Participants
Data analyzed in this study were obtained from the DIAN Data Freeze 11. The DIAN observational study is an international multi-site study that enrolls family members who have parents with a mutated gene known to cause DIAD [20]. The study participants may or may not be mutation carriers and they may or may not have cognitive symptoms. The participants underwent standardized clinical and cognitive testing, brain imaging, and biological fluid collection (blood, cerebrospinal fluid [CSF]) to determine the sequence of changes in pre-symptomatic mutation carriers who are destined to develop AD.
In this study, we selected cognitively normal DIAD mutation carriers and non-carriers from the DIAN cohort with clinical dementia rating (CDR) [21] score of 0 and mini-mental state examination [22] score ≥24 [18,23]. Individuals who were mildly symptomatic (CDR=0.5) or overtly symptomatic (CDR > 0.5) were excluded. The results of neuropsychiatric inventory-questionnaire (NPI-Q) and [18F]Flurodeoxyglucose (FDG) positron emission tomography (PET) performed at the first visit and at subsequent yearly follow ups (if available) for each participant were analyzed.
Ethical approval and patient consents
The DIAN study was approved by the Institutional Review Boards of all of the participating institutions. Informed written consent was obtained from all participants at each site.
Neuropsychiatric assessments
The NPI-Q is an informant-based assessment tool that measures the presence and severity of behavioural disturbances in 12 behavioural domains of agitation, anxiety, apathy, appetite changes, delusions, depression, disinhibition, abnormal elevated mood, hallucinations, irritability, repetitive motor behaviours, and sleep behaviour changes in clinical settings, within the past month [24]. Higher NPI-Q scores represent greater severity of NPS.
Estimated year of onset (EYO) of DIAN
The estimated age of onset of cognitive impairment in the cognitively normal individuals from the DIAN was calculated based on the mean mutation age of symptom onset and/or the parental age of symptom onset according to the following steps as described previously [25]:
(i) At any study visit, the EYO was calculated as the visit age subtracting the mean mutation age of symptom onset if the individual’s mutation is known, which is available in the DIAN database. (ii) If the individual’s mutation is not available in the DIAN database (e.g. the mutation has not been previously reported or other member age of onset not available), then at any study visit, the EYO was calculated as the visit age subtracting the parental age of symptom onset. The shorter the EYO, the closer the proximity of the individual’s time of clinical disease.
Genetic analysis
Sequencing of the APP, PSEN1 and PSEN2 genes was performed by the DIAN Genetics Core investigators as previously described [18], to reveal the presence of disease-causing mutation in AD at-risk individuals.
CSF analysis
CSF levels of Aβ1-42, total tau (t-tau) and p-tau181 were measured by immunoassay by the DIAN Biomarker Core at the Washington University, using the Luminex bead-based multiplexed xMAP technology (INNO-BIA AlzBio3™, Innogenetics, Ghent, Belgium) as previously described [26].
MRI and PET methods.
MRI and PET standard acquisition protocols have been described in the DIAN website. T1-weighted MRI images corrected for field distortions were processed with the CIVET image processing pipeline [27] and the PET images were processed with an established image processing pipeline described previously [28]. The pre-processed images from the DIAN database were spatially normalized to the Montreal Neurological Institute (MNI) 152 standardized space by using the transformations obtained for PET native to MRI native space and the MRI native to the MNI 152 space. The [18F]FDG PET standardized uptake value ratio (SUVR) maps were then generated using the pons as the reference region [29,30]. The global brain glucose uptake was calculated by averaging the [18F]FDG SUVR within several brain regions characteristic to the AD process, including the precuneus, pre-frontal, orbitofrontal, parietal, temporal, anterior, and posterior cingulate cortices.
Statistical analysis
The descriptive statistics and frequency distributions of baseline demographics, mutation characteristics and CSF AD biomarkers were summarized and compared between DIAD mutation carriers and non-carriers using family-level random-effect models for both continuous and categorical measurements using the STATA 15.0 software. Principal components were derived for the variables NPI-Q and EYO to resolve collinear relationships.
The linear mixed effect models with family-level random effects evaluated the interactions between NPI-Q (total and sub-scale scores individually and as a whole), age and EYO on FDG SUVR in the mutation carriers and non-carriers. We modelled FDG SUVR as a function of the interactions of NPI-Q, age and EYO and covariates, where FDG SUVRij denotes the FDG uptake for the jth person from the ith family, NPI-Qij indicates the severity of NPS, ageij indicates the age of participant at the time of study visit, EYOij indicates the years to estimated age of symptom onset and Xij represents fixed effect covariates for gender, education, APOE ε4 status and family mutation type (APP, PSEN1 and PSEN2):
FDG SUVRij ~ β0 + β1(NPI-Qij) + β2(EYOij) + β3(Ageij) + β4(NPI-Qij×EYOij) + β5(NPI-Qij×Ageij) + β6(EYOij×Ageij) + β7(NPI-Qij×EYOij×Ageij) + β8(Xij) + Fi + εij
where Fi represents a random effect for all individuals from family i, and εij is the residual error assumed independent and normally distributed for all individuals.
The family-level random effect accounts for the correlations between individuals within the same family. Although correlations between family members might vary with the relationship type, due to the fairly small sizes of the families, this was modelled with a single random effect.
Voxel-based statistical analyses were then performed using the R Statistical Software Package version 3.3.0 with the RMINC library [31], to test the interactions of NPI-Q, age and EYO on FDG SUVR in the DIAD mutation carriers and non-carriers. All voxel-based regression analyses were corrected for multiple comparisons using random field theory [32] at p < 0.001.
Exploratory factor analysis was performed on the sub-components of NPI-Q to identify the neuropsychiatric subsyndromes within the DIAD mutation carriers. This exploratory analysis aims to determine the multidimensional relationships of the NPI-Q sub-components and their overall effects over the individual contributions on the outcomes of interest. Linear mixed effect models with family-level random effects then evaluated the interactions of specific neuropsychiatric subsyndromes and EYO on FDG SUVR in the mutation carrier group.