Neuropsychiatric symptoms and comorbidity: Associations with dementia progression rate in a memory clinic cohort

Neuropsychiatric symptoms (NPS) are associated with dementia severity and progression rate. NPS clusters have different neurobiological underpinnings; therefore, their effect on dementia progression may differ. Furthermore, little is known about whether individual comorbidities affect progression rate. We investigated the effect of NPS clusters and individual comorbidities on dementia progression.


| INTRODUCTION
Dementia disorders are characterized by cognitive and functional decline, and many patients with dementia have neuropsychiatric symptoms (NPS) 1 and comorbidities 2 that may influence progression.
Worldwide, approximately 50 million people suffer from dementia 3 and the global prevalence is increasing. 3 Understanding the factors associated with rapid progression is important not only for designing intervention studies but also for ensuring optimal follow-up protocols and quality of life for patients.
We lack robust methods for predicting the progression rate of dementia, preventing us from providing accurate prognoses to patients. Several risk factors have been associated with rapid dementia progression, but comparisons across studies are hampered by the differences in population, the statistical methods used, and the risk factors assessed. 4 Additionally, dementia is often multifactorial, 5 and the progression rate is likely to be affected by multiple components.
However, studies typically evaluate risk factors separately, and few assess the effect of multiple determinants.
NPS are common in all dementia aetiologies and can be present at every stage of cognitive decline. 6,7 They have been associated with impairment in activities of daily living, 8 nursing home admission, 9 decreased quality of life, 10 greater caregiver burden 11 and increased informal care costs in dementia. 12 Previous progression studies link more-severe NPS with rapid progression of Alzheimer's disease (AD) 13,14 but also with more-advanced stages of dementia. 15,16 Notably, various NPS tend to cluster, indicating correlation and, possibly, shared pathology within symptom clusters. 15,17 The number of comorbidities affecting an individual could affect the dementia progression, 18,19 at least at short-time follow-up.
However, specific comorbid diseases might have distinct associations with the dementia syndrome and, thereby, may influence progression rates differently. 20 For instance, a history of hypertension or congestive heart failure has been associated with slower cognitive decline, 21 whereas a history of diabetes has been associated with slower functional and cognitive decline. 14,21 Additionally, a history of hypertension or psychiatric disorders has been associated with more NPS at the time of dementia diagnosis, 21 and the number of comorbidities might also be associated with more-severe NPS. 22 The effect of cerebrovascular comorbidities on dementia progression is not clear. 5,21,23 We seek to increase the knowledge about distinct predictors of dementia progression by building on previous results where we found that more-severe overall NPS, but not the number of comorbidities, were associated with rapid progression of all-cause dementia in a memory clinic cohort. 24 Due to the clustering of NPS, together with differences seen in relation to neuropathology and dementia severity, 6 , 7 we hypothesize that NPS clusters are valuable predictors of dementia progression rates. By including comorbidity data from the Norwegian Patient Registry, 25 we also evaluated the effect of specific comorbidities present before and at the time of the dementia diagnosis. We hypothesize that individual comorbidities will have different effects on the dementia progression rate. The findings may improve future planning for patients, and their caregivers, and may facilitate the selection of patients eligible for intervention studies.

| Subjects
The study population has been described in greater detail. 24 In brief, the patients were included in the Norwegian registry of persons assessed for cognitive symptoms (NorCog) between 12 January 2009, and 31 July 2016, at the Memory Clinic, Oslo University Hospital, were diagnosed with dementia and received at least one follow-up examination after the baseline period (6 months; number of patients [N] = 442). We restricted the follow-up period to 3 years to limit survival bias, and the participants were followed for an average of 2.2 years (standard deviations [SD] 1.5).
All participants signed an informed consent form; the project was approved by the Regional Ethics Committee (2015/1510 REK vest) and was conducted in accordance with the Helsinki Declaration.

| Assessments
Examinations were part of the standard practice at the clinic, and patients underwent an average of 3.5 (SD 1.7) consultations with varying time intervals. As a measure of cognitive and functional impairment, the Clinical Dementia Rating Scale (CDR) 26 was scored post hoc by the researchers, using all available information from the patients' records. Based on the severity of cognitive decline, a score of 0-3 was assigned within the following categories: memory, orientation, judgement and problem-solving, community affairs, home and hobbies and personal care. -961

| Diagnostic workup
At baseline, the NorCog research protocol 27 was used to assess the patients. 24 In summary, the NorCog protocol includes information on demographics, comorbidities, medication use, dementia symptoms, physical examination and blood sampling. The protocol also includes cognitive test results, for example, the MMSE 28 (0-30; lower values indicate greater cognitive impairment) and the Trail Making Test A (TMT-A) 29  At the baseline examination, the Neuropsychiatric Inventory Questionnaire (NPI-Q), 35 a brief questionnaire version of the Neuropsychiatric Inventory, was used to assess NPS. 36 The NPI-Q severity score (0-3; a higher score indicates a more-severe symptom) of the 12 NPS was included. These symptoms were as follows: delusions, hallucinations, agitation/aggression, depression/dysphoria, anxiety, elation/euphoria, apathy/indifference, disinhibition, irritability/lability, motor disturbance, night-time behaviour and appetite/ eating disturbances. For participants with two or more missing items, the NPI-Q was set to missing (N = 39).
By regulation, the specialist health care service in Norway 25 must register diagnoses in the Norwegian Patient Registry according to the 10th revision of the International Statistical Classification of Diseases and Related Health Problems (ICD-10) codes. 37 We selected the disease categories of interest based on the results of a systematic review 38 including those categories listed in the NorCog protocol. 27 For each patient, we searched both the NorCog and the Norwegian Patient Registry (from 1 January 2008 up until baseline) for comorbidities (present or absent) within the 12 selected disease categories ( Table 2). If a comorbidity was registered in either of these registries, it was denoted as present. The comorbidities were registered up to 7 years prior to baseline, but because of the chronic nature of the conditions, they were assumed to have been present at the time of the dementia diagnosis.

| Outcome measures
We have previously described the scoring of the CDR scales and the selection of trajectory groups in this cohort. 24 In summary, the CDR scales were scored post hoc by a certified rater using all available information from standardized and comprehensive patient records.

| Factor analysis
To identify NPS clusters, principal component analysis was applied to the 12 items of the NPI-Q scale, with a patient-symptom severity ranging from 0-3, using SPSS version 26. The correlation matrix revealed coefficients above 0.3. 41 The sampling adequacy, using

| Multinomial logistic regression analyses
Multinomial logistic regression analyses with the three-level trajectory group membership as the outcome variable and using the slow-progression group (Group 1) as reference were performed using Stata/IC 15.1. In multinomial logistic regression, the outcome variable Y has more than two categories, and the coefficient is, therefore, a relative risk ratio (RRR) in contrast to the odds ratio of a standard logistic regression. Spearman intercorrelation ensured that the inter-correlation between the explanatory variables were ≤0.55.
Based on previous results related to this cohort, 24 MMSE and TMT-A were used to adjust for cognitive function. In addition, all models were adjusted for age, sex, education and dementia aetiology. First, we ran preliminary models to consecutively test individual NPS clusters (continues severity score; Table S2) and comorbidity groups (present = 1 or absent = 0; Table S3). Covariates with p ≤ 0.2 were retained for further analysis in the final multivariate model (Table 3).
Patients without missing values were included in the multinomial logistic regression analyses (N = 380).

| RESULTS
Sample characteristics at the time of diagnosis are listed in Tables 1   and 2. The patients with frontotemporal dementia (FTD) showed more agitation and those with Lewy body dementia (LBD) showed more psychotic symptoms (Table S1). From the preliminary models, all of the NPS clusters (Table S2) and six of the comorbidity groups (Table S3) were included in the final model.

| Predictors associated with rate clinical progression
Results from the final model are presented in Table 3

| DISCUSSION
In the present study, we found that psychotic and affective symptoms at the time of dementia diagnosis increased the risk for rapid clinical progression, even after adjusting for demographic information, dementia aetiology, cognitive function and comorbidities. Previous diagnoses of mental and behavioural disorders (excluding dementia) were associated with slow progression.
Psychotic symptoms at the time of diagnosis were associated with rapid progression in our cohort. Patients with LBD had more psychotic symptoms, but the overall association remained significant, even after adjusting for dementia aetiology. Likewise, Gerritsen and colleagues 42 found that psychotic symptoms were associated with cognitive decline in young-onset dementia (AD, vascular dementia [VaD] and FTD). Moreover, researchers also found psychotic symptoms to be associated with rapid cognitive decline 43 find that psychotic symptoms were associated with rapid progression in AD. 45 One explanation may be that their results were attenuated by adjusting for baseline dementia severity, even though changes in the same variables were their outcome measures.
Affective symptoms at the time of diagnosis were associated with the most rapidly progressing group in our cohort. Affective symptoms have been shown to increase the risk of AD, 46 but their effect on dementia progression rate has been less studied. A review 47 found that, in AD, depression was associated with rapid cognitive decline, greater functional decline, and higher rates of nursing home placement. Affective symptoms have also been associated with mortality. 43 However, other studies did not find any associations between affective symptoms and rapid progression of dementia 14,18 and one study indicated affective symptoms to be associated with slower cognitive decline in AD, VaD and FTD. 42 This discrepancy may be due to the overlap between symptoms of AD and depression and because depressive symptoms often occur with other NPS, making the diagnosis challenging and the comparison of studies difficult. 47 When exploring individual NPS in AD, researchers have found agitation/aggression, 14,16,43 aberrant motor behaviour, 14,16 and any clinically significant NPS 43 to be associated with rapid clinical progression. We did not find such associations.
Whether psychotic and affective symptoms reflect higher neurotoxicity in the brain remains unclear. Studies have, however, indicated several neuropathological changes linked to psychosis and to affective symptoms in dementia. For instance, AD patients with psychotic symptoms have been found to have more neurofibrillary tangles in neocortical areas independent of dementia severity, 48 as well as more intraneuronal P-tau. 49 Delusions have been associated with an upregulation of postsynaptic muscarinic receptors in dementia with Lewy bodies 50 and increased striatal dopamine receptor availability in AD. 51 Genetic variations could also have an impact on psychosis in AD, 52,53 but apolipoprotein E ε4 does not seem to increase the risk. 54 Regarding affective symptoms, a literature review found that depressive symptoms in AD were associated with more cerebral neurofibrillary tangles and senile plaques, hippocampus atrophy and changes in the serotonergic pathways. 55 Moreover, alterations in the monoaminergic network in the cortex 56 and the number of corticotropin-releasing hormone neurons 57 have been associated with depression in dementia. Furthermore, apathy has been linked to increased P-tau in cerebrospinal fluid and to neurofibrillary tangles in the anterior cingulated cortex in AD. 55 The neuropathological changes related to NPS in FTD and other dementias are much less understood, but psychotic symptoms have been found to correlate with grey matter atrophy in progranulin mutation carriers with FTD. 58 The presence of previous mental and behavioural disorders was associated with slow progression, while psychotic and affective symptoms at baseline were associated with rapid progression in our cohort. This could indicate differences in underlying processes.
Psychosis in AD and VaD differs from psychosis seen in schizophrenia, with the dementia syndromes being associated with mostly persecutory and misidentification delusions and visual hallucinations. 59 A study examining the interaction between psychosis in AD and schizophrenia found a possible genetic link, but the association was strongest for delusions. 53  -965 could be that these patients develop symptoms of dementia on the basis of less neuropathology because of reduced cognitive reserve. 61 Again, this may explain why the dementia progressed more slowly, as progression rate is known to increase with dementia severity. 14,24,62 To elaborate on this, we need further research, preferably combining clinical markers with biomarkers.
A trend was identified, although not reaching statistical significance, for the association of musculoskeletal system and connective tissue disorders with a slower progression rate of dementia (RRR 0.44 [95% CI 0.19-1.02], p = 0.055), and these conditions often result in the use of non-steroidal anti-inflammatory drugs (NSAIDs). An older longitudinal study found that NSAID usage was associated with slower progression of decreased verbal fluency, spatial recognition and orientation in AD. 63 Longitudinal studies have shown that the use of NSAIDs, especially long-term use, 64 could reduce the risk of AD development. 65 The mechanism of action is uncertain, but the attenuation of over-stimulated microglia has been suggested. 64 Clinical trials have, however, failed to prove any benefit of NSAIDs in the prevention of AD. 64,66 It is not clear why epidemiological studies find this association, while clinical trials do not. Possibly NSAID usage is a marker of some unknown residual confounding, or that the potential effect of NSAIDs is difficult to prove in prevention trials. 66 We do not have information on the use of medication in our cohort, and, therefore, we cannot know if these diagnoses reflect NSAID use.
Perhaps, these diagnoses are proxies of another mechanism, thereby biasing the result.
The presence of endocrine, nutritional and metabolic diseases seemed to increase the risk of rapid progression, but again, this was a trend that was not statistically significant (

| Strengths and limitations
The patients included in the present study were from a specialized memory clinic, which may weaken the generalizability of the results.
We did, however, include all-cause dementia reflecting numerous underlying pathologies, although nearly 73% of the patients had AD. have also found NPS grouped in similar symptom clusters, 15 indicating that they might be clinically meaningful.

| CONCLUSION
We have shown that psychotic and affective symptoms at the time of dementia diagnosis were associated with rapid progression in allcause dementia. Previous diagnoses of mental and behavioural disorders were associated with slow progression. Our results contribute to the understanding of risk factors for dementia progression and show that multiple factors might affect the progression rate. This supports the need for individual assessments and follow-up protocols.