NEUROPSYCHIATRIC SYMPTOMS IN DEMENTIA
New work based on the results of genome-wide association studies in Alzheimer’s disease has identified molecular mechanisms that are shared with some psychiatric disorders. The study took advantage of omics data and has the realistic potential to elucidate unknown disease mechanisms; however, the lack of information about neuropsychiatric symptoms in participants with Alzheimer’s disease limits conclusions.
Alzheimer’s disease (AD) is clinically heterogeneous. In addition to the core features of cognitive and functional impairment, it is also characterized by an often debilitating range of neuropsychiatric symptoms. These symptoms most commonly include apathy, depression, agitation and psychosis and are present to some extent in most people with AD dementia. Treatments for the neuropsychiatric symptoms of AD are only modestly effective and, in some cases (e.g., antipsychotics), carry serious safety concerns. The need for new treatments for the neuropsychiatric symptoms of AD is clear, and addressing this problem must begin with a better understanding of the molecular mechanisms that underlie the heterogeneous
of the symptoms. There are, broadly speaking, two routes to achieve this goal. The first is the discovery of entirely new disease mechanisms in cohorts of individuals with AD in whom neuropsychiatric symptoms have been clinically characterized. The second involves the search for shared mechanisms between AD and other psychiatric disorders. This latter route is motivated by the observation that affective, psychotic, or agitated syndromes occurring in dementia share some phenomenological similarities with psychiatric disorders that generally begin earlier in life. Indeed, this concept is one of the pillars of logic behind a new study by Thomas Wingo and colleagues that integrated results from genome-wide association studies (GWAS) with transcriptomic and proteomic data to identify causal proteins that are shared between neurodegenerative and psychiatric disorders. However, there are limits to what this study can tell us about the mechanisms underlying neuropsychiatric symptoms in AD.
Wingo et al. used summary statistics from five GWAS of major neurodegenerative diseases (including AD), eight GWAS of psychiatric conditions and traits, and eleven GWAS of structural neuroimaging phenotypes to identify genetic correlations across phenotypes. Using summary statistics, the researchers performed a proteome-wide association study, Mendelian randomization based on summary data, and colocalization analysis to identify possible causal proteins for each phenotype. Finally, they looked at which of these causal proteins were shared across phenotypes. Interestingly, the authors found considerable overlap between the causal proteins identified in neurodegenerative disease and psychiatric disorders. One of the conclusions drawn by Wingo et al. was that these associations reflect shared mechanisms between AD and psychiatric disorders, so drugs targeting shared proteins could be useful in treating both primary AD and associated neuropsychiatric symptoms. However, the GWAS summary statistics used in the study were from case-control studies of AD in which comorbid neuropsychiatric symptoms were not measured. Therefore, the conclusion regarding common treatment targets should be viewed with caution, as there remains some ambiguity about what shared causal proteins reflect.
We know that AD is a clinically heterogeneous disease, but what is increasingly clear is that there is also biological heterogeneity among individuals with AD. For example, psychosis, characterized by delusions and hallucinations (but clinically distinct from schizophrenia), is present in approximately 40% of individuals with AD, and a growing body of evidence has identified genomic and neurobiological correlates of these symptoms within AD. Notably, the first risk loci for AD psychosis were reported in 2021, and differences in brain DNA methylation have been observed between individuals with AD with and without psychosis. In one study, AD psychosis was genetically correlated with major depressive disorder and years of education. In addition, evidence indicates that, compared with individuals with AD without psychosis, individuals with AD psychosis have a higher genetic risk of schizophrenia and differentially methylated genome regions in the brain.
loci associated with schizophrenia. Preliminary data indicate that the genetic risk of major depression is higher in people with AD who have depressive symptoms than in people with AD without depressive symptoms, suggesting that, with respect to neuropsychiatric syndromes, individuals with AD are not a biologically homogeneous group. In any AD GWAS, between 10% and 50% of participants with AD could reasonably be expected to have comorbid psychosis. Although this comorbidity is unlikely to affect the ability to identify genes associated with AD in a GWAS, it could have an effect on measures of genetic correlations between AD and psychiatric traits because of evidence that both AD and psychosis within AD have some genetic overlap with other psychiatric conditions. Thus, the genetic correlational signals underpinning the findings of Wingo et al. have at least two potential drivers. The first is associations between psychiatric disorders and primary AD; however, the second is correlations between psychiatric disorders and unmeasured comorbid comorbid neuropsychiatric symptoms in individuals with AD. This possibility does not necessarily make the associations reported false, but it has implications for drawing conclusions about pathways for treatment development. In order to develop treatments for neuropsychiatric symptoms in AD, ultimately, individuals with AD who have undergone clinical characterization for the presence or absence of these symptoms must be studied. At present, the identified shared molecular mechanisms do not necessarily represent a specific marker of neuropsychiatric symptoms in AD.
Overall, as a field, we certainly need more studies employing the methodology used by Wingo et al. but further validation of these results in clinically characterized AD cohorts for neuropsychiatric symptoms is required. To this end, an extension of the study would be to test whether the levels of the shared candidate proteins identified by Wingo et al. differ in brain tissue from individuals with AD with and without neuropsychiatric symptoms. At present, the authors have only inferred that proteins are involved in the disease by virtue of the association between protein levels in human brain tissue (from a variety of sources with a variety of pathology) and
single nucleotide polymorphisms from the supporting GWAS summary statistics.
Understanding the molecular mechanisms that are shared between neuropsychiatric symptoms in AD and other psychiatric conditions that occur earlier in life could help to develop new treatments for both groups of conditions, or to guide more effective use of existing psychotropic medications in AD. Harnessing omics data is a sensible route to achieve this, but, to ensure that the identified mechanisms are specific, further validation is required. Alternative study designs to elucidate more specific signals would involve association testing within groups of individuals with AD characterized by neuropsychiatric syndromes and between these groups and healthy controls. However, by capitalizing on the large-scale GWAS resources that have been accumulated in recent years, the comprehensive study by Wingo et al. paves the way for future research in this understudied area by yielding a number of new candidate proteins that can now be quantified in AD cohorts with detailed psychiatric assessments.
Creese, B., Lunnon, K. Neuropsychiatric symptoms in AD: the search for mechanisms. Nat Rev Neurol 18, 639-640 (2022). https://doi.org/10.1038/s41582-022-
Wingo, T. S. et al. Shared mechanisms across the major psychiatric and neurodegenerative diseases. Nat. Commun. 13, 4314 (2022).