A different syndrome of depression in dementia?

Depression is a heterogeneous and aetiologically complex syndrome. There are at least 256 possible unique symptom profiles that meet DSM-V criteria for a diagnosis of major depressive disorder (Buch & Liston, Reference Buch and Liston2021). This degree of clinical heterogeneity has led to efforts to define subtypes based on symptom profiles (Buch & Liston, Reference Buch and Liston2021). The reliability of different depression subtypes has not been established in the general population, largely due to transition between symptom profiles across and within depressive episodes (Lamers et al., Reference Lamers, Rhebergen, Merikangas, de Jonge, Beekman and Penninx2012). However, this has led to studies highlighting how specific symptoms of depression respond differently to interventions and may predict overall response to treatment (Buch & Liston, Reference Buch and Liston2021). For example, network analysis of individual depressive symptoms in antidepressant RCTs has identified improvement in low mood and insomnia as having greatest impact on subsequent remission and recovery (Komulainen et al., Reference Komulainen, Airaksinen, Savelieva, Gluschkoff, García Velázquez, Elovainio and Jokela2021).

Few studies have investigated how depressive symptoms differ across dementia types or in comparison with the non-dementia depression (Ehrt, Brønnick, Leentjens, Larsen, & Aarsland, Reference Ehrt, Brønnick, Leentjens, Larsen and Aarsland2006; Riedel, Heuser, Klotsche, Dodel, & Wittchen, Reference Riedel, Heuser, Klotsche, Dodel and Wittchen2010; Verdelho, Hénon, Lebert, Pasquier, & Leys, Reference Verdelho, Hénon, Lebert, Pasquier and Leys2004). However, the clinical complexity of making a distinction between a diagnosis of depression from some of the other behavioural and psychological symptoms of dementia (BPSD) can be challenging as shown by the overlap between depression rating scales with other symptoms captured by BPSD measures (See Table 1). This raises the question of whether antidepressants are ineffective because they have been used to treat different syndromes.

Table 1. Depressive symptoms across seven depression rating scales that are also captured by measures of other BPSD symptoms (not depression items) on four BPSD measures

NPI, Neuropsychiatric Inventory (Cummings et al., Reference Cummings, Mega, Gray, Rosenberg-Thompson, Carusi and Gornbein1994), Behave-AD, Behavioural Pathology in Alzheimer's Disease (Reisberg, Auer, & Monteiro, Reference Reisberg, Auer and Monteiro1997), CUSPAD, Columbia University Scale for Psychopathology in Alzheimer's disease (Devanand et al., Reference Devanand, Miller, Richards, Marder, Bell, Mayeux and Stern1992), MOUSEPAD, Manchester & Oxford University Scale for the Psychopathological Assessment of Dementia (Allen, Gordon, Hope, & Burns, Reference Allen, Gordon, Hope and Burns1996).

Coloured circles for a symptom indicate that a scale directly assesses that symptom, whereas empty circles indicate that a scale indirectly measures a symptom. The seven depression rating scales are: Beck Depression Inventory (Beck, Steer, Ball, & Ranieri, Reference Beck, Steer, Ball and Ranieri1996); Centre of Epidemiological Studies Depression Scale; Hamilton Rating Scale for Depression (Radloff, Reference Radloff1977); Inventory of Depressive Symptoms (Rush, Gullion, Basco, Jarrett, & Trivedi, Reference Rush, Gullion, Basco, Jarrett and Trivedi1996); Montgomery–Åsberg Depression Rating Scale (Montgomery & Åsberg, Reference Montgomery and Åsberg1979); Quick Inventory of Depressive Symptoms (Rush et al., Reference Rush, Trivedi, Ibrahim, Carmody, Arnow, Klein and Keller2003); Zung Self-Rating Depression Scale (Zung, Reference Zung1965). List of depressive symptoms adapted from ref – (Fried, Flake, & Robinaugh, Reference Fried, Flake and Robinaugh2022).

Studies of depressive symptom profile in dementia in long-term care facilities highlight how motivational symptoms such as apathy are common (Janzing, Hooijer, van ‘t Hof, & Zitman, Reference Janzing, Hooijer, van ‘t Hof and Zitman2002), and clinician-rated irritability is the most prevalent depressive symptom (Borza et al., Reference Borza, Engedal, Bergh, Barca, Benth and Selbæk2015). However, both apathy and irritability are common in dementia in the absence of depression, and highly influenced by environmental factors (Jao, Liu, Williams, Chaudhury, & Parajuli, Reference Jao, Liu, Williams, Chaudhury and Parajuli2019).

Apathy affects 49% (Zhao et al., Reference Zhao, Tan, Wang, Jiang, Tan, Tan and Yu2016) and 72% (Chow et al., Reference Chow, Binns, Cummings, Lam, Black, Miller and van Reekum2009) of Alzheimer's disease (AD) and FTD patients respectively, markedly higher than that reported in older adults with depression without dementia (38%) (Yuen et al., Reference Yuen, Bhutani, Lucas, Gunning, AbdelMalak, Seirup and Alexopoulos2015). Lesion studies have identified that the functional anatomy of apathy is associated with damage to fronto-striatal circuits (Levy & Dubois, Reference Levy and Dubois2006). Apathy also shares underlying neural and cognitive mechanisms with anhedonia, a core symptom of depression primarily characterised by a loss of interest in engaging in pleasurable activities. Motivational symptoms including apathy and anhedonia contribute to an ‘interest-activity’ cluster of depressive symptoms that predicts worse response to antidepressants in depression in the general population (Uher et al., Reference Uher, Perlis, Henigsberg, Zobel, Rietschel, Mors and McGuffin2012), but respond better than core mood symptoms to behavioural activation (Leontjevas et al., Reference Leontjevas, Teerenstra, Smalbrugge, Vernooij-Dassen, Bohlmeijer, Gerritsen and Koopmans2013). The presence of apathy in late-onset depression in mild cognitive impairment (MCI) also predicts subsequent development of AD (Ruthirakuhan, Herrmann, Vieira, Gallagher, & Lanctôt, Reference Ruthirakuhan, Herrmann, Vieira, Gallagher and Lanctôt2019). This indicates that apathy is not only a common symptom in depression in dementia, but is central to the presentation of depression as a prodromal stage of dementia and poor response to antidepressant treatment.

Agitation is common in dementia affecting half of patients (Livingston et al., Reference Livingston, Barber, Marston, Rapaport, Livingston, Cousins and Cooper2017), and is defined as inappropriate verbal, vocal, or motor activity that is not thought to be caused by an unmet need (Cummings et al., Reference Cummings, Mintzer, Brodaty, Sano, Banerjee, Devanand and Zhong2015). Irritability is a key component of agitation in dementia and strongly associated with depressive symptoms, both of which worsen with dementia progression (Volicer, Frijters, & Van der Steen, Reference Volicer, Frijters and Van der Steen2012). Though agitation is seen across psychiatric disorders, there is a clinical distinction between a patient with dementia and agitation consisting of wandering due to disorientation and a patient with agitated depression (Sampogna, Vecchio, Giallonardo, Luciano, & Fiorillo, Reference Sampogna, Vecchio, Giallonardo, Luciano and Fiorillo2020). It is likely that the prevalence of irritability as one of the most common symptoms in depression in dementia is due to this overlap, with agitation secondary to cognitive decline, and is a different entity to the irritability that characterises depression in the general population.

Few studies have assessed whether the phenomenology of depression in dementia is qualitatively different from depression in other populations (Ehrt et al., Reference Ehrt, Brønnick, Leentjens, Larsen and Aarsland2006; Riedel et al., Reference Riedel, Heuser, Klotsche, Dodel and Wittchen2010). However, apathy and irritability are examples of how key depressive symptoms in dementia, which are captured by most symptom scales, are also common in the context of BPSD without apparent depressed mood. This highlights just how challenging it is for clinicians to reliably diagnose depression in dementia and raises the question as to whether antidepressants are ineffective due to their being used to treat a different syndrome that isn't ‘depression’.

Due to the differing patterns of neurodegeneration across dementia subtypes it would be anticipated that the underlying mechanisms of depression, evolution of symptoms and response to treatment would be different in each disorder. However, few studies have investigated whether depressive symptom profiles or response to treatment differs across neurodegenerative disorders or stages of disease. Future research investigating the network dynamics of depressive symptoms over time could reveal how changes in specific symptoms are related, and which symptoms have the greatest impact on clinical improvement and treatment response.

Further trials of medications that treat depression as homogenous across dementia subtypes and stage of disease are likely to be ineffective. An alternative approach to understanding depression in dementia is to view it as a complex system of interacting depressive symptoms, which differ across dementia subtypes and stages of disease. Utilising this approach would promote tailored interventions for specific symptoms with greater importance at different stages of disease, development of depressive-symptom-network markers for each condition to guide interventions and diagnosis, and provide some mechanistic insights into the causes of depression.

Neurodegeneration of monoamine pathways

The monoamine hypothesis of depression is over 50 years old (Coppen, Reference Coppen1967). It proposes that diminished activity in monoamine pathways (predominantly serotonin, noradrenaline and dopamine) plays a causal role in the development of depression. In the age of complex systems neurobiology and computational neuroscience, there is a consensus that such reductionist biochemical theories are over-simplifications (Cowen & Browning, Reference Cowen and Browning2015; Harmer, Duman, & Cowen, Reference Harmer, Duman and Cowen2017). However, given that almost all effective antidepressants directly target monoamine system receptors or transporters, predominantly serotonin, monoamine function likely plays a central role in the therapeutic effect of antidepressants.

The monoamine systems project from small brainstem nuclei to widespread brain regions (Sasaki et al., Reference Sasaki, Shibata, Tohyama, Kudo, Endoh, Otsuka and Sakai2008). Combined with an array of post-synaptic receptors, this anatomical distribution enables a wide variety of functions but is also susceptible to disruption from neurodegeneration (see Table 2). A potential reason for the lack of therapeutic effect of antidepressants in dementia would be disruption of the monoamine pathways needed to mediate their effect.

Table 2. Neurodegeneration of monoaminergic nuclei in dementia subtypes

Serotonin is the monoamine that has primarily been implicated in the treatment of depression and is a target for most antidepressants, in particular SSRIs, the most widely used class of medications. The most direct evidence for serotonin's role in the pathophysiology of depression is from tryptophan depletion studies, in which transient reduction in serotonin synthesis is achieved via dietary depletion of its precursor amino acid, tryptophan. In recovered patients who are taking antidepressant medication, tryptophan depletion elicits a rapid worsening of depressive symptoms. However, tryptophan depletion has relatively little effect in healthy controls, currently depressed patients or even recovered patients who are unmedicated (Ruhé, Mason, & Schene, Reference Ruhé, Mason and Schene2007).

The dorsal raphe nucleus (DRN) is one of two brain stem nuclei which are the primary sites of serotonergic innervation in the brain. The human DRN consists of ~ 250 000 neurons, 70% of which contain serotonin (Steinbusch, Dolatkhah, & Hopkins, Reference Steinbusch, Dolatkhah, Hopkins, Di Giovanni and De Deurwaerdere2021). Projections from the DRN are topographically organised rostrocaudally and innervate regions involved in mood regulation, motivation and memory, including the striatum, entorhinal cortex, hippocampus, pre-frontal cortex and amygdala (Steinbusch et al., Reference Steinbusch, Dolatkhah, Hopkins, Di Giovanni and De Deurwaerdere2021).

During the early stages of AD pathology (Braak stage I-II), the DRN is heavily affected by tau tangles (Rüb et al., Reference Rüb, Del Tredici, Schultz, Thal, Braak and Braak2000) and exhibits marked neurodegeneration, with estimates of up to 77% neuronal cells lost depending on disease stage (Lyness, Zarow, & Chui, Reference Lyness, Zarow and Chui2003). As a consequence of the degeneration in the DRN, lower cortical serotonin, serotonin metabolites (5-hydroxyindoleacetic acid) and reduced serotonin transporter levels (Smith et al., Reference Smith, Barrett, Joo, Nassery, Savonenko, Sodums and Workman2017) have been reported in AD and other dementia syndromes (van der Zande et al., Reference van der Zande, Joling, Happach, Vriend, Scheltens, Booij and Lemstra2019). Degeneration in serotonergic pathways has also been implicated in the development of BPSD, including depression (Lanctôt, Herrmann, & Mazzotta, Reference Lanctôt, Herrmann and Mazzotta2001; Vermeiren et al., Reference Vermeiren, Van Dam, Aerts, Engelborghs, Martin and De Deyn2015).

The noradrenergic system is a further target of antidepressants, including the serotonin–noradrenaline reuptake inhibitors venlafaxine and duloxetine. The locus coeruleus (LC), a small, pigmented nucleus located in the pons is the primary site for the synthesis of noradrenaline in the brain (Holland, Robbins, & Rowe, Reference Holland, Robbins and Rowe2021). The LC projects widely and is implicated in modulating attention, memory, cognitive flexibility and sleep (Holland et al., Reference Holland, Robbins and Rowe2021; Swift et al., Reference Swift, Gross, Frazer, Bauer, Clark, Vazey and Poe2018). Degeneration of the LC occurs early in AD and Parkinson's disease (PD) (Betts et al., Reference Betts, Kirilina, Otaduy, Ivanov, Acosta-Cabronero, Callaghan and Hämmerer2019). Neuronal death of up to 30% of LC neurons has been reported in MCI, increasing to 55% in diagnosed Alzheimer's dementia (Kelly et al., Reference Kelly, He, Perez, Ginsberg, Mufson and Counts2017). Depletion of LC neurons in AD is more closely correlated with disease severity and staging than cholinergic degeneration in the nucleus basalis of Meynert (Holland et al., Reference Holland, Robbins and Rowe2021; Kelly et al., Reference Kelly, He, Perez, Ginsberg, Mufson and Counts2017). LC degeneration has been implicated in agitation and irritability in AD (Liu, Stringer, Reeves, & Howard, Reference Liu, Stringer, Reeves and Howard2018) and depression in PD (Wang et al., Reference Wang, Li, Huang, Wan, Zhang, Wang and Jin2018). However, no study to date has examined whether LC integrity predicts antidepressant response in depression in dementia.

Human and animal studies have shown that dopaminergic transmission is crucial in motivated behaviour and reward processing (Le Heron, Apps, & Husain, Reference Le Heron, Apps and Husain2018). Dopaminergic neurons project from the ventral tegmental area (VTA) and substantia nigra (SN) to the striatum (Sasaki et al., Reference Sasaki, Shibata, Tohyama, Kudo, Endoh, Otsuka and Sakai2008). Dopamine function is closely related to other neuromodulators such as serotonin which it is co-released within the striatum (Zhou et al., Reference Zhou, Liang, Salas, Zhang, De Biasi and Dani2005). This interplay with other neurotransmitters is believed to have an important role in cognitive functions and decision-making processes (Wert-Carvajal, Reneaux, Tchumatchenko, & Clopath, Reference Wert-Carvajal, Reneaux, Tchumatchenko and Clopath2022) implicated in depression (see Table 3). Though not first-line and less commonly used, dopaminergic agents such as buproprion (a noradrenaline-dopamine reuptake inhibitor), and pramipexole (a dopamine partial agonist) are effective treatments in the management of depression (Papakostas, Reference Papakostas2006).

Table 3. Examples of neurocognitive affective bias reported in depression and dementia

Dopaminergic dysfunction is associated with motivational symptoms of depression including apathy and anhedonia which are poorly responsive to antidepressant treatment (Husain & Roiser, Reference Husain and Roiser2018; Le Heron et al., Reference Le Heron, Apps and Husain2018). Greater atrophy and loss of pre-synaptic dopaminergic projections to the striatum is associated with increased apathy and anhedonia across neurodegenerative disorders (Drui et al., Reference Drui, Carnicella, Carcenac, Favier, Bertrand, Boulet and Savasta2014; Le Heron et al., Reference Le Heron, Apps and Husain2018). VTA and SN neurodegeneration is most prominent in PD where up to 80% of dopaminergic projections to the striatum are lost by the time of diagnosis (Cheng, Ulane, & Burke, Reference Cheng, Ulane and Burke2010; Kish, Shannak, & Hornykiewicz, Reference Kish, Shannak and Hornykiewicz1988). This striking loss of dopaminergic neurons may explain why apathy and anhedonia are particularly common in PD, estimated to affect 40% (den Brok et al., Reference den Brok, van Dalen, van Gool, Moll van Charante, de Bie and Richard2015) and 46% (Lemke, Brecht, Koester, Kraus, & Reichmann, Reference Lemke, Brecht, Koester, Kraus and Reichmann2005) of patients, respectively. It may also explain why depressive symptoms in PD are poorly responsive to serotonergic and noradrenergic antidepressants (Troeung, Egan, & Gasson, Reference Troeung, Egan and Gasson2013) but can respond to dopaminergic agonists used to treat motor symptoms (van der Velden, Broen, Kuijf, & Leentjens, Reference van der Velden, Broen, Kuijf and Leentjens2018).

Whether involving serotonin, noradrenaline or dopamine, antidepressant mechanisms of action involve increasing monoamine levels in the synapse by preventing either metabolism or reuptake into the pre-synaptic neuron. However, as described above, there is significant degeneration of monoamine neurons in early dementia, leading to a loss of pre-synaptic innervation (see Table 2). Following the death of these neurons, monoamine release is markedly reduced, resulting in a loss of function in these systems which presumably reduces any impact of antidepressants.

In addition to loss of innervation, degeneration of pre-synaptic monoamine nuclei would also lead to loss of feedback inhibition. For example, the serotonin-1A (5-HT_1A) autoreceptor controls serotonergic tone through feedback inhibition and has been hypothesised to lead to a reduction in synaptic serotonin levels early on in SSRI treatment (Fritze, Spanagel, & Noori, Reference Fritze, Spanagel and Noori2017). The time taken for 5-HT_1A autoreceptors to desensitise has been theorised to be an explanation of the delayed clinical effects of antidepressant treatment, and differences in 5-HT_1A receptor levels or have been associated with depression, and the response to antidepressants (Richardson-Jones et al., Reference Richardson-Jones, Craige, Guiard, Stephen, Metzger, Kung and Leonardo2010). Though it is widely accepted that 5-HT_1A receptors participate in autoinhibition, their action is far more complex and the function of feedback is regionally organised and dependent on behavioural states (Altieri, Garcia-Garcia, Leonardo, & Andrews, Reference Altieri, Garcia-Garcia, Leonardo and Andrews2012). Changes in 5-HT_1A distribution and binding have been reported in dementia but vary depending on brain region and dementia subtype (Elliott et al., Reference Elliott, Ballard, Kalaria, Perry, Hortobágyi and Francis2009; Lai et al., Reference Lai, Tsang, Francis, Esiri, Keene, Hope and Chen2003). For example, increased 5-HT_1A binding in the temporal cortex has been reported in vascular dementia (Elliott et al., Reference Elliott, Ballard, Kalaria, Perry, Hortobágyi and Francis2009) but reduced binding in the same region has been found in AD which also correlated with higher agitation levels (Lai et al., Reference Lai, Tsang, Francis, Esiri, Keene, Hope and Chen2003).

Understanding the relative deficits in monoamine function across dementia subtypes and stages of disease could guide which class of antidepressant may be most effective. However, as described above the dynamics and function of monoamine pathways are diverse, complex, and interdependent. There is considerable synergy between monoamine systems, for example co-release of dopamine and serotonin in the striatum, and innervation of the cholinergic system by the LC (Kelly et al., Reference Kelly, He, Perez, Ginsberg, Mufson and Counts2017; Zhou et al., Reference Zhou, Liang, Salas, Zhang, De Biasi and Dani2005). Further research into the relationship between these systems and association with depressive symptoms would be needed to guide best use of existing therapies or development of novel interventions.

Changes in neurocognitive processing

The cognitive neuropsychological model of depression suggests a causal role for negative affective biases in the development and treatment of depression (Roiser, Elliott, & Sahakian, Reference Roiser, Elliott and Sahakian2012). Depression is associated with several types of negative affective biases (see Table 3) including changes in reward processing, negative perception of social cues, greater attention to aversive information and increased recall of negative relative to positive information (Harmer et al., Reference Harmer, Duman and Cowen2017).

There is substantial evidence that antidepressants alter emotion and reward processing (Harmer et al., Reference Harmer, Duman and Cowen2017; Roiser et al., Reference Roiser, Elliott and Sahakian2012), thereby normalising negative affective biases. The cognitive neuropsychological model proposes that this modulation of core psychological processes in depression drives antidepressant therapeutic effect, explains why antidepressants take 4–6 weeks to improve depressive symptoms, and why they are most effective in conjunction with psychological therapy (Harmer et al., Reference Harmer, Duman and Cowen2017).

Dementia is characterised by progressive cognitive decline and it is probable that the cognitive adaptations induced by antidepressants to normalise negative biases are not sustained due to these cognitive deficits. Neurodegeneration of core neural circuits involved in affective biases (in addition to the monoamine systems innervating them) may lead to further loss of antidepressant effects (see Table 2). For example, the reward processing circuit includes the ventral striatum, anterior cingulate cortex, amygdala and prefrontal cortex (Perry & Kramer, Reference Perry and Kramer2015). Serotonin co-release with dopamine in the ventral striatum is believed to be crucial in mediating the temporal dynamics of dopamine mediated reward prediction errors which drive reinforcement learning (Zhou et al., Reference Zhou, Liang, Salas, Zhang, De Biasi and Dani2005). Activation in the ventral striatum is lower in both apathy and anhedonia (Husain & Roiser, Reference Husain and Roiser2018), and is directly altered by both serotonergic and dopaminergic medications (Admon et al., Reference Admon, Kaiser, Dillon, Beltzer, Goer, Olson and Pizzagalli2017; Stoy et al., Reference Stoy, Schlagenhauf, Sterzer, Bermpohl, Hägele, Suchotzki and Ströhle2012).

Degeneration of dopaminergic projections to the striatum is a core pathological feature of PD and LBD, with up to 80% of dopaminergic projections to the dorsal striatum lost by the time of diagnosis in PD (see Table 2) (Cheng et al., Reference Cheng, Ulane and Burke2010; Kish et al., Reference Kish, Shannak and Hornykiewicz1988). SSRIs such as fluoxetine bind at therapeutic doses in the ventral striatum and modify emotional valence and reward processing via changes in tonic and phasic activities of striatal neurons (Pasquereau et al., Reference Pasquereau, Drui, Saga, Richard, Millot, Météreau and Tremblay2021). Loss of these neurons is therefore likely to impact the pharmacodynamics and therapeutic effects of antidepressants.

Other regions such as the amygdala, involved in processing emotional aspects of information and generating emotional reactions, project to the hippocampus which participates in the creation and maintenance of emotional associations in memory (DeRubeis, Siegle, & Hollon, Reference DeRubeis, Siegle and Hollon2008). More negative recall of information is a core negative affective bias found in depression, and antidepressant therapy has been consistently shown to be associated with activation of the amygdala (Nord et al., Reference Nord, Barrett, Lindquist, Ma, Marwood, Satpute and Dalgleish2021). Prominent hippocampal and amygdala atrophy is a signature pattern of AD and FTD respectively (Barnes et al., Reference Barnes, Whitwell, Frost, Josephs, Rossor and Fox2006) and is associated with loss of emotional and behavioural regulation (Rosen et al., Reference Rosen, Perry, Murphy, Kramer, Mychack, Schuff and Miller2002). Neurodegeneration of these regions and other parts of the neural circuitry mediating affective biases is a likely key mechanism driving loss of therapeutic efficacy of antidepressants in dementia.

An alternative neurocognitive explanation for loss of antidepressant effect is that negative affective biases in depression in dementia are driven by different cognitive processes (see Table 3). For example, reward-related decision-making changes in AD have been found to be mediated by impairment in memory, inhibitory control and set-shifting rather than reward processing deficits (Perry & Kramer, Reference Perry and Kramer2015). Executive dysfunction, rather than reward valuation, has similarly been found to drive reward-based decision making in FTD (Perry & Kramer, Reference Perry and Kramer2015). In PD, reward processing impairment is dependent on dopaminergic medication state and subcategory of reward processing (Costello et al., Reference Costello, Berry, Reeves, Weil, Joyce, Howard and Roiser2021). These examples highlight how negative affective biases driving depression may have variable mechanisms which would not be targeted by existing antidepressant therapies.

Loss of neuroplasticity

Almost all antidepressants increase the expression of brain-derived neutrophic factor (BDNF) via neutrophic tyrosine kinase receptor 2 (TRKB) (Casarotto et al., Reference Casarotto, Girych, Fred, Kovaleva, Moliner, Enkavi and Castrén2021). BDNF promotes neuronal maturation, plasticity and synaptic transmission by increasing production of brain cholesterol which facilitates cell membrane phospholipid fluidity of lipid rafts where many antidepressants interact (Casarotto et al., Reference Casarotto, Girych, Fred, Kovaleva, Moliner, Enkavi and Castrén2021). Antidepressants, including SSRIs and even rapid-acting agents such as ketamine, bind to TRKB, facilitating synaptic localisation and activation by BDNF. This induces synaptic plasticity that is believed to be a key underlying mechanism of antidepressants effects on mood (Casarotto et al., Reference Casarotto, Girych, Fred, Kovaleva, Moliner, Enkavi and Castrén2021). Mutation of the TRKB antidepressant binding site leads to loss of cellular, behavioural and neuroplasticity induced changes by antidepressants in vitro and in animal models (Casarotto et al., Reference Casarotto, Girych, Fred, Kovaleva, Moliner, Enkavi and Castrén2021).

Across different dementia disorders, BDNF is markedly reduced and has been implicated as a key pathological mechanism of neurodegeneration (Ng, Ho, Tam, Kua, & Ho, Reference Ng, Ho, Tam, Kua and Ho2019; Zuccato & Cattaneo, Reference Zuccato and Cattaneo2009). Lower BDNF compared to healthy controls is seen in PD and AD, and worsens with disease progression (Zuccato & Cattaneo, Reference Zuccato and Cattaneo2009). Although, it remains unclear whether BDNF changes in PD and AD are a cause or consequence of disease, the first phase-I clinical trial of BDNF gene therapy for cognitive decline in AD commenced last year (Tuszynski, Reference Tuszynski2021). Substantial evidence supports a causal role of BDNF dysfunction in the pathogenesis of Huntington's dementia as the huntingtin protein plays a crucial role in the expression and vesicular transport of BDNF (Zuccato & Cattaneo, Reference Zuccato and Cattaneo2009).

Post-mortem studies have not supported a similar BDNF reduction in late-life depression compared to dementia (Nunes et al., Reference Nunes, Nascimento, Kim, Andreazza, Brentani, Suemoto and Lafer2018), and no study has yet investigated whether basal BDNF level predicts response to antidepressants. Further investigation of the effects of BDNF modulation on antidepressant response and mood symptoms in dementia is needed. It is likely that any BDNF-mediated therapeutic effect of antidepressants would be reduced in dementia due to marked reduction in BDNF levels. However, it could also be argued that if antidepressant efficacy is partly mediated by augmenting BDNF then strategies to increase BDNF levels could be an effective approach in dementia where it is lower.

Evidence suggests that post-synaptic modulation of calcium signalling is crucial for functional plasticity and regulation of the strength of synaptic transmission in response to neuronal activity via long-term potentiation or depression. Animal studies have shown SSRIs to protect against stress-induced synaptic neuroplasticity via direct blockade of voltage-activated L-type Ca²⁺ channels (Normann et al., Reference Normann, Frase, Haug, von Wolff, Clark, Münzer and Bischofberger2018). Epidemiological evidence has found an association between L-type calcium channel antagonist use (e.g. verapamil hydrochloride) and reduced rates of psychiatric hospitalisation and self-harm in bipolar disorder (Hayes et al., Reference Hayes, Lundin, Wicks, Lewis, Wong, Osborn and Dalman2019). This highlights how antidepressant function could be independent of their monoaminergic effects. Further investigation of the role calcium channel signalling has in mediating antidepressant effects is needed. However, this mechanism is also likely to be impacted by dementia pathology. Marked dysregulation of calcium homoeostasis is believed to play an important role in the development of AD neuropathology and hyperphosphorylated tau has been shown to directly impair L-type calcium channels (Stan et al., Reference Stan, Church, Randall, Harvey, Brown, Wilkinson and Marrion2022; Wang, Shi, & Wei, Reference Wang, Shi and Wei2017).

Why antidepressants are still used in dementia: The clinical dilemma

Given the different aetiology and potential risks of treatment, it may appear straightforward to conclude that antidepressants have no role in depression in dementia and should not be used; however, this neglects the dilemma that patients, carers and clinicians face.

First, although trials have shown no superiority of antidepressants compared to placebo, the choice to prescribe antidepressants in clinical practice is not between antidepressant and placebo, but antidepressant and nothing. Though non-pharmacological interventions are effective and should be considered first-line treatments, many patients do not respond or are unable to engage with these interventions, which often require a good level of cognitive function. Even if only acting as placebo, we know placebo is effective at improving symptoms in both depression and in dementia (Hyde, May, Xue, & Zhang, Reference Hyde, May, Xue and Zhang2017). Though clearly the use of antidepressants purely for their placebo benefit is unwise given the potential risks of harm and side effects, undoubtedly, this consideration will contribute to clinical decision making and represents an important reason why prescription of antidepressants in depressed people with dementia is continued.

Second, the lifetime risk of depression is high and it is estimated that up to half the population will experience one or more episodes of depression during their lifetime (Andrews, Poulton, & Skoog, Reference Andrews, Poulton and Skoog2005). When treating older adults with depression and dementia who have previously had episodes of depression successfully treated by antidepressants prior to developing dementia, the clinical question of whether to re-commence an antidepressant is difficult. Is this depression in the context of dementia or a relapse of existing major depressive disorder? Do antidepressants lose their efficacy in long-term depression users who develop dementia? Discontinuation of antidepressants in older adults with dementia who have remained on treatment for an extended period is similarly challenging, with evidence from one RCT showing worsening in depressive symptoms following discontinuation (Bergh, Selbæk, & Engedal, Reference Bergh, Selbæk and Engedal2012).

Third, accurate diagnosis of depression in the general population is reliant on recall of depressive symptoms over the previous two weeks and detailed description of subjective symptoms by patients. In patients with dementia who have marked cognitive impairment this is often not possible, and clinicians will be reliant on collateral information from relatives and carers. However, the accuracy of such information varies enormously and is dependent on the nature of the relationship; it can also place considerable demands on family members who may be experiencing significant carer stress and low mood themselves. High turnover of staff working shift patterns in long term-care facilities (Costello, Cooper, Marston, & Livingston, Reference Costello, Cooper, Marston and Livingston2020) can also complicate the process of making an accurate diagnosis and subsequent monitoring of depressive symptoms.

Finally, while antidepressants may not be effective in depression in dementia they are widely used for alternative indications which may overlap with depression. FTD guidelines support SSRI use to treat behavioural disinhibition and agitation despite no RCT having been conducted to date (O'Brien et al., Reference O'Brien, Holmes, Jones, Jones, Livingston, McKeith and Burns2017). Antidepressants are also used for treatment of neuropathic pain, which is highly prevalent in depression and dementia (Achterberg, Lautenbacher, Husebo, Erdal, & Herr, Reference Achterberg, Lautenbacher, Husebo, Erdal and Herr2019), and they are frequently (mis)used to treat agitation as a substitution strategy to avoid antipsychotic prescription (Maust, Kim, Chiang, & Kales, Reference Maust, Kim, Chiang and Kales2018).

The clinical complexity of diagnosis and management of depression in dementia, and use of antidepressants for other BPSD, clouds how clinicians decide whether to prescribe antidepressants. We must be honest with patients and carers about the lack of efficacy and potentially harmful consequences of antidepressant use in dementia, but also acknowledge how challenging clinical decision-making regarding prescription in this patient group can be.