Abstract
Background
Elevated high-sensitivity cardiac troponin (hs-cTn) levels are linked with cardiovascular disease and cognitive impairment, both of which are strong risk factors for late-life dementia (LLD). This study examined the association between hs-cTnI levels below the sex-specific 99th percentile for myocardial injury and the incidence of LLD in older women.
Methods
986 community-dwelling women aged ≥70 years without prior LLD and with hs-cTnI <15.6 ng/L (stratified into quartiles) were included from the Perth Longitudinal Study of Ageing Women. The primary outcome was incident LLD events, including LLD hospitalisation or death, over 14.5 years obtained from linked health records. Associations between hs-cTnI and LLD outcomes were explored using multivariable-adjusted Cox models, as part of restricted cubic splines.
Results
At baseline, participants’ mean (±SD) age was 75.2±2.7 years. Over 14.5 years of follow-up, LLD events (n=174, 17.7%), hospitalisations (n=155, 15.7%) and deaths (n=68, 6.9%) were recorded. Compared with those in the lowest quartile (Q1, median 3.1 ng/L), women in the highest quartile of hs-cTnI (Q4, median 7.3 ng/L) had a greater risk of developing LLD-related events (adjusted HR: 1.88, 95% CI: 1.22 to 2.91), hospitalisation (adjusted HR: 1.65, 95% CI: 1.04 to 2.64) and death (adjusted HR: 2.27, 95% CI: 1.13 to 4.59), after adjusting for established cardiovascular and dementia risk factors, including apolipoprotein E (APOE) genotype.
Conclusion
Among older women, hs-cTnI levels below the sex-specific 99th percentile for myocardial injury were associated with an increased risk of LLD events over 14.5 years. These findings suggest that hs-cTnI may identify older women at higher risk of LLD, capturing both cardiovascular and brain health vulnerability in older age.
Trial registration number
ACTRN12617000640303.
Keywords: Cardiovascular Diseases, Cohort Studies, Biomarkers
WHAT IS ALREADY KNOWN ON THIS TOPIC
Late-life dementia (LLD) may develop as a result of vascular and metabolic factors that accumulate over the lifespan. Cardiovascular disease, particularly coronary heart disease, has been associated with an increased risk of dementia. High-sensitivity cardiac troponin I (hs-cTnI), a biomarker of myocardial injury, has been linked to a greater risk of cardiovascular events, cerebrovascular disease and cognitive decline. Whether hs-cTnI levels below the sex-specific 99th percentile for acute myocardial injury are associated with LLD risk remains uncertain, particularly in older women.
WHAT THIS STUDY ADDS
In a cohort of 986 older women (≥70 years) with hs-cTnI levels below diagnostic thresholds, those in the highest quartile of hs-cTnI had a greater incidence of LLD events (hospitalisation or death) over 14.5 years. Notably, these findings remained significant after accounting for standard cardiovascular and dementia risk factors, including apolipoprotein E genotype.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
This study identifies an association between hs-cTnI elevations and LLD risk, suggesting that hs-cTnI might serve as a novel marker of combined cardiovascular and cognitive vulnerability in older women. Further investigation is warranted to determine how, or if, hs-cTnI testing could be integrated into dementia risk stratification and whether targeting the underlying drivers of hs-cTnI could modify LLD risk.
Introduction
Dementia affects over 57 million people globally, projected to reach 152 million by 2050.1 In 2019, dementia led to 1.62 million deaths worldwide, with women experiencing a greater burden.2 This underscores the urgent need for cost-effective and accessible biomarkers to facilitate early identification of individuals at high risk, thereby promoting timely interventions to reduce risk or prevent disease onset. While early-onset dementia (<65 years) is often driven by genetic factors,3 most cases develop after mid-life, and the more common late-life dementia (LLD) typically occurs after the age of 80 years.4 LLD appears primarily attributable to the cumulative effects of vascular, metabolic and lifestyle exposures acquired over many decades.5 In this context, identifying novel biomarkers for LLD is warranted because earlier detection might enable interventions to address modifiable risk factors, potentially lowering the risk or delaying the onset of the disease.
Links between cardiovascular disease (CVD) and dementia have previously been identified,6 with myocardial damage increasingly recognised as a contributor to cognitive dysfunction.7 High-sensitivity cardiac troponins (hs-cTn), particularly hs-cTnI and T, are widely used to rule out acute myocardial injury.8 Among these, hs-cTnI is commonly associated with myocardial infarction and coronary heart disease (CHD) in the general population.9 Although levels above the sex-specific 99th percentile (∼16 ng/L for women, depending on the assay) are indicative of acute myocardial damage,10 levels below this threshold have also been linked to increased cardiovascular risk.11 hs-cTn has been associated with higher risks of incident cerebrovascular events12 and cognitive decline,13 raising the possibility that higher hs-cTnI levels within a non-diagnostic range may be relevant for neurodegenerative diseases such as dementia. Importantly, CHD, a known risk factor for dementia,14 has been associated with elevated levels of hs-cTnI and hs-cTnT, though some studies suggest that hs-cTnI may have a closer relationship with coronary events.9
To date, only two studies have assessed the relationship between hs-cTn levels and incident dementia.15 16 However, both focused on younger populations (mean ages of 48 and 63 years), limiting their generalisability to LLD. Additionally, one study15 included participants with higher hs-cTnI levels than expected for a non-diagnostic range, while the other16 focused on hs-cTnT, rather than hs-cTnI, and did not adjust for apolipoprotein E (APOE) genotype, an important genetic risk factor for dementia.17 Thus, whether higher hs-cTnI levels below the sex-specific 99th percentile can reliably identify older women—specifically those aged ≥70 years—at heightened risk of LLD remains uncertain.
This study evaluates the association between hs-cTnI levels below the sex-specific 99th percentile and LLD in community-dwelling older women, controlling for established cardiovascular and dementia risk factors, including APOE genotype.
Methods
Study population
Data were from 1460 community-dwelling, ambulatory older women aged ≥70 years who were enrolled in the Calcium Intake Fracture Outcome Study (CAIFOS) in 1998.18 CAIFOS was a 5-year randomised clinical trial (RCT) specifically designed to assess the efficacy of oral supplementation of calcium carbonate (600 mg two times per day) for the prevention of osteoporotic fractures. After completion, participants were invited to a 10-year observational follow-up study. The RCT and observational follow-up are collectively referred to as the Perth Longitudinal Study of Ageing Women (PLSAW).
At baseline, participants had comparable medication use patterns to the general population of similar age, although they were more likely to have a relatively higher socioeconomic status.18 Of the initial sample of 1460 women, 1235 had plasma hs-cTnI measured. We excluded 56 participants who had hs-cTnI levels above the sex-specific 99th percentile (ie, >15.6 ng/L for females),11 as such elevations typically suggest myocardial injury. As we were interested in assessing LLD risk, women (n=23) who experienced any dementia-related event before the age of 80 years were excluded. Additionally, 170 had missing covariate data and were excluded. This left a sample of 986 women for this analysis (online supplemental figure 1).
Baseline cardiovascular risk factor assessment
We assessed body mass index (BMI), smoking status, physical activity, medication use, diabetes status, atherosclerotic vascular disease history (ASVD), systolic blood pressure (SBP), renal function (estimated glomerular filtration rate (eGFR)), alcohol intake, the 10-year Framingham General Cardiovascular Risk Score (FRS) and APOE genotype. Further details are provided in the online supplemental text.
Biochemistry: hs-cTnI
Plasma hs-cTnI was determined using an Abbott ARCHITECT i2000SR STAT hs-cTnI assay on samples collected at baseline (1998), stored at −80°C. The assay had a 10% coefficient of variation at a mean control of 4.7 ng/L for females, with a detection limit of 1.9 ng/L. Data below this limit (n=5; range, 1.4–1.9 ng/L) were included in the results. According to the manufacturer, the sex-specific 99th percentile cut-off for females was 15.6 ng/L, as previously noted in the Study population section. For the primary analyses, we considered approximately equal quartiles of hs-cTnI to remove the influence of extreme values and the skewed nature of the hs-cTnI distribution.
Late-life dementia outcomes
Incident dementia was tracked over 14.5 years (SD=3.3 years) using the Western Australian Data Linkage System via the Department of Health Western Australia and retrieved from the Western Australia Hospital Morbidity Data Collection. Dementia cases were classified using International Classification of Diseases (ICD)-9-CM and ICD-10-AM codes, as detailed in the online supplemental text. The primary outcome was LLD events (hospitalisation and/or death after the age of 80 years), with separate analyses for LLD hospitalisations and deaths, as used previously.19
Statistical analysis
Kaplan-Meier survival analysis was used to investigate the association between quartiles of hs-cTnI and LLD events, LLD hospitalisations and LLD deaths over 14.5 years. To account for potential non-linearity, restricted cubic splines with four knots within Cox proportional hazards models were used to examine the relationship between hs-cTnI and LLD outcomes using the ‘rms’ R package. HR estimates were referenced to the median hs-cTnI concentration of women in the lowest quartile (Q1) and were plotted against the respective outcomes with 95% confidence bands. Wald tests were used to obtain p-values.
We considered two models, an unadjusted and a multivariable model adjusted for FRS for CVD, treatment code (randomised to either calcium supplements or placebo), statin, prevalent ASVD, low-dose aspirin use, alcohol intake and APOE genotype. The proportional-hazards assumption was formally tested using scaled Schoenfeld residuals. No significant violations were observed (LLD events: global test p=0.12, hs-cTnI p=0.30; LLD hospitalisations: global p=0.26, hs-cTnI p=0.24; LLD death: global p=0.11, hs-cTnI p=0.81) and residuals plotted against time showed no systematic trend, supporting the assumption of a constant HR for hs-cTnI over time. Further details on sensitivity analyses, competing risks models and additional covariate adjustments are provided in the online supplemental text.
Statistical analyses were performed using IBM SPSS V.28.0 (IBM, Armonk, New York, USA), Stata V.18.5 (StataCorp, Texas, USA) and R software (V.4.0.4, R Foundation for Statistical Computing, Vienna, Austria). A p-value <0.05 was considered statistically significant.
Results
General characteristics of the study population
Table 1 presents the baseline participant characteristics. Women in the highest quartile (Q4) of hs-cTnI were older and had a higher BMI, mean SBP and higher FRS%. A higher prevalence of diabetes and ASVD, as well as greater use of antihypertensive and low-dose aspirin therapies, was also observed for individuals in the highest hs-cTnI quartile.
Table 1. Baseline characteristics of the participants across the approximate quartiles of hs-cTnI (n=986).
| All participants | Quartiles of hs-cTnI | ||||
|---|---|---|---|---|---|
| Quartile 1 (<3.6 ng/L) |
Quartile 2 (3.6–4.5 ng/L) |
Quartile 3 (4.6–5.8 ng/L) |
Quartile 4 (≥5.9 ng/L) |
||
| Number | 986 | 219 | 246 | 268 | 253 |
| Age, years | 75.2±2.7 | 74.5±2.4 | 74.8±2.6 | 75.5±2.8 | 75.8±2.7 |
| Calcium treatment, yes (%) | 495 (50.2) | 117 (53.4) | 127 (51.6) | 140 (52.2) | 111 (43.9) |
| BMI, kg/m2 | 27.1±4.6 | 26.0±4.1 | 26.7±4.8 | 27.1±4.3 | 28.4±4.7 |
| Smoked ever, yes (%) | 355 (36.1) | 78 (35.8) | 91 (37.0) | 90 (33.6) | 96 (38.1) |
| Diabetes, yes (%) | 56 (5.7) | 6 (2.7) | 12 (4.9) | 10 (3.7) | 28 (11.1) |
| Prevalent ASVD, yes (%) | 113 (11.5) | 15 (6.8) | 22 (8.9) | 28 (10.4) | 48 (19.0) |
| Statins, yes (%) | 185 (18.8) | 42 (19.2) | 39 (15.9) | 55 (20.5) | 49 (19.4) |
| Antihypertensives, yes (%) | 414 (42.0) | 62 (28.3) | 88 (35.8) | 119 (44.4) | 145 (57.3) |
| Mean SBP | 137.3±18.1 | 133.0±14.7 | 135.4±17.1 | 139.8±18.6 | 140.4±20.2 |
| Low dose aspirin, yes (%) | 195 (19.8) | 28 (12.8) | 43 (17.5) | 51 (19.0) | 73 (28.9) |
| Physical activity, kcal | 112.0 (37.4–203.7) | 122.6 (48.7–214.6) | 113.9 (41.6–187.1) | 96.9 (0–195.3) | 115.2 (34.4–211.6) |
| eGFR, mL/min/1.73m2 | 66.0±13.3 | 70.0±12.4 | 68.2±12.4 | 64.1±13.2 | 62.3±13.7 |
| Estimated CVD risk (Framingham), % | 19.3 (14.2–26.8) | 15.6 (12.7–21.3) | 17.3 (13.0–25.1) | 20.5 (15.2–27.5) | 23.5 (16.5–33.2) |
| Alcohol | |||||
| None (%) | 192 (19.5) | 35 (16.0) | 37 (15.0) | 64 (23.9) | 56 (22.1) |
| <10 standard drinks p/w (%) | 634 (64.3) | 140 (63.9) | 165 (67.1) | 164 (61.2) | 165 (65.2) |
| ≥10 standard drinks p/w (%) | 160 (16.2) | 44 (20.1) | 44 (17.9) | 40 (14.9) | 32 (12.6) |
| APOE genotypes | |||||
| APOE ℇ2/3, yes (%) | 156 (15.8) | 40 (18.3) | 37 (15.0) | 40 (14.9) | 39 (15.4) |
| APOE ℇ2/4, yes (%) | 23 (2.3) | 7 (3.2) | 5 (2.0) | 6 (2.2) | 5 (2.0) |
| APOE ℇ3/3, yes (%) | 611 (62.0) | 139 (63.5) | 155 (63.0) | 164 (61.2) | 153 (60.5) |
| APOE ℇ3/4, yes (%) | 181 (18.4) | 30 (13.7) | 44 (17.9) | 56 (20.9) | 51 (20.2) |
| APOE ℇ4/4, yes (%) | 15 (1.5) | 3 (1.4) | 5 (2.0) | 2 (0.7) | 5 (2.0) |
Data are expressed as mean±SD, median (IQR) or n (%).
Bolded values indicate statistical significance (p<0.05).
APOE, apolipoprotein E; ASVD, atherosclerotic vascular disease; BMI, body mass index; CVD, cardiovascular disease; eGFR, estimated glomerular filtration rate; hs-cTnI, high-sensitivity cardiac troponin I; p/w, per week; SBP, systolic blood pressure.
hs-cTnI and incident LLD dementia events
Over a 14.5-year period, the total follow-up was 12 198 person-years for LLD events and 12 562 person-years for deaths, with a mean follow-up duration of 12.4 years (SD±3.3) and 12.7 years (SD±3.1), respectively. During this time, 17.7% of older women (174 out of 986) experienced a LLD event, while 15.7% (155 out of 986) had a LLD-related hospitalisation, and 6.9% (68 out of 986) recorded a LLD death. Kaplan-Meier survival curves for the relationship between quartiles of hs-cTnI and LLD events, hospitalisations and deaths are displayed in figure 1. When the survival curves were re-estimated with age as the time scale, little divergence was seen before ∼80 years, after which the curves separated progressively in the same rank order observed in the main analysis (online supplemental figure 2).
Figure 1. Kaplan-Meier survival curves for 14.5 years of late-life dementia (LLD)-related (A) events, (B) hospitalisations and (C) deaths across quartiles of hs-cTnI. hs-cTnI, high-sensitivity cardiac troponin I.
The multivariable-adjusted HRs from Cox proportional hazards models with restricted cubic spline curves are shown in figure 2. In the multivariable-adjusted analysis, women in the highest quartile of hs-cTnI had a greater risk for a LLD event (HR: 1.88, 95% CI: 1.22 to 2.91), hospitalisation (HR: 1.65, 95% CI: 1.04 to 2.64) and death (HR: 2.27, 95% CI: 1.13 to 4.59) (table 2). No significant relationships were observed when comparing quartiles 2 and 3 to quartile 1 for any of the LLD outcomes.
Figure 2. Multivariable-adjusted HRs from Cox proportional hazards model with restrictive cubic spline curves demonstrating relationships between levels of hs-cTnI and 14.5 years of LLD-related (A) events, (B) hospitalisations and (C) deaths, comparing the specific level of hs-cTnI (horizontal axis) to the median value for women in the lowest quartile (3.1). The solid lines represent the HRs, while the grey shadings represent the 95% CI. Along the bottom of the graphs, the rug plots depict each observation. hs-cTnI, high-sensitivity cardiac troponin I; LLD, late-life dementia.
Table 2. Multivariable-adjusted HRs for the association between hs-cTnI and late-life dementia events (n=986).
| Outcome | Quartile 1 (<3.6 ng/L) |
Quartile 2 (3.6–4.5 ng/L) |
Quartile 3 (4.6–5.8 ng/L) |
Quartile 4 (≥5.9 ng/L) |
|---|---|---|---|---|
| Any late-life dementia event | ||||
| Events, n (%) | 31 (14.2) | 34 (13.8) | 50 (18.7) | 59 (23.3) |
| Unadjusted | Reference | 1.07 (0.79 to 1.46) | 1.46 (1.05 to 2.02) | 2.26 (1.50 to 3.41) |
| Multivariable-adjusted | Reference | 1.01 (0.74 to 1.37) | 1.28 (0.92 to 1.81) | 1.88 (1.22 to 2.91) |
| Any late-life dementia hospitalisation | ||||
| Events, n (%) | 27 (12.3) | 33 (13.4) | 42 (15.7) | 53 (21.0) |
| Unadjusted | Reference | 1.11 (0.80 to 1.54) | 1.42 (1.00 to 2.02) | 2.08 (1.34 to 3.22) |
| Multivariable-adjusted | Reference | 1.03 (0.74 to 1.43) | 1.21 (0.84 to 1.75) | 1.65 (1.04 to 2.64) |
| Any late-life dementia death | ||||
| Events, n (%) | 11 (5.0) | 14 (5.7) | 22 (8.2) | 21 (8.3) |
| Unadjusted | Reference | 1.21 (0.72 to 2.03) | 1.78 (1.03 to 3.08) | 2.70 (1.39 to 5.25) |
| Multivariable-adjusted | Reference | 1.15 (0.68 to 1.95) | 1.59 (0.90 to 2.82) | 2.27 (1.13 to 4.59) |
Multivariable-adjusted model included the general Framingham risk score, plus treatment code (calcium or placebo), alcohol intake, prevalent atherosclerotic vascular disease, statin use, low dose of aspirin use and apolipoprotein E genotype. The HRs and 95% CI were calculated using the median values for each quartile: 3.1, 4.1, 5.1 and 7.3 for Q1, Q2, Q3 and Q4, respectively. Bolded values indicate statistical significance (p<0.05).
hs-cTnI, high-sensitivity cardiac troponin I.
Additional analysis
In the sensitivity analysis considering LLD events, when women with prevalent ASVD were excluded, the association persisted for the highest quartile (HR: 1.65, 95% CI: 1.05 to 2.61) (online supplemental table 1), compared with the lowest quartile (Q1). Similarly, after adjusting for physical activity in the multivariable models, the associations between the highest quartile of hs-cTnI and LLD events, hospitalisations and deaths remained unchanged (online supplemental table 2). Additionally, when testing the alternative hs-cTnI cut-off strategies (<5 vs ≥5 ng/L; and <4 ng/L, 4–<8 ng/L, ≥8 ng/L), the results remained consistent in both direction and significance for the highest hs-cTnI categories in relation to any LLD events (online supplemental table 3). Similarly, when the model for LLD events was refitted with the FRS disaggregated into its individual components, the association persisted for the highest quartile: compared with Q1, the adjusted HRs were 0.99 (95 % CI 0.73 to 1.34) for Q2, 1.22 (0.87 to 1.72) for Q3 and 1.73 (1.11 to 2.70) for Q4.
In the subset of participants with available eGFR data (n=905), the addition of eGFR to the multivariable model also did not alter the association between hs-cTnI and LLD events. Only women in the highest quartile (Q4) retained a higher risk of LLD events (HR: 1.70, 95% CI: 1.06 to 2.71) compared with those in Q1. When competing risks from non-dementia-related deaths were considered in the subdistribution hazard models (online supplemental table 4), the association with LLD events was attenuated. Similar attenuation was observed when we applied the same competing-risks approach using CHD and heart failure mortality (online supplemental tables 5 and 6). Non-parametric cumulative-incidence curves are shown in the online supplemental figure 3. Additionally, we examined potential effect modification by APOE ε4 carrier status by adding interaction terms to the models; no significant interactions were detected (all p>0.05). Finally, in the multiple-imputation sensitivity analysis of the primary spline Cox model for LLD events, the pooled HRs for quartiles 2–4, compared with quartile 1, were 0.94 (95% CI: 0.88 to 1.00), 1.16 (1.03 to 1.31) and 1.64 (1.14 to 2.36), respectively, closely reflecting the complete-case estimates.
Discussion
This is the first study to consider hs-cTnI levels below the diagnostic threshold for myocardial injury in relation to incident LLD events, including hospitalisation and death, in community-dwelling older women. We showed that women in the highest quartile (Q4, median 7.3 ng/L) had higher risk (between 1.6 and 2.2 times) for LLD events, hospitalisations and dementia-related mortality. While this supports a potential link between cardiovascular health and LLD, the attenuation of this association in subdistribution-based competing risk models suggests that elevated hs-cTnI may also reflect an increased burden of cardiovascular mortality in this population. Nonetheless, these findings reinforce the importance of cardiovascular integrity for brain health, even in later life, likely reflecting shared risk factors between LLD and CVD.
To date, few studies have explored the longitudinal association between hs-cTn, cognitive decline and dementia. Schneider et al16 reported that higher hs-cTnT levels were associated with increased dementia-related hospitalisations risk among 9472 US-based participants (mean age 63 years, 59% female). Compared with hs-cTnT <3 ng/L (reference), adjusted HRs were 1.43 (95% CI: 1.07 to 1.91) for 6–8 ng/L and 1.82 (95% CI: 1.33 to 2.48) for 9–13 ng/L. This association remained significant after excluding participants with prior stroke and adjusting for major cardiovascular risk factors, such as left ventricular hypertrophy and carotid intima-media thickness.
We observed similar findings in our study focusing on LLD, but key differences must be noted. First, Schneider et al focused on hs-cTnT, whereas our study investigated hs-cTnI, which has distinct biological and clinical implications. hs-cTnI is less likely to be elevated in non-cardiac conditions20 and has been suggested to have a closer association with myocardial infarction and CHD in some studies.9 Second, their cohort was younger and included both sexes, whereas we focused exclusively on older women, a population at higher risk of LLD. Third, their primary outcome was dementia hospitalisation, while we included both dementia-related hospitalisations and deaths. Fourth, APOE ε4 genotype, a major genetic risk factor for dementia,15 17 was not accounted for in their analysis.
Tynkkynen et al15 reported an association between hs-cTnI and incident dementia in a Finnish cohort aged 25–74 years (n=7114, 48.9% female) over an 18-year follow-up. However, given their cohort’s younger mean age (47.9±13.2 years) and lack of focus on dementia after 80 years, the generalisability of these findings to LLD remains uncertain. This is particularly relevant as LLD is thought to have a distinct risk factor profile compared not only to early-onset dementia (<65 years) but also to dementia developing between 65 and 80 years.21 Furthermore, although hs-cTnI was initially associated with incident dementia, this was no longer significant after adjusting for dementia risk factors and NT-proBNP, suggesting that the observed relationship may have been influenced by underlying cardiovascular dysfunction and unmeasured shared risk factors. Importantly, Tynkkynen et al included subjects with hs-cTnI levels well into the elevated range, limiting generalisation to populations without elevated troponin, as in the present study.
Our study builds on these findings by focusing on an older female population, demonstrating that higher hs-cTnI levels, even when not considered clinically elevated, remained significantly associated with LLD events, even after excluding women with prevalent ASVD and adjusting for APOE ε4 genotype. This underscores the potential utility of hs-cTnI as a biomarker for LLD, particularly in older women, the fastest-growing demographic globally22 with the highest dementia risk.1 2 Moreover, the association remained significant after adjusting for eGFR, indicating that higher hs-cTnI levels may represent a relevant risk factor for LLD beyond kidney function. However, as only 32% of women in our cohort met the criteria for chronic kidney disease (CKD) (eGFR <60 mL/min), further research is needed to determine whether different CKD stages or longitudinal changes in kidney function influence this relationship, particularly given that CKD can elevate troponin levels due to reduced clearance and increased myocardial stress.23
Interestingly, the association with LLD events was attenuated in the subdistribution model, suggesting that elevated hs-cTnI levels may primarily reflect a broader cardiovascular risk factor burden rather than a direct pathway to dementia. Nonetheless, the cause-specific models used in our primary analysis—which treat non-dementia deaths as censoring events—remain suitable for investigating associations between hs-cTnI levels and later LLD risk, as they are unaffected by survival time. This distinction is important in the context of aetiological inference, where interest lies in understanding potential disease mechanisms rather than predicting absolute risk in the presence of competing outcomes. Supporting this, previous research has shown that higher troponin levels are strongly associated with ASVD events and all-cause mortality, even in otherwise healthy individuals.24 Given the reported overlap between cardiovascular dysfunction and neurodegeneration,25 our findings suggest that hs-cTnI may serve as an indicator of systemic risk, capturing both cardiovascular and brain health vulnerability. Moreover, similar to our findings for LLD, an earlier study in this same cohort found that hs-cTnI levels above approximately 6 ng/L were associated with a higher risk of future cardiac events.11 The consistency in this threshold across studies reinforces the role of hs-cTnI as a marker of overall cardiovascular health, linking it to both cardiac and cognitive outcomes.
LLD has a prominent vascular component,21 and thus the association between hs-cTnI and LLD observed in our study likely reflects this underlying vascular contribution. The well-established link between hs-cTnI, myocardial infarction and CHD8 may further explain its connection with LLD. Although speculative, several mechanisms may underlie this association. Myocardial injury, as assessed by hs-cTnI, can lead to inadequate left ventricular function (online supplemental file S1), which in turn may cause chronic cerebral hypoperfusion (online supplemental file S2). This condition has been associated with structural and functional alterations in the hippocampus and cerebral cortex (online supplemental file S3), regions critically involved in cognitive functions. Additionally, CHD involves significant platelet activation, contributing to cerebral inflammation and vasoconstriction (online supplemental file S4), and has been implicated in the elevation of the levels of β-amyloid peptide (online supplemental file S5), a hallmark of Alzheimer’s disease. Furthermore, current evidence suggests that elevated troponin levels may be linked with subclinical cerebrovascular damage and future risk of incident stroke, potentially contributing to cognitive vulnerability (online supplemental file S6).
In clinical practice, while hs-cTnI measurement is widely used for patients presenting with symptoms suggestive of acute coronary syndrome and other causes of myocardial injury,9 our results suggest potential utility for broader risk assessment in older individuals without overt cardiac symptoms. In fact, emerging evidence indicates that troponin levels within the normal range can refine cardiovascular risk stratification and guide more intensive preventive therapies in selected individuals.26 27 Although further evidence is needed, measuring hs-cTnI at levels below the conventional diagnostic threshold could help identify those at heightened risk of cardiovascular-related conditions and possibly dementia. Notably, even modest hs-cTnI elevations below diagnostic thresholds have been associated with an increased risk of injurious falls,28 cardiac events11 and CVD mortality in older adults. This raises the promising possibility that hs-cTnI could serve as a multi-system risk indicator in older populations. However, rigorous evidence to support community-based or routine clinical screening for hs-cTnI is lacking, and comprehensive cost-effectiveness and risk–benefit analyses are required before such strategies can be recommended.
Strengths of the study include its longitudinal design, which allowed for the observation of long-term outcomes over a 14.5-year follow-up period. Using a well-defined cohort of community-dwelling older women aged ≥70 years ensures relevance to a high-risk population for LLD. The comprehensive approach of the study, which accounted for various cardiovascular and dementia risk factors (such as APOE genotype), strengthens these findings. Importantly, the prevalence of the APOE ɛ4 allele in our study is similar to its prevalence in other populations (∼25%).17 Finally, dementia hospitalisations and deaths were captured from two separate data sources (hospitalisations and deaths), demonstrating the reproducibility of the observed associations between hs-cTnI and LLD across these different data sources in this population.
Despite these strengths, limitations exist. First, the observational design precludes establishing causality between elevated hs-cTnI and LLD outcomes. Second, although we have adjusted for several CVD and dementia risk factors, unmeasured confounding cannot be ruled out. Third, the study population consisted predominately of community-dwelling Caucasian older women, which may limit the generalisability of our findings to other populations, including men, younger individuals and dementia events before the age of 80. However, it is important to note that older women represent a high dementia-risk population.1 2 Finally, although linked hospital-discharge data are a well-established resource for clinical and epidemiological research, they may not capture every dementia diagnosis. A systematic review of 27 validation studies reported that the sensitivity of hospital or death-registry data for identifying dementia ranged from 21% to 86%, indicating that some cases in our cohort may have been missed.29 Nevertheless, an Australian audit comparing ICD-10 hospital codes with full medical-record reviews found 85–87% concordance (κ=0.71) for principal dementia diagnoses, demonstrating substantial agreement and supporting the reliability of our outcome classification.30 Moreover, the incidence of LLD events in our cohort (∼18%) is in line with rates reported for similar populations of older Australian women (∼20%) (online supplemental file S7), which further validates our findings.
In conclusion, our findings demonstrate that hs-cTnI levels, even below the threshold used to identify myocardial injury, could serve as an important prognostic indicator of LLD risk in community-dwelling older women. Although these observations highlight a potential role for hs-cTnI beyond its current application in acute settings, whether preventative strategies incorporating hs-cTnI in risk assessment can reduce the incidence of LLD remains uncertain and warrants further investigation. Determining the cost-effectiveness and risk–benefit balance of such approaches will be crucial to informing targeted management strategies that address both cardiac and brain health.
Supplementary material
Acknowledgements
The authors thank the Western Australia Data Linkage Branch, Hospital Morbidity Data Collection, the Australian Coordinating Registry, the State Registries of Births, Deaths and Marriages, the Coroners, the National Coronial Information System and the Victorian Department of Justice and Community Safety for providing the cause of death unit record file data.
Footnotes
Funding: The study was supported by project grants 254627, 303169 and 572604 from the National Health and Medical Research Council of Australia. CT-H is supported by the Forrest Research Foundation Scholarship and the ECU Higher Degree by Research Scholarship. MSim received support from the Royal Perth Hospital Research Foundation Career Advancement Fellowship (CAF 130/2020) and an Emerging Leader Fellowship from the Western Australian Future Health Research and Innovation Fund. JRL is supported by a National Heart Foundation of Australia Future Leader Fellowship (ID: 102817 and 107323). This work was made possible through Dementia Australia's support of their Chair of Dementia (author BCMS). None of the funding agencies had any role in the conduct of the study; collection, management, analysis or interpretation of the data; or preparation, review or approval of the manuscript.
Provenance and peer review: Not commissioned; externally peer reviewed.
Patient consent for publication: Not applicable.
Ethics approval: The CAIFOS and PLSAW studies were retrospectively registered with the Australian New Zealand Clinical Trials Registry, with registration numbers #ACTRN12615000750583 and #ACTRN12617000640303, and were approved by the Human Ethics Committee of the University of Western Australia. The study adhered to the principles outlined in the Declaration of Helsinki. Linked health data was authorised by the Human Research Ethics Committee of the Western Australian Department of Health (#2009/24). Participants gave informed consent to participate in the study before taking part.
Data availability free text: Data used for this paper cannot be shared unless approved by the Governing Ethics Bodies.
Patient and public involvement: Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Data availability statement
Data may be obtained from a third party and are not publicly available.
References
- 1.Nichols E, Steinmetz JD, Vollset SE, et al. Estimation of the global prevalence of dementia in 2019 and forecasted prevalence in 2050: an analysis for the Global Burden of Disease Study 2019. Lancet Public Health. 2022;7:e105–25. doi: 10.1016/S2468-2667(21)00249-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.GBD 2019 Collaborators Global mortality from dementia: Application of a new method and results from the Global Burden of Disease Study 2019. A&D Transl Res & Clin Interv. 2021;7:e12200. doi: 10.1002/trc2.12200. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Masellis M, Sherborn K, Neto P, et al. Early-onset dementias: diagnostic and etiological considerations. Alzheimers Res Ther. 2013;5:S7. doi: 10.1186/alzrt197. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Fotuhi M, Hachinski V, Whitehouse PJ. Changing perspectives regarding late-life dementia. Nat Rev Neurol. 2009;5:649–58. doi: 10.1038/nrneurol.2009.175. [DOI] [PubMed] [Google Scholar]
- 5.Whalley LJ, Dick FD, McNeill G. A life-course approach to the aetiology of late-onset dementias. Lancet Neurol. 2006;5:87–96. doi: 10.1016/S1474-4422(05)70286-6. [DOI] [PubMed] [Google Scholar]
- 6.Song R, Pan K-Y, Xu H, et al. Association of cardiovascular risk burden with risk of dementia and brain pathologies: A population-based cohort study. Alzheimers Dement. 2021;17:1914–22. doi: 10.1002/alz.12343. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Jensen M, Zeller T, Twerenbold R, et al. Circulating cardiac biomarkers, structural brain changes, and dementia: Emerging insights and perspectives. Alzheimers Dement. 2023;19:1529–48. doi: 10.1002/alz.12926. [DOI] [PubMed] [Google Scholar]
- 8.Reichlin T, Hochholzer W, Bassetti S, et al. Early Diagnosis of Myocardial Infarction with Sensitive Cardiac Troponin Assays. N Engl J Med. 2009;361:858–67. doi: 10.1056/NEJMoa0900428. [DOI] [PubMed] [Google Scholar]
- 9.Welsh P, Preiss D, Hayward C, et al. Cardiac Troponin T and Troponin I in the General Population. Circulation. 2019;139:2754–64. doi: 10.1161/CIRCULATIONAHA.118.038529. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Byrne RA, Rossello X, Coughlan JJ, et al. 2023 ESC Guidelines for the management of acute coronary syndromes: Developed by the task force on the management of acute coronary syndromes of the European Society of Cardiology (ESC) Eur Heart J. 2023;44:3720–826. doi: 10.1093/eurheartj/ehad191. [DOI] [PubMed] [Google Scholar]
- 11.Lewis JR, Lim WH, Wong G, et al. Association Between High‐Sensitivity Cardiac Troponin I and Cardiac Events in Elderly Women. JAHA. 2017;6:e004174. doi: 10.1161/JAHA.116.004174. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Camen S, Palosaari T, Reinikainen J, et al. Cardiac Troponin I and Incident Stroke in European Cohorts: Insights From the BiomarCaRE Project. Stroke. 2020;51:2770–7. doi: 10.1161/STROKEAHA.120.029452. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Wijsman LW, de Craen AJ, Trompet S, et al. High-sensitivity cardiac troponin T is associated with cognitive decline in older adults at high cardiovascular risk. Eur J Prev Cardiol. 2016;23:1383–92. doi: 10.1177/2047487316632364. [DOI] [PubMed] [Google Scholar]
- 14.Wolters FJ, Segufa RA, Darweesh SKL, et al. Coronary heart disease, heart failure, and the risk of dementia: A systematic review and meta-analysis. Alzheimers Dement. 2018;14:1493–504. doi: 10.1016/j.jalz.2018.01.007. [DOI] [PubMed] [Google Scholar]
- 15.Tynkkynen J, Hernesniemi JA, Laatikainen T, et al. High-sensitivity cardiac troponin I and NT-proBNP as predictors of incident dementia and Alzheimer’s disease: the FINRISK Study. J Neurol. 2017;264:503–11. doi: 10.1007/s00415-016-8378-7. [DOI] [PubMed] [Google Scholar]
- 16.Schneider ALC, Rawlings AM, Sharrett AR, et al. High-sensitivity cardiac troponin T and cognitive function and dementia risk: the atherosclerosis risk in communities study. Eur Heart J. 2014;35:1817–24. doi: 10.1093/eurheartj/ehu124. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Corbo RM, Scacchi R. Apolipoprotein E (APOE) allele distribution in the world. Is APOE*4 a “thrifty” allele? Ann Hum Genet. 1999;63:301–10. doi: 10.1046/j.1469-1809.1999.6340301.x. [DOI] [PubMed] [Google Scholar]
- 18.Prince RL, Devine A, Dhaliwal SS, et al. Effects of calcium supplementation on clinical fracture and bone structure: results of a 5-year, double-blind, placebo-controlled trial in elderly women. Arch Intern Med. 2006;166:869–75. doi: 10.1001/archinte.166.8.869. [DOI] [PubMed] [Google Scholar]
- 19.Porter T, Sim M, Prince RL, et al. Abdominal aortic calcification on lateral spine images captured during bone density testing and late-life dementia risk in older women: A prospective cohort study. Lancet Reg Health West Pac . 2022;26:100502. doi: 10.1016/j.lanwpc.2022.100502. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Vestergaard KR, Jespersen CB, Arnadottir A, et al. Prevalence and significance of troponin elevations in patients without acute coronary disease. Int J Cardiol. 2016;222:819–25. doi: 10.1016/j.ijcard.2016.07.166. [DOI] [PubMed] [Google Scholar]
- 21.McGrath ER, Beiser AS, O’Donnell A, et al. Determining Vascular Risk Factors for Dementia and Dementia Risk Prediction Across Mid- to Later Life: The Framingham Heart Study. Neurology (ECronicon) 2022;99:e142–53. doi: 10.1212/WNL.0000000000200521. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.United Nations . World social report 2023: leaving no one behind in an ageing world. New York: UN, Department of Economic and Social Affairs; 2023. [Google Scholar]
- 23.Parikh RH, Seliger SL, deFilippi CR. Use and interpretation of high sensitivity cardiac troponins in patients with chronic kidney disease with and without acute myocardial infarction. Clin Biochem. 2015;48:247–53. doi: 10.1016/j.clinbiochem.2015.01.004. [DOI] [PubMed] [Google Scholar]
- 24.Neumann JT, Twerenbold R, Weimann J, et al. Prognostic Value of Cardiovascular Biomarkers in the Population. JAMA. 2024;331:1898–909. doi: 10.1001/jama.2024.5596. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Saeed A, Lopez O, Cohen A, et al. Cardiovascular Disease and Alzheimer’s Disease: The Heart-Brain Axis. J Am Heart Assoc. 2023;12:e030780. doi: 10.1161/JAHA.123.030780. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Marston NA, Bonaca MP, Jarolim P, et al. Clinical Application of High-Sensitivity Troponin Testing in the Atherosclerotic Cardiovascular Disease Framework of the Current Cholesterol Guidelines. JAMA Cardiol. 2020;5:1255–62. doi: 10.1001/jamacardio.2020.2981. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Pandey A, Patel KV, Vongpatanasin W, et al. Incorporation of Biomarkers Into Risk Assessment for Allocation of Antihypertensive Medication According to the 2017 ACC/AHA High Blood Pressure Guideline: A Pooled Cohort Analysis. Circulation. 2019;140:2076–88. doi: 10.1161/CIRCULATIONAHA.119.043337. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Gebre AK, Sim M, Dalla Via J, et al. Association between high-sensitivity cardiac troponin I and fall-related hospitalisation in women aged over 70 years. Heart. 2023;110:108–14. doi: 10.1136/heartjnl-2023-322861. [DOI] [PubMed] [Google Scholar]
- 29.Wilkinson T, Ly A, Schnier C, et al. Identifying dementia cases with routinely collected health data: A systematic review. Alzheimers Dement. 2018;14:1038–51. doi: 10.1016/j.jalz.2018.02.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Henderson T, Shepheard J, Sundararajan V. Quality of diagnosis and procedure coding in ICD-10 administrative data. Med Care. 2006;44:1011–9. doi: 10.1097/01.mlr.0000228018.48783.34. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data may be obtained from a third party and are not publicly available.