Neuropsychiatric symptoms herald metabolic decline in Alzheimer’s disease

Neuropsychiatric symptoms (NPS) are increasingly recognized as early non-cognitive manifestations in the Alzheimer’s disease (AD) continuum. However, the role of NPS as an early marker of pathophysiological progression in AD remains unclear. Dominantly inherited AD (DIAD) mutation carriers are young individuals who are destined to develop AD in future due to the full penetrance of the genetic mutation. Hence, the study of DIAD mutation carriers enables the evaluation of the associations between pure AD pathophysiology and metabolic correlates of NPS without the confounding effects of co-existing pathologies. In this longitudinal study, we aim to identify regional brain metabolic dysfunctions associated with NPS in cognitively intact DIAD mutation carriers. We stratied 221 cognitively intact participants from the Dominantly Inherited Alzheimer’s Network according to their mutation carrier status. Regression mixed effect models with family-level random effects evaluated the interactions of NPS measured by the Neuropsychiatric Inventory-Questionnaire (NPI-Q), age and estimated years of onset (EYO) as a function of metabolism measured by [ 18 F]urodeoxyglucose ([ 18 F]FDG) positron emission tomography in DIAD mutation and non-mutation carriers. An exploratory factor analysis was performed to identify the neuropsychiatric subsyndromes in DIAD mutation carriers using the NPI-Q sub-components. Regression mixed effect models then evaluated the interactions of specic neuropsychiatric subsyndromes with EYO and age on metabolism. Exploratory factor analysis performed on the subcomponents of NPI-Q to identify the neuropsychiatric subsyndromes within the DIAD mutation carriers veried using factor analysis to assess the t. mixed effect models with family-level random effects then evaluated the interactions of specic neuropsychiatric subsyndromes and EYO and age on FDG SUVR in the mutation carrier groups. using family-level random-effects models for the continuous variables categorical variables, taking into account the analysis of multiple family members within families. subsyndrome 18 F]FDG uptake CI p global [ 18 F]FDG uptake decline in

Neuropsychiatric symptoms (NPS) are frequently observed in the mild cognitive impairment (MCI) and dementia stages of Alzheimer's disease (AD) [1,2] and are associated with greater functional impairment, poorer quality of life and accelerated cognitive decline [3][4][5]. In cognitively normal individuals, developing NPS later in life may potentially increase the risk of cognitive decline [6,7]. Therefore, NPS are increasingly being proposed as noncognitive manifestations in the early stages of AD when one is cognitively intact [8]. However, the role of NPS as early clinical manifestations of AD pathophysiological progression in cognitively normal individuals remains unclear. Our recent study showed that NPS in preclinical sporadic AD individuals preceded hypometabolism in the posterior cingulate cortex, a key brain region involved in the AD process [9]. Hence, further studies indicating NPS as an early manifestation of metabolic decline in an independent cognitively intact cohort known to have AD pathophysiology will further advance the emerging conceptual framework in which NPS constitute an early clinical manifestation of AD.
Dominantly inherited AD (DIAD) is a familial AD due to autosomal dominant mutations in the APP, PSEN1 or PSEN2 genes and cognitively intact individuals who are DIAD mutation carriers are destined to develop AD in future due to the full penetrance of the genetic mutation [10]. Similar to late-onset sporadic AD [11], the pathophysiology of DIAD begins to accumulate in the preclinical stage of the disease when carriers are cognitively normal [12,13]. In DIAD, early behavioural changes are reported in mildly cognitively symptomatic mutation carriers in whom the NPS increase as their disease progresses [14]. Furthermore, DIAD mutation carriers are younger compared to individuals with sporadic AD and are less likely to have other medical conditions such as cerebrovascular disease [15]. Therefore, studying cognitively normal DIAD mutation carriers constitutes a valuable strategy to evaluate the associations between pure AD pathophysiology and the metabolic correlates of NPS in the preclinical stage of AD without the confounding effects of co-existing pathologies.
Here, in a longitudinal observation of cognitively normal DIAD mutation and non-mutation carriers from the Dominantly Inherited Alzheimer Network (DIAN) [16], we will test the hypothesis that the severity of NPS is associated with greater metabolic decline in cognitively intact DIAD mutation carriers as their disease progresses. We will further test the speci c neuropsychiatric subsyndromes that drive the metabolic decline in the DIAD mutation carriers.

Study Participants
Data analysed in this study were obtained from the Dominantly Inherited Alzheimer's Network (DIAN) Data Freeze 11. The DIAN observational study is an international multi-site study that enrols biological adult children of a parent with a mutated gene known to cause DIAD [16]. Study participants may or may not be mutation carriers and they may or may not have cognitive symptoms. DIAN study participants undergo standardized clinical and cognitive testing, brain imaging, and biological uid collection (blood, cerebrospinal uid) with the goal of determining the sequence of changes in pre-symptomatic gene carriers who are destined to develop AD.
In this study, we selected cognitively normal DIAD mutation and non-mutation carriers from the DIAN cohort supported by clinical dementia rating (CDR) [17] score 0 and mini-mental state examination [18] score ≥ 24. Each participant's rst visit and subsequent yearly follow ups, if available, with completed neuropsychiatric inventory-questionnaire (NPI-Q) and [ 18 F]Flurodeoxyglucose (FDG) positron emission tomography (PET) will be analyzed.

Ethical Approvals 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 that represent a change from the baseline within the past month, in the 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 [19]. Higher NPI-Q scores represent greater severity of NPS.

DIAN estimated years of onset (EYO)
The estimated age of onset of cognitive impairment in 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 in the following ordered steps: i. At any study visit, EYO equals to the visit age minus the mean mutation age of symptom onset if the individual's mutation is known and the mean mutation age of symptom onset for this individual's mutation is available in the master DIAN database. ii. If the individual's mutation is not available in the master DIAN database (e.the mutation has not been previously reported or other member age of onset not then at any study visit, EYO equals to the visit age minus the parental age of symptom onset.
The shorter the EYO, the closer the proximity of the individual's time of clinical disease.
Genetic analysis DNA sequencing of the APP, PSEN1 and PSEN2 genes was performed by the DIAN Genetics Core investigators as previously described [14] to establish the presence of disease-causing mutation in ADAD at-risk individuals.
MRI and PET methods.
MRI and PET standard acquisition protocols were described in the DIAN website. T1-weighted MRI images corrected for eld distortions were processed with the CIVET image processing pipeline [21] and the PET images were processed with an established image processing pipeline previously described [22]. 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 [ 18 F]FDG PET standardized uptake value ratio (SUVR) maps were then generated using the pons as the reference region. The global brain glucose uptake was calculated by averaging the [ 18 F]FDG SUVR within several brain regions characteristic to the AD process, including the precuneus, prefrontal, orbitofrontal, parietal, temporal, anterior, and posterior cingulate cortices.

Statistical Analysis
Descriptive statistics and frequency distributions of baseline demographics, mutation characteristics and CSF AD biomarkers were summarized and compared between DIAD mutation and non-mutation carriers using familylevel random-effect models for both continuous and categorical measurements using STATA 15.0. Principal components were derived for the variables NPI-Q, and & EYO to resolve collinear relationships.
Linear mixed effect models with family-level random effects evaluated the interactions between NPI-Q, age and EYO on FDG SUVR in the mutation and non-mutation groups. We modelled FDG SUVR as a function of the interactions of NPI-Q, age and EYO and covariates, where FDG SUVR ij denotes the FDG uptake for the jth person from the ith family, NPI-Q ij indicates the severity of NPS, age ij indicates the age of participant at the time of study visit, EYO ij indicates the years to estimated age of symptom onset and X ij represents xed effect covariates for gender, education, APOE ε4 status and family mutation type (APP, PSEN1 and PSEN2): The family-level random effect term 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 [23]. Voxel-based regression models tested the interactions of NPI-Q, age and EYO on FDG SUVR in the DIAD mutation and non-mutation carrier groups. All voxel-based regression analyses were corrected for multiple comparisons using random eld theory [24] at p < 0.001.
Exploratory factor analysis was performed on the subcomponents of NPI-Q to identify the neuropsychiatric subsyndromes within the DIAD mutation carriers and veri ed using con rmatory factor analysis to assess the t. Linear mixed effect models with family-level random effects then evaluated the interactions of speci c neuropsychiatric subsyndromes and EYO and age on FDG SUVR in the mutation carrier groups.

Results:
Baseline demographics, mutation characteristics and CSF AD biomarkers Two hundred and twenty-one (n = 221) cognitively intact individuals (102 (46.15%) non-mutation and 119 (53.85%) DIAD mutation carriers) were included in this study. 29 individuals had year 1, 24 had year 2, 46 had year 3, 7 had year 4, 4 had year 5 and 3 had year 6 follow-up data. Baseline demographics, APOE ε4 carrier status, and AD CSF biomarker characteristics were summarized in Table 1. The DIAD mutation and non-mutation carriers did not differ signi cantly in age, gender, education and APOE ε4 status. As expected, DIAD mutation carriers had lower CSF Aβ 1−42 , higher CSF p-tau 181p and CSF t-tau levels than non-mutation carriers.

Interactions of NPI-Q and EYO on regional [ 18 F]FDG uptake
The voxel-based analysis indicated that the interaction between higher NPI-Q and shorter EYO was associated with regional [ 18 F]FDG uptake decline in the posterior cingulate cortex (PCC), ventromedial prefrontal cortex (vmPFC), bilateral parietal lobes and right insula in DIAD mutation carriers (Fig. 1). There was no statistically signi cant interaction between NPI-Q and EYO on regional [ 18 F]FDG uptake in DIAD non-mutation carriers. There was no statistically signi cant interaction between NPI-Q and age, and NPI-Q, age and EYO on regional [ 18 F]FDG uptake, in both DIAD mutation and non-mutation carriers.
Factor analysis of NPI-Q subcomponents A 4 factor solution was applied to the NPI-Q subcomponents, which explained 70% of the variance. Crossloading of the 4 factors resulted in a good t of comparative t index (CFI) = 0.837. The neuropsychiatric subsyndromes derived from the factor analysis were 1) agitation, disinhibition, irritability and depression; 2) anxiety, apathy, depression, motor behaviours, and sleep behaviour changes; 3) delusion, sleep behaviour changes and irritability; 4) appetite and anxiety (Table 2). In DIAD mutation carriers, we found that only the interaction of the neuropsychiatric subsyndrome agitation, disinhibition, irritability, depression and shorter EYO was associated with global [ 18 F]FDG uptake decline (β = -0.044, 95% CI -0.071 to -0.017, p = 0.001). The interaction of the same neuropsychiatric subsyndrome and higher age was also associated with global [ 18 F]FDG uptake decline (β = -0.049, 95% CI -0.078 to -0.020, p = 0.001) in DIAD mutation carriers.

Discussion:
The present study showed that the emergence of NPS heralded metabolic dysfunction in brain regions susceptible to AD pathophysiology in cognitively intact DIAD mutation carriers. In these individuals who were destined to develop AD, the more severe the NPS and the shorter the EYO to AD dementia onset, the greater the metabolic decline in the PCC, vmPFC, bilateral parietal lobes and right insular. We found that the metabolic dysfunctions were driven by the neuropsychiatric subsyndrome of agitation, disinhibition, irritability and depression.
Accumulating evidence have demonstrated the importance of NPS as predictors of cognitive decline in cognitively normal individuals. In the population-based Mayo Clinic Study of Aging, the presence of NPS at baseline increased the risk of incident MCI compared to those without NPS [6]. In the Alzheimer's Disease Cooperative Study (ADCS) Prevention Instrument Project, anxiety and depression at baseline predicted CDR conversion to ≥ 0.5 in cognitively intact older subjects over a 4-year follow-up [25] while in the National Alzheimer's Coordinating Center (NACC) cohort, cognitively normal participants who developed CDR > 0 during follow up had a signi cantly earlier presence of NPS [7]. NPS among cognitively normal individuals may also represent an early manifestation of progressive metabolic dysfunction. In cognitively normal persons aged > 70 years, depressive and anxiety symptoms were associated with decreased FDG uptake in AD-related regions [26].
In a recent study of preclinical sporadic AD individuals, we found that NPS were associated with metabolic dysfunctions in the limbic network and predicted hypometabolism in the PCC [9]. Our present ndings in a cohort of preclinical familial AD mutation carriers who are destined to develop AD in future further support the emerging conceptual framework that NPS are early non-cognitive manifestations of AD pathophysiology and herald subsequent metabolic decline.
The default mode network (DMN), which comprises of the PCC, vmPFC and inferior parietal lobes, plays a vital role in episodic memory processing and decreased metabolism in the DMN is observed early in the course of AD [27,28]. The salience network (SN) which is critical in detecting and integrating behavioural and emotional stimuli, has key nodes in the insular cortex and modulates the switch between the DMN and the central executive network [29,30]. The impairment of the SN can lead to numerous neuropsychiatric disorders such as psychosis [31] and depression [32]. Brain metabolic dysfunctions within the SN are also related to NPS in AD [33]. Therefore, our nding of NPS heralding greater FDG uptake decline in the PCC, vmPFC, parietal lobes, and right insula in DIAD mutation carriers with shorter EYO to onset of AD dementia supports the link between early NPS, limbic structures and brain regions involved in early AD pathophysiology.
While there is heterogeneity in the neuropsychiatric manifestations in AD, certain NPS tend to co-express. Hence, several neuropsychiatric sub-syndromes have been identi ed to characterise the clustering of NPS [34] in AD. In our study, an exploratory factor analysis revealed four neuropsychiatric subsyndromes (Table 3) and among them, only the neuropsychiatric subsyndrome "agitation, disinhibition, irritability and depression" was associated with regional metabolic decline in cognitively intact DIAD mutation carriers with shorter EYO to onset of AD dementia. This neuropsychiatric subsyndrome is consistent with ndings from a systemic review of behavioural and psychological subsyndromes among elderly individuals with sporadic AD, where 34 different clusters were found and the quartet of agitation/aggression, depression, anxiety and irritability were most commonly clustered together [35]. In addition, delusion and hallucinations, depression and anxiety, agitation and irritability and euphoria and disinhibition tend to be frequently associated symptoms. Hence, our nding is consistent with the current evidence of neuropsychiatric subsyndromes in AD. In addition, given that currently reported subsyndromes are mostly de ned among elderly individuals with sporadic AD, our results further advanced the eld by identifying speci c neuropsychiatric subsyndromes among younger cognitively intact DIAD mutation carriers who are destined to develop AD.
The neurobiology of agitation, disinhibition, irritability and depression is closely linked to dysfunctions within the PCC, vmPFC, bilateral parietal lobes and right insular. The vmPFC enables the use of emotional signals to guide decisions towards the advantageous direction, regulates behavioural responses such as changing reinforcement contingencies and emotional processing and regulation [36,37]. The manifestation of irritability is linked to abnormal emotional processing associated with vmPFC and PCC, while behavioral disinhibition and prominent emotional lability are linked to lesions in the vmPFC. Dysfunctions within the vmPFC and PCC, which are part of the neural network involved in the modulation of normal emotional behaviour, also lead to affective disorders and depressive symptoms [38]. The insular plays a key role in producing appropriate behavioral responses in a person by integrating affective, homeostatic, and higher-order cognitive processes [39]. Irritability is associated with dysfunctions within the insular [40] while AD patients with agitation also appear to have dysfunctions within the frontal cortex, anterior cingulate cortex (ACC), orbitofrontal cortex, amygdala, and insula [41]. While the neuropsychiatric subsyndrome of agitation, disinhibition, irritability and depression is linked to regional metabolic decline in cognitively normal DIAD mutation carriers, further studies are needed to evaluate if the neuropsychiatric subsyndrome also identi es cognitively normal individuals with an increased risk of pathological progression in sporadic AD.
The main strength of the present longitudinal study is the inclusion of preclinical DIAD mutation carriers who have AD pathology and are destined to develop AD in future. This allows the study of the associations between NPS, metabolism and effects of increasing AD pathology over time (EYO). Furthermore, given that individuals may be susceptible to NPS presentations due to genetic, family and environmental factors, or being at risk for DIAD, studying both DIAD mutation and non-mutation carriers enables the control of these factors.
There are limitations to our study. Firstly, while NPI-Q is commonly used to detect NPS in AD patients, the NPI-Q was not developed for patients with prodromal or preclinical AD. Hence, the sensitivity of NPI-Q in identifying early NPS in cognitively intact individuals remains unclear. In addition, given that the NPI is based on responses from an informed caregiver, the NPI scores may not accurately re ect the NPS of study participants. In this regard, the mild behavioral impairment checklist (MBI-C) [42] which is a 34-item instrument completed by either the patient, close informant, or clinician, has been developed to assess mild behavioral impairment (MBI), a construct that characterises the emergence of sustained and impactful neuropsychiatric symptoms (NPS) in predementia populations as a precursor to cognitive decline and dementia. Given that the MBI-C is sensitive in detecting MBI in people with MCI [43], future studies of NPS in pre-dementia individuals should also include the MBI-C.

Conclusions:
Our ndings further support the emerging conceptual framework that NPS, characterised by agitation, disinhibition, irritability and depression, are early clinical presentations of AD pathophysiology progression. Given that early NPS may contribute to the characterization of the preclinical AD stage, cognitively intact individuals presenting with NPS can be identi ed earlier so as to allow a personalized and timely preventive intervention. The DIAN study was approved by the Institutional Review boards of all of the participating institutions.

Consent for publication
Informed written consent was obtained from all participants at each site. Figure 1 Higher NPI-Q and shorter EYO to onset of AD is associated with higher [18F]FDG uptake decline in DIAD mutation carriers Figure Legend: Statistical parametric map overlaid on a structural MRI scan shows regions in the PCC, vmPFC, bilateral parietal lobes and right AI where higher [18F]FDG uptake decline was found in cognitively normal DIAD mutation carriers with higher NPI-Q scores and shorter EYO to onset of AD. The analysis was corrected for gender, education, APOE ε4 status and family mutation type (APP, PSEN1 and PSEN2) and multiple comparisons were corrected using random eld theory at p < 0.001. AD: Alzheimer disease; AI: anterior insula;