Using a genetic risk score to estimate the earliest age of Alzheimer’s disease-related physiologic change in Body Mass Index

Study Setting and Participants

UK Biobank is an ongoing study of over 500,000 adults. Participants aged 40-69 years old were recruited 2006-2010 from across the UK to provide detailed information about themselves via computerized questionnaires, provide biologic samples, undergo clinical measurements, and have their health followed prospectively.¹⁸ In this current analyses, we excluded those with missing genetic information (n=15,221) or who were flagged as recommended for genetic analysis exclusion (n=378), and those with missing BMI (n=1,295). We also excluded participants classified as of non-European genetic ancestry (n=78,336) based on genetic ancestry principal components, because genetic predictors of AD may differ by ancestry/population stratification.¹⁹ After exclusions, 407,386 participants were included in the analytic sample. Ethical approval for UK Biobank was obtained from the National Health Service National Research Ethics Service and all participants provided written informed consent.

Patient and Public Involvement

Patients were not involved in setting the research question, study design, data analysis or interpretation of the results.

Genotyping and Genetic Risk Scores for AD

Genotyping of UK Biobank samples was conducted with two closely related arrays (Affymetrix using a bespoke BiLEVE Axiom array and Affymetrix UK Biobank Axiom array) and is described in detail elsewhere.^20,21 Briefly, all genetic data were quality controlled and imputed by UK Biobank [downloaded on 12/1/2017] to a reference panel that merged the 1,000 Genomes Phase 3 and UK10K reference panels. A secondary imputation was completed using the Haplotype Reference Consortium (HRC) reference panel and results from the HRC imputation were preferentially used at SNPs present in both panels. Before the release of the UK Biobank genetic data a stringent QC protocol (described elsewhere) was applied at the Welcome Trust Centre for Human Genetics.²²

To construct the AD-GRS in UK Biobank, we used summary results from the 2013 International Genomics of Alzheimer’s Project (IGAP) meta-analyzed genome-wide association study (GWAS) on late-onset AD in Caucasian populations¹⁵ to calculate an AD genetic risk score (GRS) for each participant. The IGAP study identified 23 loci associated with AD, including two SNPs used to characterize APOE ε4 allele status. The GRS was based on the meta-analyzed β coefficients obtained in the IGAP’s stage 1 study, which included genotyped and imputed data (7,055,881 single nucleotide polymorphisms, 1000G phase 1 alpha imputation, Build 37, Assembly Hg19) of 17,008 Alzheimer’s disease cases and 37,154 controls. We calculated the GRS by multiplying each individual’s risk allele count for each locus by the β coefficient (expressed as the log OR) for that polymorphism (Table 1) and adding the products for all 23 loci. This step weights each SNP in proportion to the observed association with AD risk (either positive or negative). The scores can be interpreted as the log OR for AD conferred by that individual’s profile on the 23 SNPs compared to a person who had the major allele at each locus. We converted the AD-GRS into a standardized z-score based on the sample mean and variance. This score has previously been shown to predict cognition and dementia-related death in the UK Biobank.¹⁶

Body Mass Index

BMI was calculated based on height and weight (kg/m²) measured at the baseline assessment.

Other Characteristics

Age, sex, and education were reported at baseline assessment. UK Biobank provides principal components (PCs) related to genetic population stratification, we used the first 10 PCs in our analyses to adjust for population stratification. Participants underwent a touchscreen-based cognitive battery, we focused on reaction time (i.e. simple processing speed) as a measure of cognition as it was available on the largest number of participants. Participants were timed at pressing a button as soon as two identical cards are seen on the touchscreen.

Statistical Analysis

In preliminary analyses, we conducted analyses stratified by age at assessment (39-49 years, 50-59, and 60-73 years old). First, we evaluated the association between BMI and reaction time to confirm expected associations seen in prior research (e.g. a positive association between BMI and cognition in mid-life and a negative association between BMI and cognition in late-life). We used linear regression with reaction time as the outcome and BMI as the primary predictors with adjustment for age, age², sex, and education. Next, as our primary analysis, we estimated separate age-stratified linear regressions with the AD-GRS as the primary predictor and BMI as the outcome. Models included adjustment for age, age², sex, and PCs. Finally, we estimated non-linear trends for age using one linear regression model for the whole sample with AD-GRS, age, and age² as primary predictors. We tested whether age trends differed by AD-GRS using interaction terms between the linear and quadratic age terms and AD-GRS. Based on this regression equation we calculated the age at which curves began to diverge. To illustrate the effects of low and high AD genetic risk we calculated predicted BMI curves based on the 10^th percentile (low risk) and 90^th percentile (high risk) of the AD-GRS. Based on these same percentiles we also estimated the earliest age at which high AD-GRS was associated with a significantly lower BMI compared to low AD-GRS. All regression models included adjustment for sex and PCs as described above. Analyses were conducted in R (version 3.3.2). All tests were two-sided with α = 0.05 and we report 95% confidence intervals (CIs).

Evidence for BMI as an early manifestation of AD

This study expands upon prior work showing that weight loss is a predictor of subsequent AD diagnosis^4–6,12,23 and AD neuropathologic burden.²⁴ The majority of prior studies have been based on observational data in which it is not possible to establish temporal ordering of BMI relative to established markers of preclinical AD. A number of studies examining BMI in mid to late life give conflicting estimates of the age at which lower BMI is associated with AD, ranging from 20 to 8 years prior to AD diagnosis.^13,25,26 Novel approaches can help triangulate evidence around the timing of BMI change in preclinical AD.²⁷ By examining genetic risk, which is determined at birth, we provide evidence that some biological process associated with AD genetic risk affects BMI as early as age 50. Using a similar approach to our study, the Dominantly Inherited Alzheimer Network (DIAN) study assessed BMI in “preclinical AD” by comparing the point at which BMI curves began to diverge between autosomal dominant AD mutation carriers and non-carriers. They found that BMI curves for mutation carriers diverged from non-carriers as early as 17.8 years prior to expected symptom onset.⁷ Given that the average age of AD diagnosis is in the 70s we find a similarly long period (∼20 years) in which BMI may diverge between high and low AD genetic risk. Together these studies suggest processes related to AD begin many decades prior to diagnosis. We may need to target interventions in mid-life or earlier to effectively prevent AD or modify the earliest events in the AD pathologic cascade.

Our findings support the interpretation that pathophysiologic processes that culminate in AD diagnosis lead to weight loss decades earlier.²⁸ Weight loss in AD may be caused by several factors, but the exact cause is unknown. Nutritional habits are worse in individuals with mild cognitive impairment or dementia;¹⁰ however UK Biobank participants do not have dementia and divergence in BMI curves begins years prior to symptom onset. There is also evidence that neurodegeneration and neuropathologic changes in AD are associated with lower BMI and weight loss.^9,24 Brain regions and pathways important for metabolism, appetite, and weight maintenance such as the hypothalamic-pituitary-adrenal axis are affected in AD and may become disrupted early in the disease process.^8,29 Our results are consistent with BMI as an early marker for changes due to AD, however, it may also be the case that pleiotropy leads to population associations between weight loss and the likelihood of AD. Although we cannot conclusively say these BMI changes are an early manifestation of AD, our findings complement prior observational studies. Additionally, even if pleiotropic pathways link the AD-GRS to weight loss, our findings confirm that lower BMI as early as midlife can predict AD and cognitive decline.

Limitations and study strengths

There are several important caveats and limitations to our analysis. Our measurement of BMI is cross-sectional; future studies with longitudinal BMI measurements will be needed to verify age at which weight loss can begin to occur in preclinical AD. Our results are potentially influenced by selection bias due to selective survival as the AD-GRS is associated with mortality.¹⁶ However, this effect is likely limited because the sample is relatively young and healthy with a low mortality rate, especially due to dementia.¹⁶ Genetic risk for AD only explains a small percentage of variation in cognitive impairment and AD diagnosis, thus there is substantial variation in risk for AD that is not captured by our risk score. This variation reduces the ability to detect associations. These analyses were conducted in participants of European ancestry and were generally healthy and well educated; this may limit generalizability. However, there are considerable strengths to this study, particularly the large sample size, the use of measured BMI, and an innovative analytic approach. By using genetic risk, we circumvent a central challenge in interpreting prior results on weight loss, BMI and AD.