMENT
Informed consent and/or assent were obtained for all participants.
Supporting information
ACKNOWLEDGMENTS
Data collection and sharing for this project was supported by The Alzheimer's Biomarkers ConsortiumāDown Syndrome (ABCāDS) funded by the National Institute on Aging (NIA) and the Eunice Kennedy Shriver National Institute of Child Health and Human Development (U01 AG051406 and U01 AG051412). All research at the Department of Psychiatry at the University of Cambridge is supported by the National Institure for Health and Care Research (NIHR) Cambridge Biomedical Research Centre (NIHR203312) and the NIHR Applied Research Collaboration East of England. The views expressed are those of the authors and not necessarily those of the NIHR or the Department of Health and Social Care. The authors thank the adults with Down syndrome who volunteered to participate in this study for their invaluable contributions to this work, along with their families and service providers. Finally, the authors thank the staff contributing the many hours of support to the collection of data, including Jessica BeresfordāWebb, Malwina Filipczuk, Shemaya HurdāThomas, Shara Khoo, and Ellie Maycock. This research was funded by the NIA (F31AG085730; K01AG083130; R01AG070028; U01AG051412; U19AG068054). This study was supported in part by a core grant to the Waisman Center from the National Institute of Child Health and Human Development (P50HD105353). This study was also supported by the NIHR Cambridge Biomedical Research Centre (NIHR203312) and the NIHR Applied Research Collaboration East of England.
Fleming VL, Helsel BC, Ptomey LT, etĀ al. Longitudinal study of body mass index in relation to Alzheimer's disease pathology and symptomatology in Down syndrome. Alzheimer's Dement. 2025;21:e70387. 10.1002/alz.70387