Towards Defining the Neuroanatomical Basis of Late-Onset Psychiatric Symptoms

Abstract

Psychiatric symptoms, including changes in emotional processing, are a common feature of many neurodegenerative disorders, such as Alzheimer’s disease, dementia with Lewy Bodies, frontotemporal dementia, and Huntington’s disease. However, the neuroanatomical basis of emotional symptoms is not well defined; this stands in contrast to the relatively well-understood neuroanatomical correlates of cognitive and motor symptoms in neurodegenerative disorders. Furthermore, psychiatric diagnostic categories, as defined by the Diagnostic and Statistical Manual of Mental Disorders (DSM) and International Statistical Classification of Diseases and Related Health Problems (ICD), may have limited applicability in patients with late-onset psychiatric symptoms in the context of neurodegenerative disorders. In this clinical review, we suggest that early-onset and late-onset psychiatric symptoms have distinct etiologies, and that late-onset changes in emotional processing are likely underpinned by neurodegenerative disease. Furthermore, we suggest that an improved understanding of the neuroanatomical correlates of emotional changes in neurodegenerative disease may facilitate diagnosis and future treatment development. Finally, we propose a novel clinical approach, in a preliminary attempt to incorporate late-onset emotional symptoms alongside cognitive and motor symptoms into a clinical “algorithm,” with a focus on the neuroanatomy implicated when particular combinations of emotional, cognitive, and motor features are present. We anticipate that this clinical approach will assist with the diagnosis of neurodegenerative disorders, and our proposed schema represents a move towards integrating neurologic and psychiatric classification systems.

Introduction

Dementing neurodegenerative disorders commonly present with a complex phenotype that can include motor, cognitive, and psychiatric symptoms. The fields of neurology and neuropsychology have emphasized the importance of the neuroanatomical localization of motor and cognitive symptoms in neurodegenerative illnesses, with notable success. By contrast, the neuroanatomical basis of psychiatric symptoms in neurodegenerative disorders, including changes in emotional processing, has lagged behind that of motor and cognitive symptoms. It is becoming increasingly apparent that late-onset psychiatric (e.g., emotional) symptoms may signal the initial stages of a neurodegenerative illness.¹

As an example, late-life depression (onset after age 65) has been investigated more than any other late-life psychiatric syndrome. Multiple studies suggest that late-life depression and dementia share common neuropathological changes, including reduced gray matter volume² and hippocampal atrophy.³ Interestingly, some of these changes, such as hippocampal atrophy, are associated specifically with late-onset depression and not depression that begins earlier in life.^3,4 Patients with late-onset depression may also present as clinically distinct from patients with early-onset depression, and are less likely to have co-morbid personality disorders or high levels of neuroticism.⁵ Further, a recent study reported important differences in cognitive function between patients with late-onset vs early-onset depression.⁶ Patients with late-onset depression had broader cognitive deficits and experienced rapid decline in memory performance over time.⁶ Perhaps not surprisingly, late-onset depressive symptoms have also been associated with subsequent diagnosis of dementia.⁷ Clearly, a better understanding of the neuroanatomy underlying late-life psychiatric symptoms, and how these symptoms relate to neurodegenerative disorders, will lead to a better understanding of psychiatric and neurological disorders in later life. In this clinical review, we present evidence that late-onset and early-onset psychiatric symptoms are etiologically distinct, and that psychiatric symptoms with a later onset may be associated with measurable neurodegenerative and cerebrovascular processes. We discuss the neuroanatomical basis and diagnostic utility of specific psychiatric symptoms, namely, emotional over-reactivity, anxiety and dysphoria (present in depression) contrasted with emotional blunting, to propose a novel clinical approach that may prove useful in mapping the neuroanatomy of psychiatric symptoms in neurodegenerative disorders.

Neurologists generally have more experience with neurodegenerative disorders than most psychiatrists, but a conceptual framework to understand the neuroanatomical basis of emotional symptoms is important for psychiatrists, especially consult/liaison and geriatric psychiatrists, for several reasons. First, patients with neurodegenerative diseases such as behavioral variant frontotemporal dementia (bvFTD) will often present initially to psychiatrists, because emotional changes are the predominant early symptoms.^8,9 Second, recent evidence indicates that neurodegenerative disorders are far more common in older adults than we had previously thought, and limbic structures are frequently implicated.^10,11 Indeed, psychiatrists, as experts in the assessment and treatment of emotional symptoms, have much to add to the collaborative clinical assessment of these patients. Finally, identifying the neuroanatomical basis of emotional symptoms in neurodegenerative illnesses may be able to inform our understanding of the brain regions underpinning emotional symptoms in patients with psychiatric disorders that are not associated with neurodegenerative illness (i.e., primary or “idiopathic” psychiatric disorders).¹²

There are several greater challenges in identifying the neuroanatomical basis of psychiatric symptoms, compared to motor and cognitive symptoms, in individuals with neurodegenerative disorders. First, psychiatric symptoms are usually measured subjectively, with examiners most often relying on self-report, which requires a level of insight that may be impaired in patients with neurodegenerative illness. Second, there is a relatively high background prevalence of psychiatric disorders in the population, especially mood and anxiety disorders, which may pre-date the neurodegenerative process. Third, emotional experience exists on a continuum from normal to pathological emotional states (e.g., depressed mood). Finally, emotional changes in neurodegenerative disorders are bidirectional, and can manifest as the excessive experience of emotional states, such as high dysphoria and anxiety, or, at the opposite end of the spectrum, emotional blunting, which is considered a pathological lack of emotional responsiveness. We also highlight that there is an inherent difficulty in distinguishing psychiatric symptoms directly related to the biological effects of neurodegenerative illness on the brain vs the emotional impact of the negative or stressful life event that is obtaining a neurodegenerative diagnosis (e.g., “of course she is depressed, she has Alzheimer’s disease”). Nevertheless, identifying the basis of psychiatric symptoms in neurodegenerative disease is critical, due to both their ubiquity and impact. The vast majority of individuals with neurodegenerative disease (e.g., Alzheimer’s disease, frontotemporal dementia, dementia with Lewy Bodies, and Huntington’s disease) will experience psychiatric symptoms at some point during the disease course.^9,13–16 Such symptoms, including emotional changes, are important predictors of increased burden of care, caregiver distress, and out-of-home placement, as well as reduced quality of life.^13,17–23 Emotional changes (e.g., depressive symptoms) also predict accelerated disease progression and functional decline in individuals with dementia.^17,24,25

Specific factors within the field of psychiatry have contributed to the difficulty in pursuing the neuroanatomical mapping of emotional symptoms in neurodegenerative disorders. Generally, psychiatry has placed less emphasis on lesion studies to determine neuroanatomical correlates than neurology and neuropsychology,¹² and the current psychiatric classification systems (Diagnostic and Statistical Manual of Mental Disorders [DSM] and International Statistical Classification of Diseases and Related Health Problems [ICD]) are descriptive, categorical, and not based on neuroanatomical findings, as they were primarily developed to describe psychiatric illness in young people without neurodegenerative disorders.^26,27 Indeed, these psychiatric classification systems (DSM and ICD) assume that psychiatric disorders are “universal” across the adult life span; that is, psychiatric disorders in adults are classified and diagnosed the same no matter the age of the patient.^26,27 We argue that relaxing this assumption would greatly facilitate investigations into the neuroanatomical basis of psychiatric symptoms that have their onset later in life (after the age of 65).

The poor neuroanatomical localization of psychiatric symptoms in neurodegenerative disorders represents a crucial knowledge gap. At current, existing pharmacological treatments for psychiatric/emotional symptoms in neurodegenerative disorders (e.g., antidepressants) are often inadequate, eliciting poor treatment response.^28,29 Neuroanatomically-based treatments (e.g., brain stimulation) for neuropsychiatric symptoms are promising,³⁰ but the development of these treatments is hampered by our lack of understanding of their neuroanatomical basis. However, perhaps the most compelling reason to pursue investigations into the neuroanatomical basis of emotional symptoms in neurodegenerative disorders is that pinpointing anatomic underpinnings of symptoms is the sine qua non of diagnosis, not just in neurology, but in all of medicine.

Early-vs Late-Onset Psychiatric Disorders

For reasons that are not well understood, the majority of DSM- and ICD-defined primary psychiatric disorders, including mood, anxiety, psychotic, and eating disorders, most commonly have their onset in adolescence and early adulthood³¹ (see Figure 1). This notion is supported by data from the National Comorbidity Replication study and multi-generational studies of families whose members are at high risk for depression and anxiety.^32–34 However, there appears to be a “second peak” in later life, whereby we see an increasing incidence of psychiatric symptoms in individuals who did not have a primary psychiatric disorder earlier in life. This “second peak” mirrors the increasing prevalence of neurodegenerative and brain vascular processes with age³⁵ (see Figure 2 for a theoretical model). We suggest that these 2 phenomena are related; this second peak of psychiatric symptom incidence in individuals without any history of primary psychiatric disorder earlier in life may in fact represent symptoms of neurodegenerative and brain vascular processes. This notion has parallels in cognitive and motor symptoms, in that early-vs late-onset cognitive and motor symptoms usually have vastly different etiologies, manifestations, and courses. For example, if a 20-year-old is experiencing new-onset severe cognitive symptoms and is no longer able to work it is more likely to be due to psychiatric disease (e.g., schizophrenia) than Alzheimer’s disease, but the opposite is true for an 85-year-old. Similarly, the etiology and course of motor symptoms usually differ between early- and late-onset. Thus, the usefulness of our current psychiatric classification systems (DSM and ICD) to describe late-onset psychiatric symptoms is limited. In being developed to describe early-onset psychiatric disorders and assuming that psychiatric symptoms in adults should be classified and diagnosed the same regardless of age,^26,27 DSM and ICD inadequately describe and diagnose this “second wave” of late-onset psychiatric illness, necessitating an alternative approach.

Figure 1.

Age of onset of primary psychiatric disorders. From Paus et al., 2008.³¹

Figure 2.

Theoretical model of incidence and etiology of psychiatric symptoms by age.

We highlight that older individuals with psychiatric illnesses include those with early-onset psychiatric disorders who have aged, as well as those who develop new-onset psychiatric symptoms later in life. This review focuses on patients with new-onset psychiatric symptoms, defined as occurring for the first time at the age of 65 or older, as this group is etiologically and phenotypically distinct from young persons with psychiatric illness. However, we wish to acknowledge that the findings discussed may also have important implications for older persons who developed a primary psychiatric disorder in early life and have aged. The psychiatric symptoms of these patients often change late in life, possibly as a result of neurodegenerative and vascular processes.³⁶

Ubiquity of Neurodegenerative and Cerebrovascular Pathology in Older Adults

Recent evidence, from both community-based neuropathological and in vivo biomarker studies, indicates that neurodegenerative and brain vascular processes are more common in older age than previously thought. For example, participants in the Rush Memory and Aging Project have been examined post-mortem, with the majority meeting neuropathological criteria for a neurodegenerative illness and/or significant cerebrovascular disease, including those participants with normal cognition¹⁰ (Figure 3). Participants who died with normal cognition had the lowest prevalence of neurodegenerative and vascular illness, but even in this group of cognitively normal participants, it was only a minority of participants who did not meet neuropathological criteria for any neurodegenerative or vascular illness¹⁰ (Figure 3).

Figure 3.

Neuropathological diagnoses in a large community cohort by cognitive status prior to death. Abbreviations: NCI = No Cognitive Impairment; MCI = Mild Cognitive Impairment; AD = Alzheimer’s disease; V = Vascular; OD = Other Degenerative; 0 = no vascular or neurodegenerative pathology. From Kapasi et al., 2017.¹⁰

In addition to Alzheimer’s disease neuropathology, mixed dementias and non-Alzheimer neurodegenerative processes are relatively common, including alpha-synucleinopathies, primary tauopathies, and TDP-43opathies.^10,37 Furthermore, the prevalence of brain vascular processes associated with dementia appears to be higher than was previously thought¹ (Figure 3).¹⁰ Indeed, data now suggest that white matter vascular illness is associated with dementia, even if it does not fulfill full clinical criteria for vascular dementia.^39,40 In addition, small vessel vascular disease may predispose brain regions to the development of Alzheimer pathology, and patients with vascular illness are more likely to have concomitant Alzheimer’s disease than patients without vascular illness.^41,42 Data presented in Figure 3 demonstrate the commonality of mixed pathologies, including the overlap between neurodegenerative proteinopathies and brain vascular illness. These data, in conjunction with findings showing the phenotypic heterogeneity of these proteinopathies, have led some to advocate for a move away from specific diagnoses (e.g., Alzheimer’s disease dementia), to “additive” dementia risk factor models in which the risk of dementia increases with the number of different underlying neuropathological risk factors (e.g., amyloid pathology, tau pathology, vascular disease, and TDP-43 pathology) (Figure 4).^10,43 Importantly, new clinical presentations of neurodegenerative proteinopathies continue to be described (e.g., Limbic-Predominant Age-Related TDP-43 Encephalopathy [LATE]).¹¹

Figure 4.

An additive model of neuropathologies contributing to cognitive impairment. Cognitive status is defined as demented, definitely impaired, or marginal at the last examination prior to death. Lesion indices include AD changes, microvascular ischemic lesions, cortical Lewy bodies, hippocampal sclerosis, and generalized atrophy. From White et al., 2016.⁴³

Association Between Neurodegenerative Processes and Psychiatric Symptoms

What is the evidence that these common neuropathological and vascular changes in older adults are associated with late-onset psychiatric symptoms? Patients with genetic neurodegenerative disorders can inform this question, and prospectively examining individuals with pathogenic mutations can help determine the earliest symptoms of neurodegenerative illness. As primary examples, evidence suggests that in patients with genetic Huntington’s disease and bvFTD, psychiatric changes are usually the earliest symptoms, preceding motor and cognitive symptoms by 3 to 7 years.^8,16,44

The development of in vivo biomarkers for amyloid and tau has facilitated explorations of early and prodromal psychiatric symptoms of sporadic neurodegenerative disorders. In older individuals with normal cognition but amyloid positivity on PET imaging (i.e., high risk of developing dementia due to AD), the presence of anxiety significantly predicts future memory loss.⁴⁵ New-onset symptoms of depression are commonly a prodrome of dementia due to AD, preceding or co-occurring with cognitive declines.^46,47 Patients with dementia with Lewy Bodies also often present with early psychiatric symptoms, sometimes in the absence of other symptoms.⁴⁸ Therefore, there is manifold evidence that psychiatric changes occur early in neurodegenerative disease, in many cases before the onset of other cognitive or motor symptoms.

Mapping the Anatomy of Late-Onset Psychiatric Symptoms in Neurodegenerative Disease

Many neurological and neurobehavioral symptoms of neurodegenerative illness are known to preset bidirectionally, and the directionality of the symptom may provide some insight into the etiology or neuroanatomy implicated. For example, in the motor domain, there are hyperkinetic and hypokinetic movement disorders, such as Huntington’s disease and Parkinson’s disease, which differentially target basal ganglia pathways. In the language domain, a patient may present with impoverished but meaningful spontaneous speech (e.g., dynamic aphasia) or excessive but nonsensical spontaneous speech (e.g., jargon aphasia), depending on which specific regions of the frontal lobes are dysfunctional.^49,50 Sleep and appetite changes may also occur bidirectionally (i.e., over eating or sleeping vs under eating or sleeping). In the same vein, psychiatric symptoms, such as changes in emotional responsiveness, can also present bidirectionally, with either pathologically increased or decreased emotional reactivity, dysphoria and/or anxiety. In this section we propose that contrasting these “opposite” psychiatric symptoms of emotional reactivity, which differentially present across neurodegenerative disorders, may facilitate our understanding of the neuroanatomical basis of psychiatric symptoms in the context of neurodegeneration.

We suggest that one way to begin integrating neurologic and psychiatric classification systems is to focus on psychiatric symptoms that have been linked to dysfunction in a specific region of the brain. To this end, there is empiric support for, and clinical utility of, focusing on a specific symptom: emotional blunting. Emotional blunting, which is defined as inappropriately decreased emotional reactivity, dysphoria, and anxiety to stimuli that should induce these emotional reactions, is a core negative symptom of psychiatric disorders such as schizophrenia. Emotional blunting is also a hallmark symptom of bvFTD, a neurodegenerative disease on the spectrum of frontotemporal lobar degeneration (FTLD) disorders.⁵¹ In the clinical context, emotional blunting is frequently observed when patients are informed that they have bvFTD, a progressive and likely fatal neurodegenerative illness; they generally respond to this information, which would usually elicit dysphoria and/or anxiety, with indifference.

Converging evidence suggests that dysfunction of the ventromedial prefrontal cortex (vmPFC) may be associated with emotional blunting. For example, there is a lower-than-expected prevalence of mood and anxiety disorders in persons with penetrating TBI of the vmPFC.⁵² Additionally, there is some evidence of reduced depressive symptoms in patients with Major Depressive Disorder after deep brain stimulation of the medial aspect of the vmPFC (the subcallosal cingulate cortex).³⁰ The neuropathology underlying bvFTD is known to target the vmPFC early in the disease,^51,53 and, supporting this, there is a lower-than-expected prevalence of mood and anxiety disorders in both the preclinical and clinical stages of bvFTD.⁸ Similarly, decreased volume in the right vmPFC in patients with mild bvFTD is associated with decreased depressive symptoms.⁵⁴

By contrast, there is emerging evidence that degeneration of other limbic regions with strong reciprocal projections with the vmPFC, including the dorsal striatum (targeted by Huntington’s disease) and the anterior temporal lobes (targeted by semantic variant primary progressive aphasia [svPPA] or semantic dementia), are usually associated with increased emotional reactivity, anxiety, and dysphoria.^55,56

Importantly, data from patients with genetic Huntington’s disease and bvFTD demonstrate that while psychiatric symptoms are common early symptoms, they usually do not conform to DSM categories.^8,16,44 For example, characteristic psychiatric symptoms of early Huntington’s disease include obsessional thinking without compulsivity, dysphoria, irritability, and emotional lability.¹⁶ These symptoms together do not fulfill criteria for a DSM- or ICD-defined disorder, but rather appear to represent the psychiatric symptoms commonly associated with the neuronal dysfunction and early neurodegeneration of brain structures and regions targeted by early Huntington’s disease (notably the dorsal striatum).⁵⁷

Neuroanatomic regions clearly operate in conjunction with other regions as part of broader networks dedicated to supporting specific aspects of psychiatric, behavioral, and cognitive functions. Moreover, there is accumulating evidence that different neurodegenerative proteinopathies target specific structural and/or functional brain networks.^58,59 Conceptually, it is therefore critical to understand these networks, and how they become disrupted in the context of neurodegenerative disease. However, examining the entirety of a network’s structure and function remains methodologically challenging, particularly in relation to specific symptoms. Furthermore, the evidence to date suggests that damage to nodes within identified networks may produce specific symptoms. For example, within the peri-Sylvian brain network central to language abilities, individual nodes have differential functions (e.g., Broca and Wernicke’s areas) and damage to these nodes can be distinguished clinically.⁶⁰ In this section we have highlighted specific anatomic regions that appear to have key relevance for various emotional/psychiatric symptoms, appreciating the fact that these regions operate within a broader network whose integrity will ultimately need to be measured and understood. As one example, the medial PFC, discussed above, operates as a node within the default mode network (DMN), and there is growing evidence that DMN activity is disrupted in primary mood disorders; indeed, patients with Major Depressive Disorder demonstrate greater baseline activity of the DMN, especially the medial PFC.^61–65 Understanding the role of the DMN in the pathogenesis of psychiatric symptoms across neurodegenerative disorders is of key importance for the field moving forward.

In its current form, the algorithm presented below is intended to use clinical data to inform the underlying neuroanatomy rather than to incorporate the latter as part of the formulation. In the validation and further specification of this algorithm, neuroanatomic studies using structural (e.g., CT and MRI) and functional (e.g., fMRI, PET, and SPECT) neuroimaging, and/or post-mortem neuropathological examination will be invaluable in determining the extent to which the presumptive neuroanatomic regions put forth by the algorithm are implicated and can be predicted with adequate sensitivity and specificity in individual patients.

Proposed Preliminary Clinical Algorithm

Despite the challenges in mapping neuroanatomy to late-onset emotional changes, especially when considering that neurodegenerative disorders target distributed brain networks, we can begin to develop an empirically informed clinical “algorithm.” This algorithm is depicted in Figure 5 as a flowchart, and is intended to preliminarily map the neuroanatomy of certain emotional changes associated with neurodegenerative disorders. Similar schemas for the localization and diagnosis of tremor and aphasia have been published and proven clinically useful.⁶⁶ The current proposed algorithm combines motor and cognitive symptoms with emotional changes, in an attempt to preliminarily map the neuroanatomical regions that are implicated when certain emotional symptoms are present clinically, with the neurodegenerative disorders that most commonly target that region listed on the periphery in gray italic text.

Figure 5.

Flowchart depicting a clinical algorithm to neuroanatomically localize emotional changes observed with neurodegenerative disorders. Neurodegenerative disorders targeting the neuroanatomical regions are listed in italic gray text. Abbreviations: WM = white matter; vmPFC, bvFTD = behavioral variant frontotemporal dementia; svPPA = semantic variant primary progressive aphasia; SD = semantic dementia; AD = Alzheimer’s disease; PD = Parkinson’s disease; DLB = dementia with Lewy Bodies; HD = Huntington’s disease.

The first level of the algorithm asks the clinician to assess for the presence of late-onset emotional changes. Here, the emotional symptoms can be categorized as heightened emotionality (i.e., lability, dysphoria, and anxiety), emotional blunting, or pseudobulbar affect. As aforementioned, emotional blunting has been linked to dysfunction of the vmPFC in lesion studies, brain stimulation research, and neurodegenerative disease investigations.^8,30,52 Emotional blunting is a characteristic feature of bvFTD.⁵¹ By contrast, pseudobulbar affect is suggestive of disrupted frontal white matter tracts^67,68; there is no specific neurodegenerative disorder indicated here in the flowchart as it is a symptom that is present across many disorders including amyotrophic lateral sclerosis, multiple sclerosis, progressive supranuclear palsy, and others.^69,70 The presence of heightened or blunted emotionality will lead the clinician to the second level of the algorithm, where motor symptoms will be assessed. A hyperkinetic movement disorder suggests that indirect basal ganglia pathways are implicated, and this is a characteristic feature of Huntington’s disease.⁷¹ A hypokinetic movement disorder (i.e., parkinsonism) suggests that the basal ganglia is directly affected; this feature is typical of Parkinson’s disease and common in dementia with Lewy Bodies.^72,73 If no movement disorder is present, the clinician will be directed to the third level of the algorithm, which requires a cognitive assessment. A pattern of impairment where semantic memory (i.e., loss of concept, word, or object knowledge) is impaired to a greater degree than episodic memory (i.e., loss of the ability to recall events/experiences) would suggest compromise of anterior more so than medial portions of the temporal lobe.^74,75 This cognitive profile is the hallmark of svPPA or semantic dementia, in which anterior temporal atrophy (typically left greater than right) is the neuroanatomical signature.^76,77 As aforementioned, individuals with svPPA have been reported to be dysphoric and often suicidal.^55,56 Finally, if episodic memory impairments² are more pronounced than semantic memory loss, the clinician should assess for the presence of visuospatial disturbances. If visuospatial skills are impaired in the context of episodic memory loss, then it is likely that parietal regions are disrupted, along with hippocampal/parahippocampal areas.^78,79 This cognitive profile, in the context of increased emotional reactivity (e.g., anxiety and dysphoria) and no prominent movement disorder, is most indicative of Alzheimer’s disease.^80–82 An overall lack of episodic memory and visuospatial decline over time may be more consistent with a primary (i.e., “idiopathic”) psychiatric disorder, as executive dysfunction or processing speed deficits tend to be more typical of mood disorders.^83,84

Overall, certain combinations of emotional, cognitive, and motor symptoms will lead the clinician to suspect the dysfunction of particular brain regions and certain neurodegenerative disorders. However, we strongly recommend that the clinical features identified in the flowchart be interpreted in the context of the bigger clinical picture, as the algorithm is agnostic to base rates of different neurodegenerative conditions and individual risk factors. In other words, while the neurodegenerative disorders listed in gray are those that most commonly target the identified region, it is possible that a specified disorder is unlikely in a given patient. For example, a 90-year-old whose semantic knowledge is more impaired than their episodic memory is more likely to have Alzheimer’s disease than svPPA, based on population incidence. Nonetheless, the profile of cognitive function implicates a more anterior distribution of temporal neuropathology than would typically be seen in Alzheimer’s disease.

It is also critical to note that while certain combinations of symptoms should raise clinical suspicion of dysfunction of the brain regions specified in the algorithm, we cannot yet indicate how these regions are implicated (e.g., via direct atrophy and through dysfunctional network connections with another region). This will be an important target for future research. We also emphasize that, in its current state, this preliminary algorithm may be limited in the number of neurodegenerative patients it applies to. How the algorithm performs within and across different neurodegenerative patient groups should be empirically investigated in prospective studies. We anticipate that as our understanding of the neuroanatomical underpinnings of emotional changes in neurodegenerative disease improves, the algorithm can be expanded to encompass additional brain regions, which will hopefully increase its generalizability and applicability to a greater number of patients and neurodegenerative disorders.

We further highlight that this framework is not intended to be applicable to all neuropsychiatric or emotional symptoms; rather, it is intended to be a preliminary example with scope for future expansion, focused on a subset of common emotional symptoms of early neurodegenerative disease, namely, emotional blunting vs increased emotional lability, anxiety, and dysphoria. The current proposed algorithm should be modified and augmented as we learn more about the neuropsychiatric symptoms associated with dysfunction of specific brain areas (e.g., the cerebellum), and the neuroanatomical bases of other neuropsychiatric symptoms (e.g., psychosis). Future iterations could be also extended to include complex cognitive and motor functions, such as executive functioning or apraxia. It should also be adjusted as the neuropsychiatric profiles of other neurodegenerative disorders, including newly discovered neurodegenerative proteinopathies (e.g., Limbic-predominant age-related TDP-43 encephalopathy or LATE),¹¹ become better understood.

Some of the branch points in Figure 5 are binary (e.g., the presence or absence of a hyperkinetic movement disorder), while others require clinician judgment about an appropriate cutoff or a pattern of decline. For example, both episodic memory and semantic memory deficits may be present both in Alzheimer’s disease dementia and svPPA, and a judgment needs to be made as to which was impaired earliest, or which is impaired to a greater degree, taking into account the broader pattern of cognitive and language functioning. Multiple longitudinal assessments, or an in-depth assessment from a clinical neuropsychologist, may be required to ascertain this. Finally, we acknowledge that psychiatrists may have minimal exposure to rare disorders seen mostly in neurology settings (e.g., svPPA). Thus, consultation with and evaluation by relevant experts is advised, including neuropsychologists, behavioral neurologists, and movement disorder neurologists.

Although this algorithm is framed in terms of neurodegenerative disorders, in part due to their relatively high prevalence in older adults, we anticipate that studies of other causes of regional brain dysfunction, including penetrating TBI and stroke, should find similar regional associations to those detected in neurodegenerative disorders. As one example, discussed above, the evidence on the neuroanatomical basis of emotional blunting is consistent in penetrating TBI and neurodegenerative models. However, whether the brain-behavior relationships specified in this algorithm apply to other causes of brain injury/dysfunction should be investigated in future studies. Likewise, the algorithm may also be relevant to other late-onset psychiatric illness outside the context of neurodegenerative disease, but at current this remains unclear.

Conclusion

Overall, in this clinical review we suggest that late- and early-onset psychiatric illnesses have dissimilar etiologies, and that the vast majority of “primary” psychiatric disorders defined by DSM and ICD have limited applicability to late-onset psychiatric changes because they were developed in the context of early-onset psychiatric disorders. Further, emerging evidence suggests that late-onset psychiatric symptoms are associated with neurodegenerative and brain vascular processes. Against this clinical and empiric backdrop, we have proposed a preliminary algorithm prototype, in the form of a flowchart, to begin to neuroanatomically localize psychiatric symptoms in neurodegenerative disease. This novel diagnostic approach breaks with DSM and ICD, instead prioritizing localizable psychiatric symptoms, and may prove clinically useful when late-onset psychiatric symptoms are present.

By challenging the universality of psychiatric diagnoses across the adult life span, neuroanatomically based, and ultimately biomarker-driven, psychiatric diagnoses are achievable, albeit for an important subset of psychiatric patients. We believe that this reconceptualization will also aid in the development of novel therapeutics tailored to this population and could help explain differences in treatment responsiveness between early- and late-onset psychiatric illness.