Cardiometabolic multimorbidity accelerates cognitive decline and dementia progression
Abstract
Introduction
Cardiometabolic diseases (CMDs) have been individually associated with adverse cognitive outcomes, but their combined effect has not been investigated.
Methods
A total of 2577 dementia-free participants 60 years of age or older were followed for 12 years to observe changes in cognitive function and to detect incident cognitive impairment, no dementia (CIND) and dementia. CMDs (including type 2 diabetes, heart disease, and stroke) were assessed at baseline through medical records and clinical examinations. Cardiometabolic multimorbidity was defined as the presence of two or more CMDs. Data were analyzed using multi-adjusted linear mixed-effects models, Cox regression, and Laplace regression.
Results
CMD multimorbidity was associated with cognitive decline, CIND (hazard ratio [HR] 1.73; 95% confidence interval CI 1.23 to 2.44), and its progression to dementia (HR 1.86; 95% CI 1.17 to 2.97). CMD multimorbidity accelerated the onset of CIND by 2.3 years and dementia by 1.8 years.
Conclusions
CMD multimorbidity accelerates cognitive decline and increases the risk of both CIND and its conversion to dementia.
Highlights
- We explored the combined impact of cardiometabolic diseases (CMDs) on cognition.
- An increasing number of CMDs dose-dependently accelerated cognitive decline.
- CMD multimorbidity increased the risk of both cognitive impairment and dementia.
- Co-morbid CMDs could be ideal targets for interventions to protect cognitive health.
1 INTRODUCTION
Cardiometabolic diseases (CMDs) including type 2 diabetes (T2D), heart disease (HD), and stroke are well-established independent risk factors for dementia.1 The adverse impact of CMDs on cognition is also apparent in the preclinical and prodromal phases of dementia. T2D,2-4 HD,5-8 and stroke9, 10 have been individually associated with both accelerated cognitive decline and an increased risk of cognitive impairment in several longitudinal studies.
The co-occurrence of multiple chronic CMDs has become increasingly common with population aging, and one of three older adults are affected by at least two co-morbid CMDs.11 Moreover, concurrent CMDs have been shown to interact with each other, increasing the risk of negative health-related outcomes.12, 13 Despite this, most research on the link between CMDs and cognitive aging has focused on the impact of single diseases, overlooking their frequent co-occurrence.
To this end, the 2020 Lancet Commission on Dementia Prevention, Intervention, and Care highlighted, for the first time, the need to consider combinations of different cardiovascular and metabolic risk factors in relation to dementia and cognitive decline.1 Previous studies from our research group have described the association between cardiovascular multimorbidity and dementia14 and have demonstrated that the presence of CMDs increases dementia risk in a dose-dependent fashion.15 In addition, a few studies have reported associations between CMDs or CMD-related risk factors and poorer cognition.16-19 However, the combined impact of CMDs on the risk of cognitive impairment remains under-investigated, and the effect of cardiometabolic multimorbidity on the continuum of cognitive phenotypes leading up to dementia has not been examined in detail.
Here, we examined the impact of CMDs on (1) cognitive decline, (2) cognitive impairment, and (3) the progression of cognitive impairment to dementia, using 12-year follow-up data from a population-based cohort study of Swedish older adults.
2 METHODS
2.1 Study population
This study is based on data from the ongoing longitudinal Swedish National Study on Aging and Care-Kungsholmen (SNAC-K).20 The study enrolled older adults living in central Stockholm, randomly sampled from 11 age groups (60, 66, 72, 78, 81, 84, 87, 90, 93, 96, and 99+ years). A total of 3363 individuals participated in the baseline examination (2001 to 2004). Participants were followed for a maximum of 12 years (until 2016), with follow-up examinations occurring every 6 years for younger age cohorts (<78 years) and every 3 years for older age cohorts (≥78 years), given the more rapid changes in health expected with advanced age (Figure S1).
Of all participants at baseline, we excluded individuals with prevalent dementia (n = 240); Parkinson disease, schizophrenia, or developmental disorders (n = 58); and type 1 diabetes (n = 20). We additionally excluded participants with missing information on baseline dementia (n = 10) or cognitive impairment, no dementia (CIND; n = 458), leaving a total of 2577 dementia-free participants for the current study.
Participants were divided into a cognitively intact cohort (n = 1873 [72.7%]) and a CIND cohort (n = 704 [27.3%]) according to baseline cognitive status. Over the 12-year follow-up, changes in cognitive function and incident CIND and dementia were identified. In the cognitively intact cohort, 490 people (26.2%) died and 399 (21.3%; retention rate = 78.7%) dropped out of the study, leaving 984 remaining at the end of follow-up. In the CIND cohort, 324 (46.0%) people died and 148 (21.0%; retention rate = 79.0%) dropped out, leaving 232 remaining at the end of follow-up (Figure S2).
SNAC-K was approved by the Ethical Committee at Karolinska Institutet and the Regional Ethical Review Board in Stockholm, Sweden. Written informed consent was obtained from all participants, or a proxy, in the case of severe cognitive impairment.
RESEARCH IN CONTEXT
-
Systematic Review: Authors searched the literature using the PubMed database and references from relevant articles. Although there is ample evidence that type 2 diabetes, heart disease, and stroke are individually associated with cognitive decline and cognitive impairment, the combined impact of these cardiometabolic diseases (CMDs) when co-morbid remains under-investigated.
-
Interpretation: In this 12-year population-based longitudinal study, we observed a dose-dependent acceleration of cognitive decline and increased risk of cognitive impairment and dementia with the presence of an increasing number of CMDs. Moreover, CMD multimorbidity anticipated the development of both cognitive impairment and dementia by ≈2 years compared to individuals with no CMDs.
-
Future Directions: Future research should explore the mechanisms through which CMDs impact cognitive deterioration in the preclinical and prodromal phases of dementia.
2.2 Data collection
The SNAC-K protocol has been described in detail previously.20 Briefly, at each wave, trained nurses and physicians collected information on sociodemographic and lifestyle factors, anthropometrics, medical conditions, medication use, and cognitive function through structured interviews and clinical examinations. In addition, peripheral blood samples were collected for laboratory tests, including glycated hemoglobin (HbA1c) measurement and apolipoprotein E (APOE) genotyping.
Education level was assessed as the maximum years of formal schooling, and trichotomized as primary school, secondary school, or university. Smoking status was recorded as non-smoker versus current/former smoker. Alcohol consumption was recorded as no/occasional drinking versus drinking (including light-to-moderate and heavy drinking). Physical activity was assessed based on the World Health Organization recommendations and dichotomized as active (light-to-intense exercise several times per week) versus inactive (≤2 to 3 times per month).21 Body mass index (BMI) was calculated as the ratio of weight to squared height (kg/m2) and categorized as underweight (<20), normal weight (≥20 to <25), overweight (≥25 to <30), or obese (≥30).22
High cholesterol and hypertension were diagnosed based on physicians’ examinations, medical history, medication use, laboratory tests, and linkage with the Swedish National Patient Register (NPR).23 Depression was diagnosed according to the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) revised criteria. Participants’ vital status was ascertained using data from the Swedish Cause of Death Registry.
2.3 Cardiometabolic diseases
CMDs—including T2D, HD, and stroke12—were assessed at baseline using data from multiple sources. T2D was ascertained on the basis of self-reported medical history, use of glucose-lowering medications, medical records from the NPR, or HbA1c ≥6.5%, in accordance with the American Diabetes Association's standard diagnostic criteria.24, 25 Participants with prediabetes (HbA1c 5.7% to 6.4%) at baseline who developed overt diabetes at the first follow-up and did not later revert back to prediabetes or normoglycemia (n = 90) were also identified as having T2D, given the chronic, progressive nature of the disease. HD was defined as the presence of atrial fibrillation, heart failure, or ischemic heart disease based on the medical history and medical records.23 Finally, stroke was identified based on clinical examinations, medical history, and medical records.15 (See Supplementary Appendix A for more detail.)
CMD status was defined according to participants’ total number of CMDs (i.e., T2D, HD, or stroke) and categorized as CMD-free, single CMD, or CMD multimorbidity (i.e., two or more co-morbid CMDs).
2.4 Cognitive decline
At each study wave, a trained psychologist administered a battery of neuropsychological tests26 addressing performance in cognitive functions including episodic memory (free recall, recognition),27 perceptual speed (digit cancellation, pattern comparison),28, 29 verbal fluency (category and letter fluency),30 and semantic memory (vocabulary).31 The raw cognitive test scores were Z-transformed using the baseline mean and standard deviation (SD) as the standardization base. A global cognition score was then computed by taking the mean of the standardized scores for individuals with information available for at least half of the included cognitive domains.
2.5 Cognitive impairment, no dementia (CIND)
Cognitive impairment was defined as CIND, which refers to the presence of objective cognitive deficits that do not meet the threshold for dementia.32 The operationalization of CIND has been described previously for SNAC-K.33 First, raw scores from the individual cognitive tests in the neuropsychological battery were standardized into Z-scores using baseline means and SDs from an internal random sample of cognitively healthy participants. Next, five cognitive domains were defined by averaging the Z-scores of multiple tests or using individual test Z-scores: episodic memory (free recall),27 executive function (Trail Making Test-B),34 perceptual speed (digit cancellation, pattern comparison),30 visuospatial abilities (mental rotation task),30 and verbal fluency (category and letter fluency).30 Based on previous reports,32, 33, 35 CIND was defined as having a Z-score ≥1.5 standard deviations below age group-specific means in at least one cognitive domain, in the absence of a dementia diagnosis. The same procedure, using the cutoffs determined at baseline, was used to identify incident CIND during follow-up.
2.6 Dementia
Dementia was clinically diagnosed according to the DSM-IV revised criteria through a validated three-step procedure.36 Briefly, two physicians independently made preliminary diagnoses of dementia based on the participant's physical, neurological, and cognitive status (steps 1 and 2). In the event of discrepancy between the two diagnoses, a senior neurologist external to the data collection process made the final determination (step 3). For participants who died during follow-up and who did not have a previous dementia diagnosis, dementia status was corroborated by medical records and/or death certificates.
2.7 Statistical analysis
Baseline characteristics of the study participants were assessed by CMD status using χ2 tests for categorical variables and one-way analysis of variance (ANOVA) for continuous variables.
First, linear mixed-effects models were used to determine the β coefficients of annual changes in global cognition as a function of CMD status at baseline. Models included baseline CMD status, linear annual follow-up time (in years), and time-by-CMD status interactions. To avoid reverse causality and to ensure the temporality of the association, participants with CIND at baseline were excluded, and this analysis was restricted to individuals in the cognitively intact cohort.
Second, Cox regression models were used to estimate the hazard ratios (HRs) of two outcomes according to CMD status: (1) incident CIND in the cognitively intact cohort and (2) incident dementia in the CIND cohort. Follow-up time was calculated as the time from study entry until CIND/dementia diagnosis, death, or last examination. The proportional hazard assumption was tested using Schoenfeld residuals regressed against follow-up time. A violation of proportionality was observed for age in the model for incident CIND. To deal with this, a Cox model stratified according to age was used, allowing different baseline hazards for the strata variables to be adjusted for without estimating the effect. No violations of proportionality were observed for the analysis of incident dementia.
Cox analyses were repeated after stratifying by baseline age (younger-old [<78 years; n = 1955] versus older-old [≥78 years; n = 622]). We further assessed the interaction between CMD status and age (as a continuous variable) by combining the variables’ cross-product term (CMD status × age) in the same model.
Finally, we used Laplace regression to model the time until the development of CIND/dementia as a function of CMD status.37 As a complement to Cox regression analysis, which estimates the risk of experiencing an event, Laplace regression estimates the difference in time to that event. This analysis therefore allowed us to quantify the absolute differences in the speed of cognitive deterioration among people with different CMD profiles. Because <30% of participants in each cohort developed the outcome, we assessed differences in the median time (in year) until the first 20% of participants in the cognitively intact cohort developed CIND and the first 20% of participants in the CIND cohort developed dementia.
All analyses were repeated after grouping participants by total number of CMDs (ranging from 0 to 3) and by specific combination of individual CMDs. To assess whether an increasing burden of CMDs impacts cognition in a dose-dependent manner, we also re-ran all models using total CMD number as a continuous variable.
In sensitivity analyses, we: (1) assessed the impact of participant attrition, (2) repeated the analyses of cognitive decline after excluding individuals who developed dementia or had no repeated measures of global cognition score available, (3) addressed the competing risk of death, and (4) accounted for reverse causality related to preclinical dementia. (More information is available in Supplementary Appendix A.)
The following factors were considered potential confounders and controlled for in all analyses: age, sex, education, BMI, physical activity, hypertension, alcohol consumption, and APOE ɛ4 carrier status (see Supplementary Appendix A for more detail). P-values <.05 were considered statistically significant. All statistical analyses were performed using Stata SE 16.0 (StataCorp, College Station, TX).
3 RESULTS
3.1 Characteristics of the study population
Baseline sociodemographic, clinical, and lifestyle characteristics of the 2577 study participants (mean age 72.4 ± 10.0, range 60 to 102 years) are shown in Table 1. At baseline, 1730 (67.1%) were CMD-free and 847 (32.9%) had at least one CMD. Compared to people with no CMDs, those with CMDs were more likely to be older, male, have lower education, have overweight/obesity, be physically inactive, and abstain from drinking alcohol. In addition, within the cognitively intact cohort, people with CMDs were more likely to have hypertension and high cholesterol. No differences among the groups were found in terms of smoking status, depression, or APOE ɛ4 carrier status.
| Cognitively intact cohort (n = 1873) | CIND cohort (n = 704) | |||||||
|---|---|---|---|---|---|---|---|---|
| CMD-free (n = 1332) | Single CMD (n = 440) | CMD multimorbidity (n = 101) | P-value | CMD-free (n = 398) | Single CMD (n = 215) | CMD multimorbidity (n = 91) | P-value | |
| Age, years | 69.3 ± 8.9 | 75.0 ± 9.6a | 77.0 ± 8.1a | <.001 | 72.6 ± 9.8 | 81.2 ± 10.2a | 78.9 ± 9.6a | <.001 |
| 60 and 66 | 782 (58.7) | 143 (32.5) | 20 (19.8) | <.001 | 169 (42.5) | 30 (14.0) | 16 (17.6) | <.001 |
| 72 and 78 | 358 (26.9) | 157 (35.7) | 47 (46.5) | 133 (33.4) | 64 (29.8) | 36 (39.6) | ||
| 81, 84, and 87 | 160 (12.0) | 108 (24.6) | 27 (26.7) | 64 (16.1) | 54 (25.1) | 20 (22.0) | ||
| 90+ | 32 (2.4) | 32 (7.3) | 7 (6.9) | 32 (8.0) | 67 (31.2) | 19 (20.9) | ||
| Women | 843 (63.3) | 224 (50.9) | 52 (51.5) | <.001 | 288 (72.4) | 137 (63.7) | 49 (53.9) | <.001 |
| Educationb | ||||||||
| Primary | 115 (8.6) | 64 (14.6) | 16 (15.8) | <.001 | 84 (21.1) | 61 (28.4) | 24 (26.4) | .047 |
| Secondary | 621 (46.6) | 206 (46.8) | 57 (56.4) | 210 (52.8) | 116 (54.0) | 52 (57.1) | ||
| University | 596 (44.7) | 170 (38.6) | 28 (27.7) | 104 (26.1) | 38 (17.7) | 15 (16.5) | ||
| Current/former smokersb | 749 (56.5) | 236 (53.9) | 55 (55.0) | .627 | 203 (51.3) | 116 (54.5) | 46 (51.1) | .734 |
| Current alcohol drinkersb | 1,045 (78.6) | 304 (69.1) | 58 (58.0) | <.001 | 234 (58.9) | 94 (44.1) | 42 (47.2) | .001 |
| Physically active | 1,078 (80.9) | 324 (73.6) | 60 (59.4) | <.001 | 278 (69.9) | 119 (55.4) | 47 (51.7) | <.001 |
| BMI (kg/m2) | 25.8 ± 3.8 | 26.1 ± 4.0 | 26.8 ± 4.1a | .014 | 25.3 ± 3.8 | 25.1 ± 4.5 | 26.4 ± 5.6 | .038 |
| Underweight (<20) | 51 (3.8) | 23 (5.2) | 3 (3.0) | .009 | 20 (5.0) | 26 (12.1) | 8 (8.8) | .004 |
| Normal (20 to ≤25) | 562 (42.2) | 153 (34.8) | 31 (30.7) | 194 (48.7) | 89 (41.4) | 35 (38.5) | ||
| Overweight (≥25 to≤30) | 552 (41.4) | 193 (43.9) | 45 (444.6) | 142 (35.7) | 75 (34.9) | 29 (31.9) | ||
| Obese (≥30) | 167 (12.5) | 71 (16.1) | 22 (21.8) | 42 (10.6) | 25 (11.6) | 19 (20.9) | ||
| Hypertension | 868 (65.2) | 340 (77.3) | 76 (75.3) | <.001 | 287 (72.1) | 164 (76.3) | 70 (76.9) | .422 |
| High cholesterolb | 667 (51.2) | 233 (54.6) | 63 (64.3) | .028 | 189 (49.4) | 95 (46.1) | 43 (50.0) | .721 |
| Depressionb | 44 (3.3) | 15 (3.4) | 5 (5.0) | .684 | 31 (7.8) | 17 (8.0) | 8 (8.8) | .955 |
| Any APOE ε4 alleleb | 364 (28.5) | 121 (29.4) | 28 (29.2) | .940 | 108 (30.1) | 58 (30.2) | 24 (28.6) | .958 |
| Global cognition scoreb | 0.31 ± 0.54 | 0.02 ± 0.55a | -0.07 ± 0.51a | <.001 | -0.39 ± 0.72 | -0.74 ± 0.69a | -0.75 ± 0.68a | <.001 |
- Note: Data are presented as means ± standard deviations or number (proportion %).
- Abbreviations: APOE ε4; apolipoprotein E ε4 allele; BMI, body mass index; CIND, cognitive impairment, no dementia.
- a Pairwise means comparison using the Bonferroni correction: P < .05 (reference group = CMD-free).
- b Missing variables: 14 were missing data on smoking, 8 on alcohol consumption, 73 on high cholesterol, 10 on depression, 157 on APOE ɛ4, and 84 on global cognition score.
3.2 CMDs and cognitive decline
Among participants in the cognitively intact cohort, the presence of an increasing number of CMDs was dose-dependently (P < .001) associated with faster annual decline in global cognition score over the follow-up (median: 5.5 years), independent of age, sex, and education level (Model 1) as well as BMI, physical activity, hypertension, alcohol consumption, and APOE ε4 carrier status (Model 2) (Table 2 and Figure 1).
| n | Model 1a | Model 2b | |
|---|---|---|---|
| β (95% CI) | β (95% CI) | ||
| CMD-free × time | 1280 | Reference | Reference |
| Single CMD × time | 440 | −0.02 (-0.02 to -0.01) | −0.02 (-0.03 to -0.01) |
| T2D × time | 120 | −0.01 (-0.02 to 0.00) | −0.01 (-0.02 to 0.00) |
| HD × time | 272 | −0.02 (-0.03 to -0.01) | −0.02 (-0.03 to -0.01) |
| Stroke × time | 48 | −0.03 (-0.05 to -0.02) | −0.04 (-0.05 to -0.02) |
| CMD multimorbidity × time | 101 | −0.03 (-0.04 to -0.02) | −0.03 (-0.04 to -0.02) |
| Two CMDs × time | 94 | −0.03 (-0.04 to -0.01) | −0.03 (-0.04 to -0.02) |
| T2D + HD × time | 56 | −0.02 (-0.04 to -0.00) | −0.02 (-0.04 to -0.01) |
| HD + Stroke × time | 32 | −0.03 (-0.05 to -0.00) | −0.03 (-0.05 to -0.01) |
| T2D + Stroke × time | 6 | −0.07 (-0.12 to -0.02) | −0.07 (-0.12 to -0.02) |
| Three CMDs (T2D + HD + Stroke) × time | 7 | −0.05 (-0.09 to -0.00) | −0.05 (-0.10 to -0.01) |
| P for dose-response trend | <0.001 | <0.001 |
- Note: Study population excludes 52 participants from the cognitively intact cohort who were missing data on global cognition score at baseline. Global cognition score is expressed in units of standard deviation relative to the baseline sample.
- Abbreviations: HD, heart disease; T2D, type 2 diabetes.
- a Adjusted for age, sex, and education.
- b Adjusted for Model 1, body mass index (BMI), physical activity, hypertension, alcohol consumption, and apolipoprotein E (APOE ε4) carrier status.
3.3 CMDs and incident CIND
During follow-up (median: 11.2 years), 539 participants (28.7%) in the cognitively intact cohort developed CIND. In multi-adjusted Cox regression models, CMD multimorbidity was associated with nearly double the risk of incident CIND (HR 1.73; 95% confidence interval [CI] 1.23 to 2.44) relative to participants who were CMD-free. CIND risk increased dose-dependently with an increasing number of CMDs (P = .011) (Table 3). In Laplace regression analysis, compared to CMD-free participants, people with CMD multimorbidity developed incident CIND 2.2 (20th percentile difference [PD]; 95% CI −3.48 to −0.91) years earlier (Figure 2). Specifically, the time to CIND was shortened by 2.0 (20th PD −1.96; 95% CI −3.14 to −0.77) and 3.7 (20th PD −3.73; 95% CI −6.53 to −0.92) years among people with two or three CMDs, respectively.
| Incident CIND in the cognitively intact cohort (n = 1873) | Incident dementia in the CIND cohort (n = 704) | |||||
|---|---|---|---|---|---|---|
| N | IR (95% CI) | HR (95% CI)a | n | IR (95% CI) | HR (95% CI)a | |
| CMD-free | 1332 | 35.6 (32.2–39.4) | Reference | 398 | 22.7 (18.1–28.4) | Reference |
| Single CMD | 440 | 44.6 (37.6–53.0) | 1.06 (0.86–1.31) | 215 | 40.8 (31.2–53.5) | 1.14 (0.78–1.67) |
| T2D alone | 120 | 44.4 (32.8–60.1) | 1.17 (0.84–1.63) | 43 | 14.1 (5.9–34.0) | 0.42 (0.16–1.07) |
| HD alone | 272 | 46.9 (37.5–58.8) | 1.08 (0.83–1.39) | 149 | 49.9 (36.6–68.0) | 1.37 (0.91–2.01) |
| Stroke alone | 48 | 33.8 (18.7–61.0) | 0.76 (0.41–1.38) | 23 | 56.3 (28.2–112.6) | 1.26 (0.59–2.72) |
| CMD multimorbidity | 101 | 66.1 (48.3–90.5) | 1.73 (1.23–2.44) | 91 | 52.5 (36.0–76.6) | 1.86 (1.17–2.97) |
| Two CMDs | 94 | 63.2 (45.4–88.0) | 1.62 (1.13–2.31) | 82 | 49.2 (32.7–74.0) | 1.71 (1.04–2.81) |
| T2D+HD | 56 | 70.8 (46.6–107.6) | 1.82 (1.17–2.82) | 51 | 46.8 (28.2–77.6) | 1.73 (0.97–3.08) |
| HD + Stroke | 32 | 63.8 (36.3–112.4) | 1.73 (0.96–3.12) | 27 | 69.0 (34.5–138.0) | 1.77 (0.80–3.93) |
| T2D + Stroke | 6 | 18.0 (2.5–127.8) | 0.38 (0.05–2.71) | 4 | - | - |
|
Three CMDs T2D + HD + Stroke |
7 | 112.3 (42.1–299.2) | 5.59 (1.97–15.87) | 9 | 86.4 (32.4–230.2) | 3.50 (1.23–9.95) |
| P for dose-response trend | 0.011 | 0.010 | ||||
- a Multi-adjusted for baseline age, sex, education, body mass index (BMI), physical activity, hypertension, alcohol consumption, and apolipoprotein E (APOE) ԑ4 carrier status.
- Abbreviations: HD, heart disease; T2D, type 2 diabetes.
3.4 CMDs and CIND's progression to dementia
In the CIND cohort, 155 participants (22.0%) developed dementia during the follow-up (median: 7.0 years). Compared to CMD-free participants, those with CMD multimorbidity had almost twice the risk of CIND progressing to dementia (HR 1.86; 95% CI 1.17 to 2.97). Dementia risk increased dose-dependently with an increasing number of CMDs (P = .010) (Table 3). Moreover, in Laplace regression analysis, the onset of dementia was anticipated by 1.84 years (20th PD; 95% CI −4.49 to 0.82) for individuals with CMD multimorbidity (Figure 2). Specifically, the time to dementia was shortened by 1.6 (20th PD1.56; 95% CI −3.38 to 0.25; P = .091) and 4.2 (20th PD −4.23; 95% CI −6.39 to −2.07) years for individuals with two or three CMDs, respectively.
3.5 Role of age in the CMD-CIND/dementia association
In analyses stratified by age (<78 years [n = 1955] vs ≥78 years [n = 622]), younger-old participants with CMD multimorbidity tended to have an increased risk of both CIND and CIND's progression to dementia, whereas the association was weaker and not statistically significant among older-old participants (Table 4). However, there was no significant interaction between CMD multimorbidity and age (as a continuous variable) on CIND/dementia risk (P = .599 and P = .581, respectively); thus the role of age in the association between CMDs and CIND/dementia might not be evident.
| Incident CIND in the cognitively intact cohort | Incident dementia in the CIND cohort | |||||||
|---|---|---|---|---|---|---|---|---|
| Younger-old (<78 years) | Older-old (≥78 years) | Younger-old (<78 years) | Older-old (≥78 years) | |||||
| N | HR (95% CI)a | n | HR (95% CI)a | n | HR (95% CI)a | n | HR (95% CI)a | |
| CMD-free | 1140 | Reference | 192 | Reference | 302 | Reference | 96 | Reference |
| Single CMD | 300 | 1.18 (0.93–1.50) | 140 | 1.04 (0.68–1.58) | 94 | 1.27 (0.64–2.50) | 121 | 1.37 (0.86–2.19) |
| CMD multimorbidity | 67 | 2.11 (1.42–3.14) | 34 | 1.57 (0.80–3.08) | 52 | 2.71 (1.36–5.40) | 39 | 1.50 (0.75–2.99) |
- a Multi-adjusted for baseline age, sex, education, body mass index (BMI), physical activity, hypertension, alcohol consumption, and apolipoprotein E (APOE) ԑ4 carrier status.
4 DISCUSSION
In this population-based cohort study of Swedish older adults, we found that: (1) cognitive decline accelerates dose-dependently with the presence of an increasing number of CMDs; (2) CMD multimorbidity nearly doubles the risk of cognitive impairment and its progression to dementia; and (3) CMD multimorbidity anticipates both the onset of cognitive impairment and its progression to dementia by ≈2 years.
T2D, HD, and stroke are widely recognized risk factors for dementia,1 and a growing body of evidence suggests that these individual CMDs also have an adverse impact on the preclinical and prodromal phases of dementia. T2D is associated with cognitive decline that is up to 50% faster than what is expected in normal aging,2 and poor glycemic control (i.e., HbA1c ≥7.0% to 7.5%) has been linked to increased risk of cognitive impairment.3, 4 The chronic hyperglycemia that characterizes T2D has been linked to several hallmarks of brain aging, including oxidative stress, the accumulation of advanced glycation end-products, and brain micro- and macro-vascular pathology.38 Furthermore, disruptions in insulin signaling have been proposed to interfere with the degradation of amyloid beta (Aβ), a key pathological feature of Alzheimer's disease (AD).38 HD has also been associated with accelerated cognitive decline,5 and recent studies have reported that a history of atrial fibrillation, coronary heart disease, or congestive heart failure is related to a 77% increased risk of non-amnestic mild cognitive impairment (MCI).7 The damaging impact of HD on cognition may be driven by reductions in cerebral blood flow or disruption of the blood-brain barrier, which can lead to neurological insults such as brain infarcts and an increased load of white matter lesions.39 Finally, a large body of literature has described the increased risk of cognitive decline9 and cognitive impairment10 following stroke, with an estimated 20% to 80% (depending on geography, follow-up time, and diagnostic criteria) of stroke survivors developing cognitive impairment. Cerebrovascular lesions resulting from stroke may adversely impact cognition via changes in cerebral blood flow and oxygen supply, elevated inflammation, and altered cortical connectivity.40
T2D, HD, and stroke are driven by common risk factors and interact with one another in complex ways across the life course.11 Despite this, only a handful of studies have investigated the combined impact of multiple co-morbid CMDs on the preclinical and prodromal phases of dementia. A large cross-sectional study from the UK Biobank (n = 474,129) reported dose-dependent associations between CMDs (including hypertension, coronary artery disease, and T2D) and poor performance in cognitive reasoning tasks.16 In addition, previous studies from our research group have demonstrated accelerated cognitive decline among older adults with higher Framingham general cardiovascular risk scores (FGCRS) in both the SNAC-K study (n = 2189)17 and the Rush Memory and Aging Project (n = 1588).18
Our study takes the further step of examining the impact of cardiometabolic multimorbidity on the continuum of cognitive phenotypes leading up to dementia. We provide evidence that an increasing number of comorbid CMDs predisposes older adults to accelerated cognitive decline, the development of cognitive impairment, and the progression of cognitive impairment to overt dementia in a dose-response fashion.
Notably, participants with only one CMD did not have a significantly increased risk of CIND or its progression to dementia over the 12-year follow-up. This result suggests that it may be the accumulation of multiple comorbid CMDs, rather than CMDs per se, that drives—or at least accelerates—cognitive deterioration. This finding also highlights the importance of preventing the development of CMD multimorbidity among older individuals—particularly those who already have T2D, HD, or stroke alone—in order to maintain cognitive health.
Our results additionally suggest that the damaging effect of CMDs on cognitive impairment and dementia may be more pronounced earlier (<78 years) as opposed to later (≥78 years) in old age. This is consistent with previous investigations demonstrating a greater impact of cardiovascular risk profile on cognition in earlier old age.17, 19 However, our results might be due to a smaller sample size of older-old adults, thereby limiting the statistical power to detect an effect in this group. In addition, selective survival could have impacted the CMD-CIND/dementia association among older-old adults because very old people with could CMDs die earlier than those without CMDs.
Given the heterogeneity of CMDs as chronic diseases with a potentially decades-long time course, additional studies are warranted to better understand how CMDs and cognitive health interact throughout the life course. In addition, further investigations are needed to elucidate the possible genetic, metabolic, and environmental factors underlying the connection between CMDs and adverse cognitive outcomes, which will involve the integration of longitudinal measures of cognitive function with neuropathological, genetic, and biomarker data.
4.1 Strengths and limitations
Strengths of this study include the longitudinal design with a long follow-up (12 years), large sample size (n = 2577), high retention rate (79.7%), and integration of data from multiple sources. One limitation is a risk of selection bias due to non-response over follow-up, leading to a younger and relatively healthier sample. That said, our results were not much altered after excluding participants who dropped out of the study (Table S1). In addition, the longitudinal association between CMDs and cognitive decline could have been influenced by individuals who developed dementia during follow-up. However, the given associations were not much altered after excluding all individuals who were diagnosed with dementia at later assessments (Table S2). The CMD-cognitive decline association was also unchanged after we excluded participants who had no repeated measures of global cognition score beyond the baseline assessment (n = 348) (Table S2). It is also possible that death acted as a competing risk for the development of CIND or that the results were influenced by reverse causality related to prodromal dementia, but we accounted for these possibilities in sensitivity analyses (Tables S3 and S4). Another limitation is the small sample size for the subgroup of participants with two or more CMDs, as well as participants in the older-old age group. Finally, we cannot rule out the influence of residual confounding due to unmeasured factors.
5 CONCLUSIONS
To the best of our knowledge, this is the first study to examine the combined impact of multiple CMDs on cognitive decline and the prodromal phase of dementia. Cardiometabolic multimorbidity accelerates cognitive decline and increases the risk of both cognitive impairment and its conversion to dementia. Our findings highlight comorbid T2D, HD, and stroke as ideal targets for preventive measures to slow cognitive decline and postpone the development of cognitive impairment and dementia.
ACKNOWLEDGMENTS
The authors would like to express their gratitude to the SNAC-K study participants and the staff involved in the SNAC-K data collection and management. The Swedish National Study on Aging and Care-Kungsholmen (http://www.snac.org) is financially supported by the Swedish Ministry of Health and Social Affairs, the participating County Councils and Municipalities, the Swedish Research Council (No. 2017–06088 and No. 2016-01705), and the Swedish Council for Health Working Life and Welfare (No. 2016-01705). W.X. received grants from the Swedish Research Council (No. 2017-00981 and No. 2021-01647), the Swedish Council for Health Working Life and Welfare (No. 2021-01826), and Lindhés Advokatbyrå AB (No. 2021-0134). A.D. additionally received support from Lindhés Advokatbyrå AB (No. 2021-0122). This study was accomplished within the context of the Swedish National Graduate School on Aging and Health (SWEAH).
CONFLICT OF INTEREST
The authors have no interests to disclose.
AUTHOR CONTRIBUTIONS
Abigail Dove, Anna Marseglia, and Weili Xu contributed to the conception and design of the study. Abigail Dove conducted the statistical analyses and drafted the manuscript. All authors interpreted the data, provided critical revisions, and approved the final version for publication. Abigail Dove is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
