Background

The role of emergent neuropsychiatric symptoms (NPS) as a risk factor for all-cause dementia is increasingly being recognized. Evidence links NPS in advance of dementia to a greater likelihood of cognitive impairment (Mortby et al., Reference Mortby, Burns, Eramudugolla, Ismail and Anstey2017) and cognitive decline (Geda et al., Reference Geda2014) compared to those without NPS. Affective and emotional dysregulation symptoms are among the most common NPS presenting in mild cognitive impairment (MCI) (Peters et al., Reference Peters2012), and are, along with other NPS, associated with poorer outcomes overall (Cerejeira et al., Reference Cerejeira, Lagarto and Mukaetova-Ladinska2012) and higher conversion rates from MCI to dementia (Palmer et al., Reference Palmer, Berger, Monastero, Winblad, Backman and Fratiglioni2007).

While the early presentation of NPS in the course of neurodegenerative disease has traditionally raised suspicion of behavioral variant frontotemporal dementia (bvFTD), evidence also suggests that NPS can emerge in advance of any dementia syndrome (Taragano et al., Reference Taragano, Allegri, Krupitzki, Sarasola, Serrano and Lyketsos2009). Historically, these symptoms have been viewed through the lens of psychiatric nosology with insufficient attention given to their late life onset and consequently their frequent misclassification as idiopathic psychiatric illness (Woolley et al., Reference Woolley, Khan, Murthy, Miller and Rankin2011), potentially exposing patients to inappropriate medications or delays in dementia diagnosis (Jalal et al., Reference Jalal, Ganesh, Lau, Lysack and Ismail2014). Cognitive patterns alone may not adequately differentiate between psychiatric and neurodegenerative disease (Ting et al., Reference Ting2010).

Mild Behavioral Impairment (MBI) is characterized by later life acquired, sustained and impactful NPS of any severity that cannot be better accounted for by other formal medical and psychiatric nosology. MBI is an “at risk” state for incident cognitive decline and dementia, and for some, MBI is the index manifestation of neurodegeneration, observed in advance of cognitive impairment (Ismail et al., Reference Ismail2016). Importantly, MBI distinguishes between chronic psychiatric symptomatology and formal psychiatric illness, versus new onset psychiatric symptoms in older adults, the latter of which are core to the MBI construct of the at-risk state. MBI has been formally described in the International Society to Advance Alzheimer's Research and Treatment – Alzheimer's Association (ISTAART-AA), MBI proposed research diagnostic criteria, which classify MBI into the domains of decreased drive/motivation, affective/emotional dysregulation, impulse dyscontrol and agitation, social inappropriateness, and delusions and hallucinations (Ismail et al., Reference Ismail2016). Assessment of MBI has been operationalized with the development of the MBI checklist (MBI-C), which was tailored specifically to the MBI criteria, including explicit observations of symptoms being later life in onset, and sustained for 6 months – these requirements are not explicit in many NPS rating scales (Ismail et al., Reference Ismail2017a). The affective/emotional dysregulation domain is a core feature of MBI, and includes symptoms of anxiety, depression and dysphoria, euphoria, and irritability. In this review, we will explore the epidemiology and neurobiological links between affective and emotional symptoms and incident cognitive impairment and all-cause dementia.

Depression and dysphoria

Increasing evidence in both cognitively normal (CN) older adults (without objective cognitive impairment) and in MCI suggests that emergent depression is a risk factor for dementia (Rosenberg et al., Reference Rosenberg, Mielke, Appleby, Oh, Geda and Lyketsos2013) and may constitute a prodrome (Ismail et al., Reference Ismail, Malick, Smith, Schweizer and Fischer2014). This is true for both Alzheimer's disease (AD) dementia and all-cause dementia (Barnes et al., Reference Barnes, Yaffe, Byers, McCormick, Schaefer and Whitmer2012; Steenland et al., Reference Steenland, Karnes, Seals, Carnevale, Hermida and Levey2012; Geda et al., Reference Geda2014). In a meta-analysis of 57 MCI studies, depression was found to be common, with a prevalence of 25% in community samples and 40% in clinical samples (Ismail et al., Reference Ismail2017b), and associated with increased risk of progression to AD dementia (Hermida et al., Reference Hermida, McDonald, Steenland and Levey2012; Lopez-Anton et al., Reference Lopez-Anton2015). Another recent study assessed the association of subsyndromal depression (SSD) with cognitive decline over four years in a sample of MCI patients from the Alzheimer's Disease Neuroimaging Initiative (ADNI). Compared with individuals without depressive symptoms, the SSD group exhibited accelerated decline on cognitive measures (Gonzales et al., Reference Gonzales2017). This suggests that in early stages of AD, depressive symptoms may be clinical indicators of AD pathology and/or impact AD pathogenesis (Steenland et al., Reference Steenland, Karnes, Seals, Carnevale, Hermida and Levey2012; Donovan et al., Reference Donovan2014; Geda et al., Reference Geda2014; Gonzales et al., Reference Gonzales2017).

Very recent research has moved the field rapidly forward, and underscored the importance of incipient depressive symptoms in emergence of dementia syndromes. A 28-year follow up of a 10,189 person cohort from the United Kingdom described the emergence of sustained depressive symptoms to increase the risk of incident dementia, but longstanding sustained depressive symptoms offered no increased risk (Singh-Manoux et al., Reference Singh-Manoux2017). A 14-year longitudinal Australian study of 4,922 cognitively intact men demonstrated that depression is more likely to be a marker of incipient dementia than a truly modifiable risk factor due to a time-dependent link between onset of depression and incidence of dementia. The association between depression and dementia was only apparent during the initial five years of follow-up, and was not observed if depression emerged more than five years before the onset of dementia (Almeida et al., Reference Almeida, Hankey, Yeap, Golledge and Flicker2017). Similarly, a Finnish nationwide nested case-control study of 27,948 pairs also established the importance of the time window between psychiatric symptoms and dementia. Hospital treated behavioral and mental disorders, including depression and other mood disorders, were associated with a higher risk of AD in the five-year but not the ten-year time window (Tapiainen et al., Reference Tapiainen, Hartikainen, Taipale, Tiihonen and Tolppanen2017). The authors suggested that some of these disorders may have represented misdiagnosed prodromal symptoms of AD. This highlights the importance of proper differential diagnostics among older persons and the importance of appropriate time window in psychiatric and neuroepidemiology research (Tapiainen et al., Reference Tapiainen, Hartikainen, Taipale, Tiihonen and Tolppanen2017).

Despite converging evidence for the relationship between depression and dementia, the neurobiology of depressive symptoms in preclinical dementia syndromes has not been clearly established. According to one hypothesis, which enjoys some empirical support, clinically significant symptoms of major depressive disorder and subclinical depressive symptoms, in preclinical and prodromal AD, may indicate AD pathological changes in selectively vulnerable brain regions. In CN older adults, clinically significant depressive symptoms have been associated with AD-related changes: thinning of the entorhinal cortex, hippocampal volume reduction, and decreased cerebrospinal fluid (CSF) amyloid-beta 42 (Ballmaier et al., Reference Ballmaier2008; Gerritsen et al., Reference Gerritsen, Comijs, van der Graaf, Knoops, Penninx and Geerlings2011; Pomara et al., Reference Pomara2012). A cross-sectional study in late life major depression found reduced hippocampal volume in depressed versus control participants but no differences in cortical amyloid (measured in vivo by positron emission tomography (PET) imaging) (De Winter et al., Reference De Winter2016). In the aforementioned ADNI sample of participants with SSD and MCI, frontal and anterior cingulate atrophy was associated with global cognitive decline. These regions were postulated to govern cognitive decline in this highly vulnerable population (Gonzales et al., Reference Gonzales2017).

Studies of subclinical depressive symptoms in CN older adults and across the AD spectrum have similarly shown associations with underlying structural and functional changes and more variably, with AD proteinopathies, A-beta and tau (Donovan et al., Reference Donovan2015; Babulal et al., Reference Babulal2016; Krell-Roesch et al., Reference Krell-Roesch2016; McCutcheon et al., Reference McCutcheon2016; Gatchel et al., Reference Gatchel2017). Donovan and colleagues, in a cross-sectional study of CN older adults, found that subclinical depressive symptoms on the Geriatric depression scale (GDS) were associated with lower hippocampal volume, while GDS symptom clusters grouping with dysphoria, apathy, and anhedonia were associated with decreased hippocampal volume and reduced cerebral 18F-fluorodeoxyglucose (FDG) metabolism (for apathy-anhedonia symptoms); these associations were independent of cortical amyloid burden (Donovan et al., Reference Donovan2015). Similarly, Babulal and colleagues, assessing subclinical positive and negative affective symptoms by the GDS and other measures, found a longitudinal but not cross-sectional association between GDS and CSF and amyloid (Pittsburgh Compound B (PiB)) PET AD biomarkers (Babulal et al., Reference Babulal2016). In contrast, clinico-pathologic studies in mixed populations of CN and AD participants (Wilson et al., Reference Wilson, Schneider, Bienias, Arnold, Evans and Bennett2003, Reference Wilson2014) and in MCI and mild AD participants (McCutcheon et al., Reference McCutcheon2016) have not shown robust relationships between depressive symptoms and amyloid plaques or neurofibrillary tangles (Wilson et al., Reference Wilson, Schneider, Bienias, Arnold, Evans and Bennett2003; Wilson et al., Reference Wilson2014; McCutcheon et al., Reference McCutcheon2016). Together, these findings suggest that the neurobiology of depressive symptoms may differ based on depressive symptom severity and disease stage, and may not be mediated solely by AD proteinopathies.

In addition to AD, converging evidence supports depression as a risk factor and/or prodrome for vascular dementia (Barnes et al., Reference Barnes, Yaffe, Byers, McCormick, Schaefer and Whitmer2012; Lin et al., Reference Lin, Hu, Tsai, Yang and Shen2017). Depressive symptoms may be more common in vascular dementia than in AD dementia (O'Brien, Reference O'Brien2003), and depression has been associated with amnestic (Steenland et al., Reference Steenland, Karnes, Seals, Carnevale, Hermida and Levey2012; Geda et al., Reference Geda2014) as well as both amnestic and non-amnestic MCI (Hermida et al., Reference Hermida, McDonald, Steenland and Levey2012). The robust relationship between cerebrovascular disease and depressive symptoms may underlie this association (Alexopoulos et al., Reference Alexopoulos1997; Krishnan et al., Reference Krishnan, Hays and Blazer1997). In particular, disruption of fronto-subcortical circuits, whether by vascular lesions or in FTD, is one mechanism hypothesized to underlie the association between depression and vascular and FTD pre-dementia syndromes.

Anxiety

There has been a growing body of research supporting an association between anxiety and reduced cognitive function in older adults, indicating clinically elevated anxiety symptoms to be associated with poorer global cognition, episodic memory, and executive functioning (Beaudreau and O'Hara, Reference Beaudreau and O'Hara2008). Anxiety symptoms are commonly associated with neurodegenerative diseases, and prevalence estimates range between 8% and 71% (Seignourel et al., Reference Seignourel, Kunik, Snow, Wilson and Stanley2008). Clinically elevated anxiety has also been linked to cognitive decline in older adults (Sinoff and Werner, Reference Sinoff and Werner2003; DeLuca et al., Reference DeLuca2005; Gallacher et al., Reference Gallacher2009), leading some to conclude that anxiety symptoms may predict cognitive decline in older adults without dementia (Pietrzak et al., Reference Pietrzak2012). Anxiety symptoms are common in MCI (Apostolova and Cummings, Reference Apostolova and Cummings2008; Spalletta et al., Reference Spalletta, Musicco, Padovani, Rozzini, Perri and Fadda2010), and co-morbid presentation of anxiety in MCI has been shown to increase the risk of progressing to AD (Palmer et al., Reference Palmer, Berger, Monastero, Winblad, Backman and Fratiglioni2007; Palmer et al., Reference Palmer2010; Rabins et al., Reference Rabins2013; Mah et al., Reference Mah, Binns and Steffens2015).

Systematic reviews and meta-analyses are less conclusive, however, with some providing support for anxiety as a risk factor for dementia (Gulpers et al., Reference Gulpers, Ramakers, Hamel, Kohler, Oude Voshaar and Verhey2016), while others provide mixed results (Cooper et al., Reference Cooper, Sommerlad, Lyketsos and Livingston2015). For instance, Gulpers et al. (Reference Gulpers, Ramakers, Hamel, Kohler, Oude Voshaar and Verhey2016) conducted a systematic review and meta-analysis of 20 studies demonstrating anxiety predicts incident cognitive impairment (relative risk [RR]: 1.77, 95% CI: 1.02–2.42, z = 2.05, p = 0.040) in the community, with greater age being a driver of results, suggesting anxiety to be a prodromal symptom. However, co-morbid anxiety in subject with MCI seen in the clinical setting did not predict conversion to dementia in this study (RR: 1.21, 95% CI: 0.90–1.63, z = 1.28, p = 0.200) (Gulpers et al., Reference Gulpers, Ramakers, Hamel, Kohler, Oude Voshaar and Verhey2016). Conversely, Cooper et al. (Reference Cooper, Sommerlad, Lyketsos and Livingston2015) provided mixed results, identifying only one higher quality epidemiological study that showed anxiety to predict AD (Palmer et al., Reference Palmer, Berger, Monastero, Winblad, Backman and Fratiglioni2007), while in three clinical studies (Robert et al., Reference Robert2006; Rozzini et al., Reference Rozzini2007; Devier et al., Reference Devier2009) anxiety was found not to predict conversion (pooled odds ratio [OR] from clinical studies: −0.11, −0.34 to 0.11). Nonetheless, a recent meta-analysis by Li and Li (Reference Li and Li2017) of 11 studies showed a pooled hazard ratio of conversion to dementia for co-morbid anxiety and MCI compared to those without anxiety (hazard ratio (HR): 1.18, 95% CI: 1.07–1.31, p = 0.002). Heterogeneity was substantial in these studies, and confounding the results is the fact that rating scales used to assess anxiety did not reliably distinguish between anxiety as part of a psychiatric illness, or anxiety as a new-onset phenomenon, with little information to determine whether psychiatric symptoms were brief and episodic, or sustained symptoms representing a change from baseline state.

While the manifestations of anxiety in prodromal patients are still under investigation, worried appearance, fearfulness, tension, restlessness and fidgeting are all common manifestations of anxiety in people with dementia (Ferretti et al., Reference Ferretti, McCurry, Logsdon, Gibbons and Teri2001). What remains unclear is the degree to which anxiety symptoms reflect a psychopathological reaction to cognitive and functional decline, or whether they arise from biological changes in emotion-relevant neural circuits (Levenson et al., Reference Levenson, Sturm and Haase2014). Levenson et al. (Reference Levenson, Sturm and Haase2014) conclude that there is strong evidence to support the latter, based on findings of heightened intrinsic connectivity in the salience networks in AD (Balthazar et al., Reference Balthazar2013) and hypo-metabolism in the medial temporal lobe, superior temporal gyrus and insula (Hashimoto et al., Reference Hashimoto2006), both which have been linked to increases in anxiety. Further evidence comes from the National Alzheimer Coordinating Center (NACC) registry in a study of 2,416 CN participants over the age of 50 assessing NPS and conversion from a Clinical Dementia Rating Scale (CDR) of 0 to >0 (Masters et al., Reference Masters, Morris and Roe2015). The Hazard Ratio of conversion in participants with anxiety was 2.83 (CI: 1.93–3.19) compared to those without anxiety. Additionally, the authors described a three-phase progression of NPS within those with cognitive decline: first, irritability, depression, and nighttime behavior changes; next, anxiety, appetite changes, agitation, and apathy; and last, elation, motor disturbances, hallucinations, delusions, and disinhibition, suggesting anxiety to be an emerging phenomenon with progressive neurodegeneration (Masters et al., Reference Masters, Morris and Roe2015). Findings from the ADNI have further provided mechanistic evidence linking NPI-Q measured anxiety symptoms in amnestic MCI and conversion to AD (Mah et al., Reference Mah, Binns and Steffens2015). Mah et al. (Reference Mah, Binns and Steffens2015) found that anxiety, over and beyond depression or memory decline predicted a greater decline in entorhinal cortex volume. However, it was not unequivocally determined in this study whether anxiety symptoms were causative, mediating or a consequence of neurodegeneration.

A number of potential hypotheses for a causal pathway between anxiety and cognitive impairment have been described including: (1) hypercortisolism, (2) cardiovascular disease, (3) low-grade inflammation, (4) brain-derived neurotrophic factor suppression, and (5) depletion of cognitive reserves (as reviewed in Gulpers et al. (Reference Gulpers, Ramakers, Hamel, Kohler, Oude Voshaar and Verhey2016)). According to the hypercortisolism hypothesis, anxiety is a state of heightened stress and thus higher cortisol levels. Cortisol levels are in turn related to symptom severity and poorer outcomes on neuropsychological tests (Erickson et al., Reference Erickson, Drevets and Schulkin2003; Leininger and Skeel, Reference Leininger and Skeel2012; Mantella et al. Reference Mantella2008; Rosnick et al., Reference Rosnick, Rawson, Butters and Lenze2013). Further, cortisol-induced overstimulation of glucocorticoid receptors in the medial temporal lobe have also been linked to hippocampal atrophy (Sapolsky, Reference Sapolsky2000; Erickson et al., Reference Erickson, Drevets and Schulkin2003), and stress has been postulated as a mechanism for hippocampal injury and volume loss (Sheline et al., Reference Sheline, Sanghavi, Mintun and Gado1999). Animal studies have shown higher cortisol administration levels to increase amyloid and tau accumulation, both processes involved in AD pathology (Green et al., Reference Green, Billings, Roozendaal, McGaugh and LaFerla2006). Anxiety has also been linked to coronary artery disease and stroke, with anxiety implicated in the triggering of physiological reactions (e.g. increased heart rate, blood pressure, vasoconstriction, and platelet activity) that are associated with cardiovascular disease and ultimately vascular dementia (Sheps and Sheffield, Reference Sheps and Sheffield2001; Lambiase et al., Reference Lambiase, Kubzansky and Thurston2014; Batelaan et al., Reference Batelaan, Seldenrijk, Bot, van Balkom and Penninx2016). Chronic low-level inflammation is another related causal hypothesis, in which elevated cytokines such as interleukin-6 and tumor necrosis factor are observed in stress-related states such as anxiety, and in turn shown to be linked to negative effects on cognitive functioning (Furtado and Katzman (Reference Furtado and Katzman2015); Reichenberg et al., Reference Reichenberg2001). Furthermore, anxiety disorders have also been linked to decreased brain derived neurotrophic factor (BDNF) levels or polymorphisms (Domingos da Silveira da Luz et al., Reference Domingos da Silveira da Luz2013). BDNF is needed for synaptic plasticity, learning and memory, and neuronal repair and is decreased in AD and MCI (Teixeira et al., Reference Teixeira, Barbosa, Diniz and Kummer2010). BDNF signaling is necessary for antidepressant response, and BDNF deletion attenuates this response (Castrén and Kojima, Reference Castrén and Kojima2017). The Framingham Heart Study identified higher stratified BDNF levels as a protective factor for dementia, emphasizing its importance as a marker and target (Weinstein et al., Reference Weinstein2014). Finally, it has been proposed that chronic and recurrent anxiety disorders throughout the life-course may lead to avoidance behaviors, which in turn may result in lowered cognitive reserve as a result of less mental and social stimulation, therefore increasing the risk of dementia (Stern, Reference Stern2012).

Elation, euphoria, and mania

A constellation of elation, euphoria, and mania can represent a pre-dementia syndrome. “Secondary mania,” mania due to an etiology other than bipolar disorder, causes the majority of new onset mania in older adults (Krauthammer and Klerman, Reference Krauthammer and Klerman1978; Brooks and Hoblyn, Reference Brooks and Hoblyn2005), which include neurodegenerative disorders, such as pre-dementia syndromes and dementias (Krauthammer and Klerman, Reference Krauthammer and Klerman1978; Brooks and Hoblyn, Reference Brooks and Hoblyn2005).

Evidence from cohort studies as well as retrospective data suggests that those with late onset mania are at an increased risk for development of cognitive impairment and dementia. Patients who develop their first manic episode after age 58 have higher rates of cognitive impairment, which is variably reversible (Young and Klerman, Reference Young and Klerman1992; Brooks and Hoblyn, Reference Brooks and Hoblyn2005). A large cohort study of 37,768 men aged 65–85 years with bipolar disorder had an adjusted hazard ratio of 2.3 (HR = 2.30, 95% CI 1.80–2.94) for development of dementia (Almeida et al., Reference Almeida, McCaul, Hankey, Yeap, Golledge and Flicker2016). Subgroups, who developed bipolar disorder over the age of 70 or within the past five years had the greatest risk for development of dementia (Almeida et al., Reference Almeida, McCaul, Hankey, Yeap, Golledge and Flicker2016), very similar to the evidence for depressive symptoms (Almeida et al., Reference Almeida, Hankey, Yeap, Golledge and Flicker2017). A retrospective case series of patients over the age of 65 admitted for mania, found 22 of 92 (24%) had evidence of cerebral organic impairment (Stone, Reference Stone1989; Almeida et al., Reference Almeida, McCaul, Hankey, Yeap, Golledge and Flicker2016). Another case series showed that 8 of 25 (32%) older adults with mania went on to develop cognitive impairment within 5–7 years (Dhingra and Rabins, Reference Dhingra and Rabins1991).

CN individuals with symptoms of elation may be at elevated risk of developing AD. A survival analysis of 11,453 CN individuals from the NACC database showed that positive ε4 carrier status and symptoms of elation, among other NPS, conferred greater risk of development of AD (Burke et al., Reference Burke, Maramaldi, Cadet and Kukull2016). Another NACC database study showed those with amnestic MCI had higher rates of elation and aggression as compared to non-amnestic MCI (Apostolova et al., Reference Apostolova2014). However, symptoms of elation, euphoria, and mania are relatively rare in AD type dementia as compared to FTD, with an estimated prevalence of 2.2–3.5% (Burns, Reference Burns1992; Lyketsos et al., Reference Lyketsos, Corazzini and Steele1995).

The pathophysiology of late onset bipolar disorder is thought to be different from early onset bipolar disorder with the late onset group having higher rates of neurologic and vascular brain abnormalities (Young and Klerman, Reference Young and Klerman1992; Tohen et al., Reference Tohen, Shulman and Satlin1994; Fujikawa et al., Reference Fujikawa, Yamawaki and Touhouda1995; Sami et al., Reference Sami, Khan and Nilforooshan2015). Vascular mania has been suggested as a subtype of secondary mania, when it occurs in the presence of co-morbid cerebrovascular disease or cognitive impairment (Steffens and Krishnan, Reference Steffens and Krishnan1998; Sajatovic et al., Reference Sajatovic2015). The association of cerebrovascular disease with late onset mania implies a possible prodrome to vascular dementia. Lesions in a number of different brain regions have been associated with the development of secondary mania, including bilateral orbitofrontal, right temporoparietal, right basal and medial temporal lobe, basal ganglia, thalamic and right frontotemporal (Bakchine et al., Reference Bakchine, Lacomblez, Benoit, Parisot, Chain and Lhermitte1989; Danel et al., Reference Danel1989; Turecki et al., Reference Turecki, Mari Jde and Del Porto1993; Robinson, Reference Robinson1997; Starkstein and Robinson, Reference Starkstein and Robinson1997; Brooks and Hoblyn, Reference Brooks and Hoblyn2005; Cerami et al., Reference Cerami, Marcone, Galimberti, Villa, Scarpini and Cappa2011). A case control series by Ramirez-Bermudez et al. (Reference Ramirez-Bermudez2016) showed that when compared to healthy normal adults, a group with first onset mania after age 50 had higher rates of white matter hyperintensities in the right frontal and the left temporal brain regions. A number of scientists hypothesize that damage to the right orbitofrontal region is a unifying finding in the development of secondary mania (Starkstein and Robinson, Reference Starkstein and Robinson1997; Brooks and Hoblyn, Reference Brooks and Hoblyn2005).

Disruption to orbitofrontal neurocircuitry appears common to both to secondary mania and bvFTD. More recently, two cases of sporadic FTLD with the same progranulin mutation and one case of FTLD with a C9ORF72 gene hexanucleotide expansion mutation developed symptomatology meeting criteria for a bipolar disorder diagnosis prior to clinical presentation of Frontotemporal lobar degeneration (FTLD) (Cerami et al., Reference Cerami, Marcone, Galimberti, Villa, Scarpini and Cappa2011; Floris et al., Reference Floris2013). There have also been case reports of bipolar disorder mimicking BvFTD, with failure to progress to “probable” BvFTD after 3–7 years of observation (Dols et al., Reference Dols, Krudop, Moller, Shulman, Sajatovic and Pijnenburg2016).

Irritability

Irritability is common among neurobehavioral symptoms in neurodegenerative disease and causes increased burden on patients, caregivers, and the community alike (Sousa et al., Reference Sousa, Santos, Turro-Garriga, Dias, Dourado and Conde-Sala2016). Among cognitively impaired individuals without dementia in the Cache County cohort, irritability/lability was, among all NPI domains, second only to depression/dysphoria in prevalence (Peters et al., Reference Peters2013). Irritability was the third most prevalent NPS regardless of dementia in a population-based longitudinal study of aging conducted in England and Wales, with sleep problems more prevalent in people without dementia and apathy more prevalent in people with dementia (Savva et al., Reference Savva2009). It has now been shown that irritability is among the neurobehavioral symptoms which portends more rapid decline in several datasets including NACC (Leoutsakos et al., Reference Leoutsakos, Forrester, Lyketsos and Smith2015; Forrester et al., Reference Forrester, Gallo, Smith and Leoutsakos2016); the Spanish ZARADEMP cohort (Lobo et al., Reference Lobo2008) and the Mayo Clinic Study of aging (Geda et al., Reference Geda2014). Hence, as with other emergent behavioral symptoms, monitoring emergent irritability in the clinic would seem of the utmost importance.

Whether emergent irritability observed in the dementia prodrome should be regarded as a manifestation of a mood disturbance or one of behavioral dysregulation/disinhibition is unclear phenomenologically as there is support for both possibilities from cluster analyses (Leoutsakos et al., Reference Leoutsakos, Forrester, Lyketsos and Smith2015). It might be hoped that studies of the neurobiology of irritability would ultimately serve to clarify this question thereby offering improved treatment strategies. However, such studies have suffered from this ambiguous phenomenology as well. Disinhibition together with emotional lability in dementia has been associated with orbitofrontal-subcortical circuit dysfunction (Tascone and Bottino, Reference Tascone and Bottino2013). The same can be said when irritability falls within agitation in AD. i.e. even with AD, associations have been made with deficits in frontal, anterior cingulate, and posterior cingulate cortices as well as the amygdala and hippocampus (Rosenberg et al., Reference Rosenberg, Nowrangi and Lyketsos2015). A specific association between irritability on the NPI scale has been made with lower fractional anisotropy of the anterior cingulate in a sample of MCI and AD patients (Tighe et al., Reference Tighe2012). On the other hand, when irritability is grouped with affective symptoms (depression) or psychosis, no clear evidence for specific neurobiological substrates emerges (Tascone and Bottino, Reference Tascone and Bottino2013). Recently, however, an FDG-PET study in AD demonstrated that irritability had common metabolic changes to agitation (right temporal, right frontal, bilateral middle, and posterior cingulate gyri) but differed in specific regions (right insular, precentral, and postcentral gyri) (Weissberger et al., Reference Weissberger, Melrose, Narvaez, Harwood, Mandelkern and Sultzer2017), reflecting neurodegeneration in regions associated with core AD pathology (Rosenberg, Reference Rosenberg2017). It is noteworthy that to date, most of the work on the neurobiology of irritability comes from the dementia literature with little data addressing emergent irritability in the predementia stages in spite of its prevalence. That said, a very recent ADNI study of biomarker positive preclinical AD participants (amyloid and tau positive, MMSE ~29,) assessed relationships between NPS and two-year interval change in metabolism measured by FDG-PET. Sleep behavior and irritability predicted posterior cingulate hypometabolism at two years, “supporting the emerging conceptual framework in which NPS constitute an early clinical manifestation of AD pathophysiology” (Ng et al., Reference Ng2017).

Conflict of interest declaration

Z. Ismail has received consultation/advisory board funding from Eli Lilly and Merck. M. Cantillon has had research support from Pfizer, Lilly, Lundbeck, Takeda, Novartis, Impax, Prevacus, CogRx, and the Critical Path Institute that includes FDA funding.

Description of authors’ roles

All authors contributed to manuscript preparation and revisions.