Research report: Admission high-sensitivity cardiac troponin T and clinical outcomes after acute ischemic stroke: a retrospective cohort study with exploratory nonlinear analysis

In a retrospective study of 564 adults hospitalized with acute ischemic stroke, cardiac troponin T measured within 24 hours of admission was associated with both poorer functional recovery at 90 days and higher all-cause mortality during follow-up. The association remained after adjustment for major clinical factors, but its pattern was nonlinear: risk rose most sharply at lower troponin concentrations, while the estimated increase appeared to flatten above approximately 28–32 ng/L. The investigators described those values as exploratory turning points rather than validated clinical cutoffs. The single-center cohort, conducted at Shanghang County Hospital in Fujian, China, included patients admitted from January 2022 through October 2024.

Eligible participants were at least 18 years old, had acute ischemic stroke confirmed by cranial computed tomography or magnetic resonance imaging, reached the hospital within 72 hours of symptom onset, had a high-sensitivity cardiac troponin T, or hs-cTnT, result obtained within 24 hours of admission, and had the clinical data required for the prespecified analyses. Of 1,447 people initially screened, 883 were excluded under the predefined eligibility criteria and 564 entered the final cohort. Patients with documented acute coronary syndrome, myocardial infarction, severe heart failure, advanced kidney or liver disease, recent surgery or trauma, and other major conditions were excluded.

Blood samples were analyzed with the Roche Elecsys Troponin T high-sensitivity assay on a Cobas e601 automated electrochemiluminescence analyzer. The assay had a reported measurement range of 3–10,000 ng/L, a lower detection limit of 3 ng/L, and a lowest concentration with a coefficient of variation below 10% of 13 ng/L. For the main categorical comparison, researchers classified concentrations above the assay-specific 99th-percentile upper reference limit of 14 ng/L as elevated. They also retained hs-cTnT as a continuous variable, assessed it by quartile and in 10 ng/L increments, and used restricted cubic splines to examine the shape of its association with outcomes rather than assuming a simple linear relationship.

Among the 564 participants, 212, or 37.6%, had hs-cTnT above 14 ng/L. This group was older on average than patients with concentrations at or below the reference limit: 78.28 versus 69.03 years. It also had more severe neurological deficits, a higher prevalence of hypertension, coronary heart disease and prior cerebral infarction, and lower mean estimated glomerular filtration rate, or eGFR, at 65.49 versus 79.62 mL/min/1.73 m². Median hs-cTnT was 20.41 ng/L in the elevated group and 6.53 ng/L in the lower group.

Infarcts involving multiple territories were also more frequent in the elevated group, at 19.81% versus 11.08%. These baseline differences were addressed statistically, but they show why an observational association cannot be interpreted as evidence that troponin elevation itself caused the outcomes. The primary functional endpoint was a poor 90-day outcome, defined as a modified Rankin Scale score from 3 to 6.

Multivariable logistic regression showed that patients with hs-cTnT above 14 ng/L had nearly twice the adjusted odds of a poor outcome compared with those at or below 14 ng/L: an odds ratio of 1.93, with a 95% confidence interval of 1.22–3.05 and p = 0.005. The estimate was 3.20 in the unadjusted model and 2.83 after adjustment for age, sex and smoking status. The fully adjusted model also included admission National Institutes of Health Stroke Scale score, hypertension, diabetes, coronary heart disease, previous cerebral infarction, renal impairment and eGFR.

The analysis showed an association; it did not establish that increasing troponin directly caused functional impairment or that troponin-guided management would improve outcomes. For mortality, the researchers used Cox proportional hazards models with admission as the starting point for follow-up. Patients with hs-cTnT above 14 ng/L had a fully adjusted hazard ratio of 4.12 for all-cause death, with a 95% confidence interval of 2.31–7.48 and p < 0.001.

The corresponding unadjusted hazard ratio was 6.24, and the result remained significant in a parsimonious sensitivity model that included hs-cTnT, age, sex, admission stroke severity and eGFR. Schoenfeld residual tests did not indicate a statistically significant violation of the proportional-hazards assumption. Changing the time origin from admission to stroke onset, with or without delayed entry at admission, produced closely similar fully adjusted estimates of 4.17 and 4.41, respectively.

During a median follow-up of 17 months, the cohort recorded 78 all-cause deaths and 71 recurrent strokes, contributing 893.8 person-years of observation. The mortality rate was 8.7 per 100 person-years, while the recurrent-stroke rate was 7.9 per 100 person-years. Kaplan–Meier curves showed markedly lower overall survival among patients with hs-cTnT above 14 ng/L, with a log-rank p value below 0.001.

Recurrent cerebrovascular events, however, did not differ significantly between the groups after death was treated as a competing event. The adjusted subdistribution hazard ratio was 1.05, with a 95% confidence interval of 0.56–1.96 and p = 0.877. Restricted cubic spline analyses indicated significant nonlinear relationships between hs-cTnT and both primary endpoints, with p < 0.001 for each.

Segmented regression estimated a turning point of 28.42 ng/L for 90-day functional outcome. Below that level, each 1 ng/L increase in hs-cTnT was associated with a 4.8% increase in the odds of a poor outcome, with an adjusted odds ratio of 1.048 per 1 ng/L and a 95% confidence interval of 1.005–1.093. Above the turning point, the model showed no statistically significant further increase, with p = 0.328.

The estimated slope therefore describes the pattern in this dataset rather than demonstrating a biological threshold. The investigators also noted that the study’s continuous and quartile analyses supported the broader proposition that information may extend beyond a simple normal-versus-elevated division. Mortality followed a similar pattern.

The exploratory mortality turning point was 31.68 ng/L. Below this concentration, the hazard of death increased by 9.6% for each 1 ng/L rise in hs-cTnT, corresponding to a hazard ratio of 1.096 and a 95% confidence interval of 1.045–1.149. No significant further increase was detected above the turning point, with p = 0.176. Only 67 patients, or 11.9%, had values above 28.42 ng/L, and 52 patients, or 9.2%, exceeded 31.68 ng/L. Because relatively few observations lay above these thresholds, the apparent flattening could reflect limited statistical power as much as a genuine change in risk.

The study therefore supports describing the concentration range of 28–32 ng/L as a model-derived signal, not as a validated cutoff for diagnosis or treatment. The biological processes behind the association were not directly measured. The investigators discussed neurogenic cardiac stress, in which brain ischemia and autonomic activation could produce catecholamine-driven myocardial injury, as one possible explanation.

Greater cardiovascular comorbidity, more severe stroke, systemic inflammation and reduced renal clearance were also considered. Yet catecholamine concentrations, serial echocardiography, inflammatory markers and repeated troponin measurements were unavailable, so the relative contributions of these pathways cannot be separated. Patients with elevated hs-cTnT had lower eGFR, but the mortality and functional associations persisted after adjustment for renal function.

The findings therefore show that admission hs-cTnT contained prognostic information in this cohort, but they do not establish the mechanism linking troponin elevation to death or disability. Several constraints limit how far the results can be generalized. The study included one county hospital in China, relied on retrospective records, and excluded 49 otherwise eligible patients because required data were missing.

It also included only patients who survived long enough for blood sampling, creating potential survivor bias during the earliest post-stroke period. Troponin was measured only once, causes of death were unavailable, residual confounding remained possible, and the median follow-up was relatively short at 17 months. The study did not test whether troponin-guided monitoring, cardiac imaging or specialist referral improves outcomes, and the marker’s incremental predictive value beyond existing prognostic information was not formally assessed.

The results are best viewed as evidence warranting prospective, multicenter validation rather than as grounds for a treatment rule.

Subject of Research: Association of admission high-sensitivity cardiac troponin T with 90-day function and all-cause mortality after acute ischemic stroke

Article Title: Admission high-sensitivity cardiac troponin T and clinical outcomes after acute ischemic stroke: a retrospective cohort study with exploratory nonlinear analysis

Article References: Chen, Y., Liu, S., Xue, H., Wang, R., Wen, R., Xiao, Z., Sheng, X., Chen, P., Song, N., & Li, Y. (2026). Admission high-sensitivity cardiac troponin T and clinical outcomes after acute ischemic stroke: a retrospective cohort study with exploratory nonlinear analysis. Frontiers in Neurology, 17, Article 1882492. https://doi.org/10.3389/fneur.2026.1882492

Image Credits: AI Generated

DOI: 10.3389/fneur.2026.1882492

Keywords: acute ischemic stroke, cardiac troponin T, stroke prognosis, 90-day functional outcome, all-cause mortality, nonlinear association, restricted cubic splines, retrospective cohort, cardiovascular risk, Admission, high-sensitivity, cardiac