OUP user menu

★ Editor's choice ★

High population prevalence of cardiac troponin I measured by a high-sensitivity assay and cardiovascular risk estimation: the MORGAM Biomarker Project Scottish Cohort

Tanja Zeller, Hugh Tunstall-Pedoe, Olli Saarela, Francisco Ojeda, Renate B. Schnabel, Tarja Tuovinen, Mark Woodward, Allan Struthers, Maria Hughes, Frank Kee, Veikko Salomaa, Kari Kuulasmaa, Stefan Blankenberg,
DOI: http://dx.doi.org/10.1093/eurheartj/eht406 271-281 First published online: 8 October 2013


Aims Our aim was to test the prediction and clinical applicability of high-sensitivity assayed troponin I for incident cardiovascular events in a general middle-aged European population.

Methods and results High-sensitivity assayed troponin I was measured in the Scottish Heart Health Extended Cohort (n = 15 340) with 2171 cardiovascular events (including acute coronary heart disease and probable ischaemic strokes), 714 coronary deaths (25% of all deaths), 1980 myocardial infarctions, and 797 strokes of all kinds during an average of 20 years follow-up. Detection rate above the limit of detection (LoD) was 74.8% in the overall population and 82.6% in men and 67.0% in women. Troponin I assayed by the high-sensitivity method was associated with future cardiovascular risk after full adjustment such as that individuals in the fourth category had 2.5 times the risk compared with those without detectable troponin I (P < 0.0001). These associations remained significant even for those individuals in whom levels of contemporary-sensitivity troponin I measures were not detectable. Addition of troponin I levels to clinical variables led to significant increases in risk prediction with significant improvement of the c-statistic (P < 0.0001) and net reclassification (P < 0.0001). A threshold of 4.7 pg/mL in women and 7.0 pg/mL in men is suggested to detect individuals at high risk for future cardiovascular events.

Conclusion Troponin I, measured with a high-sensitivity assay, is an independent predictor of cardiovascular events and might support selection of at risk individuals.

  • High-sensitivity assayed troponin I
  • Cardiovascular risk
  • Mortality
  • Scottish Heart Health Extended Cohort (SHHEC)
  • MONICA Risk Genetics Archiving and Monograph (MORGAM)

See page 268 for the editorial comment on this article (doi:10.1093/eurheartj/eht479)


Troponin is a structural protein released into the circulation following cardiac myocyte damage. The advent of high-sensitivity assays with reliable measurements at very low troponin levels has facilitated earlier exclusion or confirmation of the diagnosis myocardial infarction.14 Troponin T, measured by a high-sensitivity assay, detect concentrations in up to 30% in the general population aged 30–65,5 whereas the application of the newly available high-sensitivity troponin I assays enables troponin I detection in 80–90% in similar populations.6 Whether these detectable levels at very low concentration reflect myocyte (micro-) necrosis or are raised in pathologies other than coronary heart disease remain unclear.7,8 Recent work indicates that detectable troponin levels are associated with silent ischaemia and other cardiac abnormalities, such as left ventricular hypertrophy, left ventricular dysfunction, and left atrial enlargement.9 Irrespective of its origin, troponin detected by a high-sensitivity assay currently allows for risk stratification in stable coronary artery disease patients and the general elderly population.1013

The objective of the present study was to evaluate whether the detection of troponin levels measured by a high-sensitivity assay improves cardiovascular risk prediction in the general population of the Scottish Heart Health Extended Cohort (SHHEC) study14 and to directly identify applicable thresholds for risk prediction and potential adaption of clinical management.


All analyses and new biomarker measurements presented were performed within the frame of the MORGAM (MONICA Risk, Genetics, Archiving, and Monograph) Biomarker study15,16 and the BiomarCaRE consortium (www.biomarcare.eu).

Study description

SHHEC combines the Scottish Heart Health Study (subjects recruited from Scotland between 1984 and 1987) and the Scottish MONICA (subjects recruited in Edinburgh and North Glasgow in 1986, and in North Glasgow in 1989, 1992, and 1995).14 The study was approved by all appropriate ethics committees; all participants signed an informed consent for baseline examination and consented to follow-up of their medical records. For the present study, samples and data were used from all subcohorts of the SHHEC.

At recruitment, pre-existing cardiovascular disease was identified by self-report. Subsequent mortality and hospital illness episodes were followed to the end of 2005 through record linkage of deaths and hospital admissions.17

The main outcome measures were first occurrences of a major cardiovascular event (including first fatal or non-fatal definite or possible myocardial infarction or unclassifiable coronary death, unstable angina, cardiac revascularization, and probable ischaemic stroke) and coronary death. Additionally, single endpoints of myocardial infarction and stroke were assessed. Events ascertained through 2005 were available for analyses. For all analyses, participants with a history of myocardial infarction and ischaemic or haemorrhagic stroke were excluded. Record linkage was with all hospital admissions in Scotland and all deaths in the UK. Thirty-two persons were lost to follow-up.

The FINRISK study was used to validate the cut-off values established in the SHHEC to differentiate individuals at risk for incident cardiovascular events. FINRISK 1997 is a population-based health examination survey carried out in five geographical areas of Finland and followed up for 14 years. A detailed description of the FINRISK study is given in the Supplementary Material.

Laboratory procedures

Serum and EDTA plasma were separated from venous blood left at room temperature for a maximum of 4 h, and then stored at 4°C (1–3 days) during transfer to the laboratory for long-term storage. Routine measurements were performed and aliquots were stored at −40°C. Serum samples were collected from all surveys of the SHHEC study, but EDTA plasma was only stored in two of the Glasgow MONICA surveys.

All biomarkers were measured in the MORGAM/BiomarCaRE laboratory.

High-sensitivity measurements of troponin I levels were determined by the ARCHITECT STAT highly sensitive Troponin I immunoassay (Abbott Diagnostics, USA, ARCHITECT i2000SR). The LoD for the assay was 1.9 pg/mL. For subsequent analyses, observed values below this limit were included (assay range 0–50 000 pg/mL). The assay supported a 10% coefficient of variation at a concentration of 5.2 pg/mL. Intra-assay and inter-assay coefficients of variation were 4.26 and 6.29%, respectively.

Contemporary-sensitivity measurements of troponin I levels were determined by the STAT Troponin I immunoassay (Abbott Diagnostics, USA; ARCHITECT i2000SR). Measurements were considered valid for values above the LoD of 10 pg/mL, but observed values below this limit were also used in the analyses (assay range 0–50 000 pg/mL). The 10% coefficient of variation corresponded to 32 pg/mL.15 Intra-assay and inter-assay coefficients of variation were 2.36 and 4.80%, respectively.

C-reactive protein was determined by Latex immunoassay (CRP Vario CRP16, Abbott Diagnostics, USA, ARCHITECT c8000) and its intra-assay and inter-assay coefficients of variation were 0.93 and 3.91%, respectively.

Measurement of B-type natriuretic peptide (BNP) was performed in EDTA-plasma using a chemiluminescent microparticle immunoassay (BNP, Abbott Diagnostics, USA, ARCHITECT i2000SR). Its intra-assay and inter-assay coefficients of variation were 2.11 and 4.28%, respectively. Because EDTA plasma samples were not available from all SHHEC subjects, measurement of BNP was performed in only 2609 EDTA plasma samples.

Statistical methods

Statistical procedures followed those described previously.15 Baseline characteristics are expressed as counts and percentages for categorical variables, means, and standard deviations for normally distributed continuous variables or medians and first and third quartiles for skewed continuous variables. Spearman's rank-order correlation coefficients between variables were calculated.

Because of severely skewed distributions troponin I, BNP and C-reactive protein values were transformed to their cube roots and the highest measured values were replaced with the second highest value. Multiple imputation was used for missing values of the biomarkers.15,18 Imputation was performed with the use of WinBUGS software.19 The numbers of missing values for the biomarkers are shown in Supplementary material online, Table S1.

For some of the analyses, categories were defined for high-sensitivity assayed troponin I values using the cut-points 1.9 pg/mL (considered LoD), 4.8 pg/mL (median value between 1.9 and 12.7), and 12.7 pg/mL [which corresponds to the LoD for the contemporary-sensitivity troponin assay (10 pg/mL)]. The associations between biomarkers and survival were estimated by a Cox proportional hazards model using age as survival time and illustrated by age-adjusted survival curves using follow-up years as survival time. In the fully adjusted analysis, models were adjusted for cohorts, sex, and the cardiovascular risk factors including history of diabetes, antihypertensive medication, daily smoking, body mass index, systolic blood pressure, non-HDL and HDL cholesterol, deprivation index, and family history of coronary heart disease; these factors had been published previously as components of the ASSIGN score.20

In prediction modelling, 10-year absolute risks were estimated by a Weibull curve fitted over age and adjusted by the linear predictor of the estimated Cox model with and without a biomarker. Ten-fold cross-validation was used to avoid overfitting. The data set was split randomly into 10 equal-sized validation sets. The prediction model was fitted to each of the 10 data sets obtained by leaving out each of the validation sets in turn. Each fitted model was then applied to the corresponding validation set to obtain 10-year risk estimates.13

Discrimination criteria used to evaluate the prediction models were the net reclassification improvement (NRI), integrated discrimination improvement (IDI),21 and c-index improvement.22 The risk categories used for the NRI analysis were 0–5%, 5–10%, 10–20%, and >20%.

The reclassification probabilities, sensitivities, specificities, and area under the receiver operating characteristic (ROC) curve were calculated using the Kaplan–Meier approach proposed by Heagerty et al.23 taking into account the censoring in time-to-event data. For each troponin I ROC curve, the cut-off where the maximum Youden index24 is achieved was computed. The sensitivity and specificity of the Youden cut-off, computed in SHHEC, was validated in the FINRISK study.

A two-sided P-value of <0.05 was considered statistically significant. All statistical methods were implemented in R statistical software version 2.15.025 (www.R-project.org).


Study population and baseline characteristics

The baseline characteristics of the 15 340 SHHEC study participants are shown in Table 1. During an average of 20 years of follow-up, 2171 participants (14.2%) experienced a cardiovascular event and 714 (4.7%) coronary deaths occurred. Non-fatal myocardial infarction and stroke were documented for 1980 (12.9%) and 797 (5.2%) participants, respectively.

View this table:
Table 1

Characteristics of the SHHEC study population

When using the high-sensitivity troponin I assay, 82.6% of men and 67.0% of women had troponin I concentrations above the LoD of 1.9 pg/mL. To test the impact of long-term storage of samples in SHHEC, we compared the troponin I concentrations in SHHEC with those obtained from 3 years stored samples (−80°C) of a population-based study26 (Supplementary material online, Figure S1). The overall concentration of troponin I assessed by the high-sensitivity assay had a very similar distribution in both studies. A correlation matrix for baseline risk factors and biomarkers is shown in Supplementary material online, Figure S2.

High-sensitivity assayed troponin I and association with cardiovascular outcomes

Figure 1 displays fully adjusted survival curves according to the categories of troponin I measured by the high-sensitivity assay, separately for men and women. The relationship between high-sensitivity assayed troponin I and incident cardiovascular events was stronger in older male subjects. Supplementary material online, Table S2 outlines the distribution of the troponin I categories and their relationship with outcome according to age group and gender.

Figure 1

Survival curves assessing the time to event across troponin I categories measured by a high-sensitivity assay (hsTnI). Cox model based survival curves adjusted for age, cohort and all risk factors in the ASSIGN score.20 P-values are for the comparison of the highest and lowest hsTnI categories. Total number of individuals at risk when the follow-up started was 7598 men and 7742 women. At year 10 there were 6399 and 6856 men and 7075 and 7331 women in the at risk population for cardiovascular events and coronary death, respectively. At year 20 the size of the population at risk were 1154 and 1358 men and 1413 and 1514 women.

The fully adjusted hazard ratios of cardiovascular events increased with increasing troponin I categories and individuals within the highest category had 2.5 times the risk of those with non-detectable troponin I levels (95% CI 1.60, 2.61). Those individuals within the lowest detectable troponin category were independently associated with the cardiovascular risk (Figure 2). Of particular interest, this association remained significant when restricting the analysis to individuals in whom troponin I levels were not detectable by the use of the contemporary-sensitivity assay. Detailed results are provided in Table 2 and Supplementary material online, Figure S3. The association of high-sensitivity assayed troponin I with incident cardiovascular events and coronary deaths according to the time of follow-up is shown in Figure 3. These associations remained significant throughout most of the 20-year follow-up.

View this table:
Table 2

Hazard ratios according to cubic root transformed troponin I measured by a high-sensitivity assay in the SHHEC cohort (all) and considering only individuals with contemporary-sensitivity troponin I measures at 0 pg/mL (cTnI = 0)

Figure 2

Hazard ratios and 95% confidence intervals in four categories of troponin I measured by a high-sensitivity assay (hsTnI) over a follow-up of 20 years. Cox models were adjusted for sex, cohort and all risk factors in the ASSIGN score20 using age as the survival time. Category hsTnI < 1.9 pg/mL is used as a reference.

Figure 3

Hazard ratios according to cubic root transformed baseline troponin I measured by a high-sensitivity assay (hsTnI) over a follow-up of 20 years. Cox models were adjusted for sex, cohort and all risk factors in the ASSIGN score,20 fitted in windows of five-year follow-up periods and using age as the survival time.

Further to the combined endpoint and coronary death, troponin measured by the high-sensitivity assay associated significantly with the single endpoints myocardial infarction and stroke. Corresponding hazard ratios for those single endpoints are given in Supplementary material online, Tables S3 and S4.

High-sensitivity troponin I measurements and prediction of cardiovascular events

The ability of the fully adjusted prediction model to discriminate events from non-events was significantly improved after inclusion of high-sensitivity assayed troponin I. The NRI was estimated at 3.2% (P = 0.0005) for cardiovascular events and 4.5% (P = 0.012) for coronary death in the overall population. Addition of troponin I to the prediction model led to small but significant increases in the c-statistics for cardiovascular events and coronary death [c-index difference 0.0044 (P < 0.0001) and 0.0077 (P < 0.0001), respectively]. Detailed prediction metrics for coronary death and combined cardiovascular events for the biomarkers assessed are outlined in Table 3. High-sensitivity troponin I values measured in only those individuals with non-detectable contemporary-sensitivity troponin I levels still provided a modest but significant improvement in risk prediction (Supplementary material online, Table S5). Supplementary material online, Table S6 outlines prediction data for all biomarkers with respect to the single endpoint events myocardial infarction and stroke.

View this table:
Table 3

Improvement of 10-year risk prediction by troponin I measured by high-sensitivity assay for endpoints coronary death and major cardiovascular events over first 10 years of follow-up in the SHHEC cohort

Determination of a clinically applicable threshold of high-sensitivity measured troponin I

We assessed an optimal cut-off value of high-sensitivity assayed troponin I for cardiovascular events with a ROC analysis separately applied in both genders in SHHEC. A cut-off value of 7.0 pg/mL with a sensitivity of 39.1% (95% CI: 34.9–43.4%) and a specificity of 81.8% (95% CI: 80.8–82.8%) for men and 4.7 pg/mL with a sensitivity of 52.4% (95% CI: 46.5–58.3%) and a specificity of 68.6% (95% CI: 67.4–69.7%) for women was detected. Validation of these different cut-off values was carried out in the FINRISK study as displayed in Figure 4. Next, we applied the gender-specific cut-off values in SHHEC and calculated the improvement in cardiovascular risk prediction. This resulted in a NRI of 13.4% (P < 0.0001) in men and 13.1% (P < 0.0001) in women. Details of this prediction analyses are shown in Table 4.

View this table:
Table 4

Application of a gender-specific clinical cut-off value for risk prediction of cardiovascular events in SHHEC in men and women

Figure 4

Receiver operating characteristic curve for various cut off levels of high-sensitivity measured troponin I in differentiating risk of cardiovascular events in males and females. Optimal cut-off values were derived from SHHEC, sensitivities and specificities of the different cut-off values were calculated separately in men and women and validated in FINRISK. The optimal hsTnI cut-off in SHHEC is 7.0 pg/mL in men and 4.7 pg/mL in women. SHHEC, Scottish Heart Health Extended Cohort; *indicates optimal hsTnI cut-off in SHHEC is 7.0 pg/mL in men and 4.7 pg/mL in women, CI, confidence interval.


The main objective of the present study was to evaluate the usefulness of high-sensitivity troponin I measurements for prediction of major cardiovascular events and death in the general population over a wide age range. Assessment of troponin I levels detected in nearly 75% of the overall population added predictive information beyond established cardiovascular risk models in the general population. This information was obtained in a Scottish population with a high incidence of cardiovascular disease and very long follow-up. These findings expand upon previous studies5,11,12,15 in several ways. First, we used the latest generation of a robust high-sensitivity cardiac troponin I assay, allowing us to reliably detect troponin I concentrations in the vast majority of the general population, not restricted to elderly or high-risk individuals. Second, we demonstrated the predictive value across a wide range of ages in men and women for fatal and non-fatal cardiovascular events and coronary death. The use of this assay led to a significant reclassification of risk in the entire population with the strongest effect observed for fatal events. Third, there is some indication that the improvement in risk discrimination between using the high-sensitivity troponin assays and those assays detecting troponin in a lower number of individuals is explained by the low troponin levels that could not be detected reliably using an assay with lower sensitivity. However, in the overall population, the accuracy of high-sensitivity measured troponin I in risk predication was not superior to that of contemporary-sensitivity troponin I measurements.

Whenever additional risk factors, however powerful, are added to risk scores, it becomes increasingly difficult to improve discrimination.21 Therefore, the significant independent contribution to cardiovascular risk assessment of circulating biomarkers over and above the multiple pre-existing cardiovascular risk factors is of considerable interest. However, the absolute improvement is more modest than that found previously when combining the three biomarker hs-C-reactive protein, NT-proBNP, and a contemporary-sensitivity assayed troponin.15

Studies from at least three large epidemiological cohorts and a follow-up study of a randomized trial have been published that used a more sensitive troponin T assay, and detected those levels in up to 255 and 35%27 in the general population and 66.2%,12 respectively, 66.5%11 in the elderly. Detectable troponin T concentrations were related to incident events such as heart failure, overall death, and cardiac mortality in these studies. The present large-scale analyses of the Scottish population, however, used a robust high-sensitivity troponin I assay that allows reliable troponin I detection at very low levels in the general population. These low detected values improved risk stratification modestly but significantly. However, common to all studies, including the present study, is the finding that troponins measured by high-sensitivity assays did not lead to a superior risk prediction achieved by BNP.

Of further interest is that biomarkers might predict cardiovascular risk differently in women and men. Given the different baseline values for BNP, C-reactive protein, and high-sensitivity assayed troponin in the population, it might appear reasonable to expect that they will render different predictive information in women and men separately. So far, insufficient data are available to draw a conclusion about gender-specific risk prediction in relation to biomarkers, but this will be an area of intensive research in the near future.

With respect to clinical application beyond risk estimation, diagnostic thresholds need to be identified and therapeutic consequences in those individuals above the diagnostic threshold need to be assessed. We explored and validated an assay specific cut-off of 7.0 pg/mL in men and 4.7 pg/mL in women in two population-based studies for cardiovascular events (SHHEC and FINRISK). Individuals above this threshold had a higher risk of suffering from cardiovascular events in the future. When applying these dichotomous cut-off values separately in men and women, substantial reclassification information was obtained. Obviously, this information needs to be confirmed in independent studies. The next step required is to retrospectively and—ultimately—prospectively test whether high-risk individuals identified by troponin thresholds benefit from preventive therapies such as aspirin and/or statin medication.


Some limitations merit consideration: (i) samples had been in long-term storage at −40°C in the SHHEC. However, troponin I levels measured by the high-sensitivity assay were comparable with those from a reference cohort26 (Supplementary material online, Figure S1), where samples were stored at −80°C for only 3 years, implying that archived material had not deteriorated, and the biomarker assessment was robust. (ii) Only a limited number of BNP measures are available in the present study due to limited availability of EDTA-plasma samples. (iii) Validated heart failure events are not available for these analyses limiting our ability to test any association of troponin I high-sensitivity measures to heart failure. (iv) Although analyses for likely ischaemic stroke as an endpoint had also been performed (data not shown), the number of such cases was less than a third of all strokes, and it was not possible to conclude a difference in the results. (v) Detection of very low troponin I concentrations below the 10% CV is associated with higher imprecision. However, given the predictive capacity of those values, detection of this low concentration still adds to the predictive information.


Determination of troponin I by a high-sensitivity assay, with measurable results in about 75% of a middle-aged population, predicts cardiovascular disease over a 20-year time period in the general population in both men and women. Our results provide evidence that high-sensitivity assayed troponin I is a cardiac-specific marker of global cardiovascular risk and might be considered for incorporation into risk scores. It may reflect pathophysiological progression from myocardial health to myocardial damage. From a clinical perspective, the application of high-sensitivity assayed troponin I for cardiovascular risk stratification and decision making in routine clinical practice is not yet ready for prime time. Although improving prediction metrics, it still needs to be proven that any changes in treatment consequent upon measurement and risk reclassification can lead to improved patient outcomes. This would constitute a milestone in personalized cardiovascular medicine based on biomarker assessment.


This work was supported by the Scottish Health Department Chief Scientist Organization; the British Heart Foundation; the PF Charitable Trust (for the Scottish Heart Health Extended Cohort); the European Union's Seventh Framework Programme FP7/2007–2013 (HEALTH-F4-2007-2014113, ENGAGE and HEALTH-F3-2010-242244, CHANCES) (for MORGAM); the Medical Research Council London (G0601463, No 80983: Serum Biomarkers in the MORGAM Populations); the European Union's Seventh Framework Programme FP7/2007-2013 (HEALTH-F2-2011-278913, BiomarCaRE) (for the MORGAM Biomarker study); the Australian National Health and Medical Research Council (grant 571281 and fellowship 1020812 to M.W.). V.S. was supported by the Finnish Foundation for Cardiovascular Research and the Academy of Finland (grant #139635).

Conflict of interest: Abbott Diagnostics provided test reagents for highly and contemporary sensitive troponin I, C-reactive protein, and BNP for determinations within the frame of the MORGAM Biomarker Study and the BiomarCaRE project. S.B. has received honoraria from Abbott Diagnostics, SIEMENS, Thermo Fisher, and Roche Diagnostics and is a consultant for Thermo Fisher. All other co-authors reported no conflicts of interest.


This work has been sustained by the MORGAM Project's current funding: European Community FP 7 projects ENGAGE, CHANCES and BiomarCaRE. This has supported central coordination, workshops and part of the activities of The MORGAM Data Centre, at THL in Helsinki, Finland. MORGAM Participating Centres are funded by regional and national governments, research councils, charities, and other local sources.


View Abstract