RELATIONSHIPS BETWEEN BASELINE BRAIN IMAGING BIOMARKER MEASUREMENTS AND AGE IN THE API AUTOSOMAL DOMINANT ALZHEIMER’S DISEASE COLOMBIA TRIAL
Yi Su,Silvia Rios-Romenets,Pierre N. Tariot,Kaycee Sink,David Clayton,Nan Hu,Heather Guthrie,Jillian Smith,William Cho,Jessica B. Langbaum,Ronald G. Thomas,Margarita Giraldo,Carlos Tobon,Natalia Acosta-Baena,Alexander Navarro,Francisco Piedrahita,Sergio Alvarez,Kewei Chen,Dhruman D. Goradia,Pradeep Thiyyagura
DOI: https://doi.org/10.1016/j.jalz.2019.06.4727
2019-01-01
Abstract:We previously used data from 54 cognitively unimpaired and impaired 20–59 year-old presenilin 1 (PSEN1) E280A mutation carriers and noncarriers from the world's largest autosomal dominant kindred (median age of 44 at mild cognitive impairment [MCI] onset) to characterize associations with age and estimated ages at onset (AAO) of brain imaging and fluid biomarker abnormalities in the mutation carrier group (Fleisher, 2015). Here, we analyzed baseline brain imaging measurements of amyloid-β (Aβ) plaque deposition, precuneus glucose hypometabolism, and hippocampal volume from 242 cognitively unimpaired 30–53 year-old PSEN1 E280A mutation carriers and non-carriers in the Alzheimer's Prevention Initiative (API) ADAD Colombia Trial, characterized their relationships with age, and estimated their AAO in the mutation carrier group. Baseline florbetapir PET, fluorodeoxyglucose (FDG) PET, and volumetric MRI images from 167 mutation carriers and 75 age matched noncarriers were analyzed using previously established pipelines (Fleisher, 2015; Su 2015) to characterize 1) Aβ plaque burden using mean-cortical-to-pontine florbetapir standard uptake value ratios (SUVRs), 2) cerebral glucose hypometabolism using precuneus-to-whole brain FDG SUVRs, and 3) neurodegeneration using hippocampal-to-intracranial volume ratios. Linear regression models were used to characterize biomarker associations with age in the two genetic groups. AAOs, defined as the age at which mean biomarker values diverge significantly between the mutation carrier and non-carrier groups, were estimated using both approximate t-tests and the 95% confidence interval (CI) band of age associations. AAO of mean cortical increases in florbetapir SUVR, decreases in precuneus FDG SUVR, and declines in hippocampal volume were less than 30 years, 36 (95%CI 30-40) years, and 42 (95% CI 37-48) years, respectively. This study provides information about associations between baseline brain imaging measurements and age in a large number of cognitively unimpaired PSEN1 E280A mutation carriers and non-carriers from the API ADAD Colombia Trial. Findings are roughly consistent with observational studies of ADAD (Bateman, 2012; Fleisher 2015), and the extent to which they support the hypothetical ordering of biomarker changes in preclinical AD (Jack 2018) will be presented.
What problem does this paper attempt to address?