Risk of dementia associated with cardiometabolic abnormalities and depressive symptoms: a longitudinal cohort study using the English longitudinal study of ageing

Objectives Depression and cardiometabolic abnormalities are independently associated with a high risk of dementia. This study aimed to examine the association of comorbid depressive symptoms and cardiometabolic abnormalities with risk of dementia. Methods The sample comprised 4859 participants aged 50 or older without baseline dementia who took part in the English Longitudinal Study of Ageing (waves 2‐7). Depressive symptoms were assessed using the Center for Epidemiologic Studies‐Depression tool. Cardiometabolic abnormalities were defined as three or more cardiometabolic risk factors (inflammation, central obesity, raised triglycerides, low high‐density lipoprotein [HDL] cholesterol, hypertension, and hyperglycaemia or diabetes). Participants were classified into four groups based on presence of depressive symptoms and cardiometabolic abnormalities. Results were analysed using the Cox proportional hazards regression adjusted for covariates. Results A total of 216 cases of incident dementia were reported over 10 years of follow‐up. The group with high depressive symptoms only had an increased hazard of developing incident dementia during follow‐up (HR = 2.68; 95%CI, 1.70‐4.23), which was attenuated after adjustment for baseline cognition. No evidence was found for an association of overall cardiometabolic abnormalities with incident dementia; though hyperglycaemia, hypertension, and abdominal obesity with depressive symptoms had an unadjusted association with incident dementia. Only low‐HDL cholesterol with depressive symptoms had an adjusted association with incident dementia (HR = 0.18; 95%CI, 0.04‐0.75). Conclusions This work confirms depressive symptoms as a risk factor for incident dementia. However, low HDL‐cholesterol with depressive symptoms may be protective against dementia, though more work is required to confirm this association.


| INTRODUCTION
Dementia is a neurodegenerative disorder of later life characterised by progressive deterioration of cognitive function, beyond what might be expected from normal ageing. 1 As our population ages, it is projected that the number of people affected by dementia will increase. 2 There is currently no cure and neuropathological mechanisms, underlying dementia, take place over many years before the overt dementia symptoms emerge. 3 Recently, great attention has been paid to the effect of cardiometabolic abnormalities as a potential trigger of dementia. 4,5 Cardiometabolic abnormalities include cardiovascular risk factors such as central obesity, raised triglycerides, low high-density lipoprotein (HDL) cholesterol, hypertension, high blood glucose, and increased inflammation. 6,7 However, the evidence-base regarding the association of cardiometabolic abnormalities with cognitive decline and incident dementia is mixed. [8][9][10][11][12] Some studies indicate cardiometabolic abnormalities increase the risk of dementia, 9,12 whereas others find no association. 10,11 Some researchers, instead, propose that individual markers of cardiometabolic abnormalities are linked with dementia rather than combined cardiometabolic abnormalities. 11 Given that there are mixed findings about the relationship of each cardiometabolic factor and their additive impact on risk of dementia, more research is needed to determine this potential link.
Cardiometabolic abnormalities have been shown to be strongly linked with depression, 13 which is also an independent risk factor for incident dementia. 14,15 Previous work suggests that the co-occurrence of cardiometabolic abnormalities and depression might represent a subtype of depression called "metabolic depression." 16,17 This comorbidity is associated with lifestyle and health-related risk factors (eg, poor self-care) for cardiovascular and metabolic diseases. 18,19 These cardiovascular and metabolic diseases have been shown to be risk factors for development of future dementia. 20 Furthermore, work shows that co-occurring cardiometabolic abnormalities and depressive symptoms are risk factors for cognitive decline. 21 However, to the best of our knowledge, no previous work has examined whether co-occurring cardiometabolic abnormalities and depression might be associated with incident dementia. The aim of the present study was to examine the impact of elevated depressive symptoms and cardiometabolic abnormalities on the incidence of dementia in a large community-dwelling sample of adults aged 50 years and older for more than 10 years. complete data on depressive symptoms, with valid blood sample, and no more than two missing values for the six cardiometabolic risk factors assessed. The study restricted the analyses to participants who attended at least one subsequent follow-up session (wave 3-7) after the baseline assessment (wave 2). Participants with dementia or Alzheimer's disease (AD) at or before the baseline assessment were excluded.
All participants provided signed full informed consent, and ethical approval for all the ELSA waves was granted by the London multicentre research ethics committee.
The outcome variable was defined as time-to-dementia from the index date identified as the date of the wave 2 interview. The incidence of dementia was determined during the follow-up period between waves 3 to 7. This was determined in one of two ways. Firstly, the identification of newly diagnosed dementia based on participant or informant reported physician-diagnosed dementia or AD. Date to time of event was determined by self-report, or where date of diagnosis was not provided. This was calculated as the midpoint between the previous interview and current interview in line with previous work. 20

Key points
• High depressive symptoms without cardiometabolic abnormalities are associated with an increased hazard of dementia.
• Cardiometabolic abnormalities (with or without depressive symptoms) are not associated with an increased hazard of dementia.
• Individual cardiometabolic abnormalities with depressive symptoms (such as hyperglycemia, hypertension, and abdominal obesity) are associated with an increased unadjusted hazard of dementia; though these associations are attenuated after adjustment.
• After adjusting for all confounders low HDL-cholesterol with depressive symptoms are predictive of a decreased risk of developing dementia.
Those people who did not develop dementia during the follow-up or before they were lost to follow-up were treated as censored.
The study did not distinguish between the reasons for losses to follow-up (eg, due to death).

| Depressive symptoms
The 8-item Center for Epidemiologic Studies-Depression (CES-D) tool was used to evaluate self-reported depressive symptomology.
Participants were asked to respond with a binary option (yes/no) whether they had experienced different symptoms of depression, such as much of the time feeling sad during the past week. The total sum score ranged from 0 to 8. Participants were dichotomized into those with "no or low" (CES-D < 4) and those with "high" (CES-D ≥ 4) depressive symptomatology, in line with previous ELSA studies. 19,25

| Cardiometabolic abnormalities
Six cardiometabolic risk factors were evaluated including C-reactive protein (CRP), central obesity, raised triglycerides, low HDL cholesterol, hypertension, and hyperglycaemia or diabetes. These risk factors were chosen as they have been used in previous work to identify cardiometabolic abnormalities. 18,19,25 Central obesity was defined as having waist circumference of greater than or equal to 94 cm for European men and greater than or equal to 80 cm for European women. Raised triglycerides were greater than or equal to 1.7 mmol/L and reduced HDL cholesterol is less than 1.0 mmol/L in males and and less than 1.3 mmol/L in females.
Hypertension risk was defined as having systolic blood pressure of greater than or equal to 130 mm Hg or diastolic blood pressure of greater than or equal to 85 mm Hg, or use of antihypertensive treatment. Hyperglycaemia was defined as having raised fasting plasma glucose (FPG) of greater than or equal to 5.6 mmol/L and/or having a previous diagnosis of type 2 diabetes. The cutoffs used for obesity, triglycerides, HDL-cholesterol, hypertension, and hyperglycaemia were based on the International Diabetes Federation (IDF) criteria. 6 Finally, systemic inflammation was determined as having CRP levels of greater than 3 mg/L, according to the American Heart Association and the Center for Disease Control. 26 Participants were categorised as having cardiometabolic abnormalities if they exhibited a minimum of three of the above cardiometabolic risk factors in line with previous work. 19

| Clinical variables
Self-reported physician-diagnosis of cardiovascular comorbidities (angina, myocardial infarction, congestive heart failure, heart murmur, arrhythmia, and stroke) were also recorded. Reporting one or more of these conditions was defined as cardiovascular comorbidity.

| Lifestyle variables
Self-reported health-related behaviours included physical activity measured using the self-reported participation in vigorous, moderate or low impact physical activities (sedentary or low/moderate/high see previous study by Smith et al 27 for more information on categorisation), and self-reported smoking status (never/former/current).

| Cognitive function
Cognitive function was defined as the total cognitive index derived from the total scores on memory and executive tests and ranged from 0 to 50. 22 Specifically, memory tests included a word list learning task, which measured verbal learning, immediate and delayed memory (recall of 10 nouns), and a prospective memory task (remembering to remember). The combined scores from memory tests ranged from 0 to 27. Executive function tests measured verbal semantic fluency (naming as many animals as possible in 1 min) and attention (letter cancellation task). The combined scores from the executive function index ranged from 0 to 23.

| STATISTICAL ANALYSIS
The study sample was categorised into four groups based on CES-D scores (less than 4 low-to-no depressive symptomatology versus greater than or equal to 4 elevated depressive symptomatology) and cardiometabolic abnormalities (less than 3 absence of cardiometabolic risk factors versus greater than or equal to 3 presence of cardiometabolic risk factors) in line with previous studies. 19,25 The first group had no or low depressive symptoms and no cardiometabolic abnormalities.

| Sensitivity analysis
Three sensitivity analyses were conducted in order to assess the robustness of the findings of the primary analysis. To minimise the possibility that depressive symptoms occur as a prodrome of dementia, we conducted a sensitivity analysis that repeated the above proportional hazard models, excluding incident cases of dementia during the two-year period after baseline (wave 3). This approach to assessing depression as a prodrome has also been used in previous research. 28 We conducted a second sensitivity analysis removing those with cardiovascular comorbidity at baseline. This analysis is based on the hypothesis that a cardiovascular comorbidity may influence the risk of developing dementia in older age. 29 A third sensitivity analysis evaluated robustness of the findings against the reliability of dementia diagnosis by excluding cases

| Results
A total of 7666 individuals participated in the nurse baseline assessment in wave 2. However, the final analytical sample comprised 4859 participants free of dementia at baseline who had sufficient data to be placed in one of the four groups (see Figure 1 for participant flowchart).
The sociodemographic, health, lifestyle, and cognitive characteristics of the analytical sample according to the four depressive symptom and cardiometabolic abnormalities groups are shown in Table 1.
Notably, having both high depressive symptoms and cardiometabolic abnormalities was significantly related to being older, female, and separated. This group were less likely to have attained a higher education and be from a wealthier quintile. Participants with depressive symptoms only were also more likely to be female, less educated, and less wealthy. Participants with comorbid depression and cardiometabolic abnormalities were significantly more likely to have reported at least one cardiovascular comorbidity, as compared with other groups. As well, co-occurrence of depression and cardiometabolic risk factors was significantly associated with lower cognitive performance compared with other three groups. Similarly, participants with high depressive symptoms only were more likely to suffer from comorbid cardiovascular disease and to have lower cognitive ability, compared with the reference and cardiometabolic abnormalities groups.  Table 2).

| Stratification by individual cardiometabolic markers
Across all stratified analyses, high depressive symptoms alone remained a significant predictor of incident dementia. However, there were some observable differences in results based on different groupings of individual cardiometabolic abnormalities (see Table 3). Groups with both high depressive symptoms and hyperglycemia/diabetes (HR = 2.50; 95% CI, 1.31-4.76) and high depressive symptoms with abdominal obesity (HR = 1.72; 95% CI, 1.05-2.81) had an increased hazard of incident dementia compared with their respective reference groups (see Table 3). Furthermore, groups with either hypertension alone (HR = 1.86; 95% CI, 1.29-2.69) or both hypertension and depressive symptoms (HR = 2.92; 95%CI, 1.79-4.76) also had an increased hazard of incident dementia compared with a group with no hypertension and no depressive symptoms. However, these associations were attenuated after adjustment for all covariates (see Table 3 ). Conversely, in the fully adjusted model, low HDL-cholesterol and depressive symptoms were associated with a decreased risk of incident dementia compared with the reference group (HR = 0.18; 95% CI, 0.04-0.75).

| SENSITIVITY ANALYSES
The sensitivity analysis that excluded dementia diagnoses in the twoyear period after baseline resulted in a loss of 37 cases of dementia. In this analysis, the increased risk of dementia associated with depressive symptoms alone was slightly lower in both unadjusted (HR = 2.57; 95% CI, 1.54-4.29) and adjusted models for sociodemographic, lifestyle, and cardiovascular comorbidity covariates (HR = 1.79; 95% CI, 1.05-3.06). However, this association was attenuated after adjust-  Table A).
We performed an additional sensitivity analysis to assess the risk of dementia in those without cardiovascular comorbidity at baseline.
The analysis resulted in a loss of 80 dementia cases. The groups with showed a significant association with incident dementia in the unadjusted model, though these associations were attenuated after adjusting for confounders. No significant associations were found for the group with both depressive symptoms and cardiometabolic abnormalities for either the unadjusted or adjusted models (see Supporting Information Table B).
A third sensitivity analysis excluded cases diagnosed using the IQCODE and limited the analysis to self-reported physician-diagnosed dementia. This resulted in a loss of 23 events of dementia in the follow-up period. The group with high depressive symptoms but no cardiometabolic abnormalities was the only group that was significantly associated with dementia incidence in the unadjusted model (HR = 2.14; 95% CI, 1.27-3.58) (see Supporting Information Table C) though this association was attenuated in the fully adjusted model. This study lends further support that depression may be a potential risk factor for dementia. The findings of the present study are in agreement with meta-analyses, which demonstrate that individuals with depression are more than double at the risk of developing dementia. 14,15 In our sensitivity analysis, that excluded dementia cases in the 2 years after baseline, depressive symptoms alone were associated with a slightly lower risk of dementia than was observed in the main analysis, however it did not substantially affect estimates.
Furthermore, our sensitivity analysis that excluded dementia cases diagnosed via IQCODE still showed high depressive symptomatology as a potential risk factor for developing subsequent dementia, although again, this association was only detected in unadjusted analyses. At present, the possible mechanisms behind the role of depressive symptoms in the development of dementia remain uncertain. A possible pathway linking depression with dementia could be through neurobiological changes and neuronal brain damage. 30 Previous research suggests the relationship between depressive symptoms and dementia may also be explained in part by cardiovascular comorbidity, in which vascular damage in the brain might predispose to depression in the elderly (vascular depression hypothesis). 29 However, our results suggest that the association between depression and incident dementia is independent of co-occurring cardiometabolic abnormalities. Furthermore, in our sensitivity analysis, which excluded those with a cardiovascular comorbidity at baseline, depressive symptoms alone were still associated with an increased risk of dementia in unadjusted analyses (which was attenuated after adjustment). This suggests that the link between depressive symptoms and dementia may also be independent of cardiovascular comorbidity, although more work is needed to confirm this.
Prior research on the possible relationship between cardiometabolic risk factors and risk of dementia reports contradictory results. [10][11][12] This discrepancy may be attributable to methodological differences and heterogeneity in the study design, population selection, criteria used for the diagnosis of dementia, thresholds used for the definition of cardiometabolic risk factors, and differences in confounder adjustment. 31 Notably, vascular dementia (VaD), rather than AD and other dementias, is the subtype of dementia most commonly linked with cardiometabolic risk factors. 32,33 However, we were unable to assess this subtype of dementia within our study. Our   research is in agreement with those studies that find no association between cardiometabolic abnormalities and risk of future dementia. 10,11,34 The only significant adjusted result that we uncovered in this study regarding cardiometabolic abnormalities was that those people with low HDL-cholesterol (indicative of hypercholesterolemia) and depressive symptoms were less likely to develop incident dementia. Some previous work has suggested that hypercholesterolemia may be protective against dementia risk. 11,35 Previous work has shown that lower blood fats are linked with an increased risk of mortality in older adults, and this is speculated to be due to low blood fats being linked with markers for poorer health such as subclinical disease and frailty. 36 Thus, it is possible that hypercholesterolemia is protective because low levels of cholesterol are linked with other risk factors for dementia. This could be examined in future work.
Previously, researchers have suggested that investigating cardiometabolic abnormalities as a whole might not offer any additive value in the prediction of dementia because of individual indicators of cardiometabolic abnormalities working in opposing directions. 37,38 However, this assumption is complicated by the fact that different studies will find opposing results when examining the same cardiometabolic abnormality. For instance, some studies have found that abdominal obesity, 11 hypertension, 39 and high lipids 11,35 are protective against risk of dementia. Other studies find that obesity, 40,41 hypertension, 42 diabetes, 43 and high lipids 44 are risk factors for incident dementia. Finally, some studies find no association between any individual cardiometabolic risk factors with incident dementia. 10 Therefore, there is a need for a lot more work to determine how cardiometabolic abnormalities as a whole and individually are linked with the risk of dementia and the effect of modifiers that may explain why these differences exist.
We found that the co-occurrence of high depressive symptoms Survival bias is also a perennial concern in longitudinal studies examining older individuals. It is possible that the oldest groups with cardiometabolic abnormalities died before developing dementia, and the sample may be representative of healthy survivors who are less susceptible to dementia caused by cardiometabolic risk factors. 11 There is also the possibility that many cases of dementia may have been censored as people developing this disorder could have been more likely to drop out. Thus, there is a possibility of underestimation of the association between depression, cardiometabolic abnormalities, and dementia incidence. ing the underlying mechanisms between these two conditions will not only provide substantial insights into the causes of dementia but will also inspire novel strategies for preventing and treating dementia.

| CONCLUSION
Overall, this study provides additional evidence for the link between depression and future risk of dementia, supporting that older individuals with high depressive symptoms are associated with an increased risk of subsequent dementia. However, cardiometabolic abnormalities both with and without co-occurring depressive symptoms were not linked with incident dementia.
Study of Ageing (ELSA) data. ELSA was created by a team of researchers based at University College London, the Institute of Fiscal Studies, and the National Centre for Social Research. The developers and funders of ELSA and the UK Archive do not bear any responsibility for the study design, the analyses or interpretations presented in the present study. We would also like to thank Professor Chris Fife-Schaw for reading over the draft manuscript and providing feedback.