Article Text

Download PDFPDF

Original article
Sex-specific versus overall cut points for a high sensitivity troponin I assay in predicting 1-year outcomes in emergency patients presenting with chest pain
  1. Louise Cullen1,2,3,
  2. Jaimi H Greenslade1,2,3,
  3. Edward W Carlton4,
  4. Martin Than5,
  5. John W Pickering6,
  6. Ariel Ho1,
  7. Kim Greaves2,7,
  8. Sara L Berndt1,
  9. Richard Body8,
  10. Kimberley Ryan1,
  11. William A Parsonage1
  1. 1Department of Emergency Medicine, Royal Brisbane and Women's Hospital, Brisbane, Queensland, Australia
  2. 2School of Medicine, The University of Queensland, Brisbane, Queensland, Australia
  3. 3School of Public Health, Queensland University of Technology, Brisbane, Queensland, Australia
  4. 4Emergency Department, Southmead Hospital, Bristol, UK
  5. 5Department of Emergency Medicine, Christchurch Hospital, Christchurch, New Zealand
  6. 6Emergency Care Foundation, Christchurch Hospital, Christchurch, New Zealand
  7. 7Department of Cardiology, Sunshine Coast Hospital and Health Services, University of the Sunshine Coast, Nambour, Australia
  8. 8Emergency Department, Central Manchester University Hospitals NHS Foundation Trust, Manchester, UK
  1. Correspondence to Dr Louise Cullen, Department of Emergency Medicine, Royal Brisbane and Women's Hospital, Butterfield St, Herston Brisbane, QLD 4029, Australia; louise_cullen{at}health.qld.gov.au

Abstract

Objective To evaluate the incidence of major adverse cardiac events (MACE) at 1 year in emergency department (ED) patients with possible acute coronary syndromes, stratified by high sensitivity troponin (hs-cTnI) concentrations using sex-specific cut points compared with overall cut points.

Methods In a multicentre observational study of 2841 patients, presentation hs-cTnI concentrations were categorised using sex-specific (women 16 ng/L; men 34 ng/L) and overall (26 ng/L) cut points. The primary outcome was MACE occurring within 1 year of presentation. Patients with hs-cTnI values concentrations within these categories were reported by sex and 1-year MACE. Net reclassification improvement (NRI) was computed to measure the change in prediction after altering the hs-cTnI cut points, and was calculated separately for events and non-events.

Results Application of sex-specific 99th percentile cut points rather than the overall cut point of 26 ng/L, reclassified 25 females from having a non-elevated troponin to having an elevated troponin, and 29 males from having an elevated troponin value to having a non-elevated troponin value on presentation. Of these, 7 (28.0%) females and 12 (41.4%) males had a 1-year MACE. There was no reclassification improvement for those with or without 1-year MACE (NRIevents=−1.5%, 95% CI −4.0% to 1.1%; NRInon-events −0.04%, 95% CI −0.5% to 0.4%).

Conclusions Sex-specific cut points improve the identification of women but not men at risk for 1-year MACE. The net-effect across the whole ED population with possible cardiac chest pain is minimal. Lowering the clinical cut point for both sexes may be appropriate for prognostic purposes.

Trial registration number ISRCTN No. 21109279, ACTRN12609000283279.

View Full Text

Statistics from Altmetric.com

Introduction

Improvements in the analytical performance of troponin assays have led to the identification of sex-specific normal ranges in healthy populations, with the 99th percentile of females in a healthy reference population being lower than that of males.1 ,2 To distinguish between elevated and normal troponin values for the clinical definition of acute myocardial infarction (AMI), a single cut point of the 99th percentile has been recommended.3 This cut off falls between the female and male 99th percentiles. Within the emergency department (ED), the initial troponin concentration often dictates further assessment processes and influences disposition, underpinning the importance of the identification of patients with elevated troponin results.

Sex-dependent cut point values for other assays exist; some are widely used clinically (eg, haemoglobin), while others are not (eg, natriuretic peptides). Clinical health professionals are largely unaware of the potential implications of sex-specific cut points for cardiac troponin assays. There are possible benefits, particularly for women, in improved identification of patients at risk of short-term and long-term adverse cardiac events, including AMI and death.4 ,5 The exact implications of the clinical use of sex-specific cut points remain unclear.

Across all ages, women have a lower incidence of AMI in comparison to men.6 Women who present with acute coronary syndrome (ACS) are less frequently diagnosed with the condition and often have worse outcomes than men.7–9 Although the reasons for this are multifactorial, cardiac injury in women may be under-recognised. In particular, this may occur in women who present with initial troponin concentrations below the overall 99th percentile cut point but higher than the sex-specific 99th percentile value for women. Recent studies have had mixed findings, with some reporting a benefit in using sex-specific cut points for females4 ,10 and others suggesting this may have little clinical impact.5 ,11 Longer-term studies of prediction of risk which could identify the utility of the use of sex-specific cut points have yet to be undertaken.

Sex-specific cut points for highly sensitive assays are in clinical use, however different cut points are being used and reported in sites using the Abbott hs-cTnI assay (Hans Schneider, Director of Pathology, Alfred Health, 2015, personal communication). In sites in New Zealand, both the Abbott recommended reference sex-specific (female 16 ng/L and male 34 ng/L) and overall (26 ng/L) intervals are used in clinical practice. In Australia, a hybrid of these values is the widely agreed cut points, with 16 ng/L for females and 26 ng/L for males in clinical use.

This study investigated the use of sex-specific cut points in the prediction of major adverse cardiac events (MACE) in a multicentre ED population presenting with possible cardiac chest pain 1 year after presentation. We used sex-specific cut points in comparison to overall cut points for initial highly sensitive troponin assay results. We also investigated the net effect of classification with different cut points.

Method

Design

The study population consisted of patients recruited into two prospective trials. One cohort was enrolled at two tertiary teaching hospitals (Royal Brisbane and Women's Hospital, Australia and Christchurch Hospital, New Zealand), in the 2-Hour Accelerated Diagnostic Protocol to Assess Patients with Chest Pain Symptoms Using Contemporary Troponins as the Only Biomarker (ADAPT) study. The second cohort was enrolled at Poole Hospital NHS Foundation Trust, UK in the Triage Rule-out Using high-Sensitivity Troponin (TRUST) Study. All patients recruited in the trials were eligible for enrolment if they presented with symptoms suggestive of cardiac ischaemia (acute chest, epigastric, neck, jaw or arm pain or discomfort or pressure without an apparent non-cardiac source). Patients were excluded if any of the following were present: ST-segment elevation myocardial infarction or new left bundle branch block, ECG changes diagnostic of ischaemia (ST segment depression ≥1 mm or T-wave inversion consistent with ischaemia), significant arrhythmias (new-onset atrial fibrillation, atrial flutter, sustained supraventricular tachycardia, second-degree or complete heart block, or sustained or recurrent ventricular arrhythmias), age <18 years, a clear cause other than ACS for the symptoms, pregnancy, inappropriate recruitment (eg, terminal illness), transferred from another hospital, unwillingness to consent, and if follow-up was considered impossible. Six patients were excluded from the TRUST as they were lost to follow up at 1 year. Four hundred and seventy-four patients from the ADAPT cohort were excluded from this analysis because they did not consent to be followed up at 1 year. Written informed consent was obtained from all patients and the study protocol had been approved by the respective local ethics committees (HREC 2008/101 and HREC/10/QRBW/403 at Royal Brisbane and Women's Hospital, URA/07/06/048/AM01 at Christchurch and TRUST 12/SW/0133 at Poole Hospital).

Laboratory troponin (cTn) levels were measured at presentation (0 h) and again at least 6 h later, with additional samples stored at presentation (0 h) for later analysis. The cTn (sensitive and highly sensitive) assays used for clinical management and adjudication of outcomes are described in the online supplementary web appendix 1. In line with manufacturer recommendations at the time of the study, adjudication was based on the 99th percentile for the cTn assays used at two of the hospitals and the 10% coefficient of variation for the cTn assay used at the third hospital. Research samples taken on presentation were centrifuged and serum stored frozen at −80°C for later analysis in a blinded fashion using the hs-cTnI (Abbott Diagnostics, ARCHITECT STAT hs-cTnI assay, LoD 1.2 ng/L, 10% Coefficient of variation (CV) at 4.7 ng/L). The overall decision cut point, the 99th percentile of a healthy reference population, as recommended by the manufacturer, was 26 ng/L. The sex-specific cut points evaluated were also those recommended by the manufacturer. These were female and male cut points defined as 16 ng/L and 34 ng/L, respectively.4

ECGs were recorded at presentation and ≥6 h later, during recurrent symptomatic episodes and if requested by medical staff. Patients were managed according to local protocols, as previously published.12–14 All clinical management including the decision to perform stress testing, coronary angiography and/or other management was at the discretion of the attending clinician.

Patient data were recorded according to standardised data collection forms using a published data dictionary.15 Patient diagnoses on admission were independently adjudicated by cardiologists with knowledge of the local cTn assay results, ECGs and all other clinical information available up to 30 days after presentation. Patients were followed up for 1 year for all-cause mortality and new diagnoses of ACS (defined as for the index admission). One-year follow-up events were monitored by dedicated research staff through a combination of both telephone contact, corroboration by review of the hospital online patient management system and query to the national death registries at least 12 months after index presentation.

Outcomes

The primary outcome was MACE occurring within 1 year of presentation, including events that occurred during the index presentation. MACE was defined as AMI, emergency coronary revascularisation (including percutaneous coronary intervention and coronary artery bypass grafting) or death (unless clearly non-cardiac). The diagnostic criteria for AMI as defined by the Universal Definition were the detection of a rise and/or fall of cTnI with at least one value above the overall 99th percentile in the setting of symptoms of ischaemia, with or without new or presumed new significant ST-segment/T wave changes/new left bundle branch block/development of pathological Q waves or imaging evidence of new loss of viable myocardium or new regional wall motion abnormality.3

Data analysis

Data were analysed using Stata V.12 (StataCorp, College Station, Texas, USA). Descriptive statistics were used to compare the baseline characteristics of the sample across 1-year MACE and non-MACE groups. To identify the performance of the hs-cTnI assay, receiver operating characteristic (ROC) curves for 1-year MACE were computed separately for males and females. Optimal cut points were derived using two methods. The first was the Youden index. This index finds the cut point that maximises the vertical point between the ROC curve and the line of equality. The second was based on logistic regression modelling of 1-year MACE. Using this approach, each cut point value for the hs-cTnI variable is entered as a covariate in a logistic regression model and the c-statistic is computed. The model with the highest c-statistic is considered to be optimal. Hs-cTnI data were categorised as ≤16 ng/L, 17–26 ng/L, 27–34 ng/L and >34 ng/L to allow reporting of different patient groups. The number and percentage of individuals with hs-cTnI values within these categories was reported by sex and 1-year MACE. Net reclassification improvement (NRI) was computed to measure the change in prediction after altering the hs-cTnI cut points. The NRI was calculated separately for events and non-events.

Logistic regression was used to test the relationship between hs-cTnI and 1-year MACE. Logistic regression analyses were adjusted for hospital, age and sex. An interaction term between hs-cTnI and sex was included in the model to test whether the relationship between troponin levels and MACE differed by sex. Model fit was assessed using the Hosmer–Lemeshow (HL) goodness of fit test.

Results

Data were available for 2841 patients, including 1164 from Australia, 810 from New Zealand and 867 from UK; 1661 (58.5%) were male and the mean age was 57 years (SD=14, range=18–95). Characteristics of the sample by outcome, sex and by the hospital are shown in tables 1 and 2, respectively. Patients with ACS were older, more likely to have risk factors and more likely to have a prior cardiovascular history than patients without ACS. Patients from New Zealand had the highest event rate, were older and had more risk factors than the Australian sample. Patients from UK were similar to New Zealand in terms of risk factors and fell between the New Zealand and Australian cohorts in terms of the history of reported medical conditions. Figure 1 provides the presentation hs-cTnI for males and females by the hospital and shows that females have lower hs-cTnI on presentation compared with males.

Table 1

Characteristics of the cohort by 1-year MACE

Table 2

Characteristics of the cohort by hospital

Figure 1

hs-cTnI for males and females by hospital. Data are presented on a logarithmic scale.

ROC analysis for 1-year MACE had an area under the ROC curve of 0.88 (95% CI 0.84 to 0.90) for males and 0.92 (95% CI 0.89 to 0.96) for females (figure 2). Derived optimal cut point concentrations based on the Youden index were ≥6 ng/L for females (sensitivity 89.3%; specificity 84.5%) and ≥7 ng/L for males (sensitivity 77.8%; specificity 84.9%). Optimal concentrations based on logistic regression modelling were ≥6 ng/L for females (sensitivity 89.3%; specificity 84.5%) and ≥9 ng/L for males (sensitivity 74.8%; specificity 88.7%).

Figure 2

Receiver operating characteristic (ROC) analysis for 1-year major adverse cardiac events for males and females. Embedded Image Males: Area under the Receiver Operating Characteristic (AUROC) 0.88. Embedded Image Females: AUROC 0.92. p=0.03.

The number of male and female patients with troponin values in each of the categories ≤16 ng/L, 17–26 ng/L, 27–34 ng/L and >34 ng/L are presented in table 3. Application of the sex-specific 99th percentile cut points (16 ng/L and 34 ng/L, female and males respectively) compared with the overall cut point of 26 ng/L, reclassified 25 females from having a non-elevated troponin to having an elevated troponin, 7 (28.0%) of whom had 1-year MACE. Twenty-nine males would be reclassified from having an elevated troponin value to having a non-elevated troponin value on presentation, 12 (41.4%) of whom had MACE. The proportion of patients considered to have hs-cTnI elevation changed from 9.29% with the overall cut point to 9.15% by the use of sex-specific cut points. Using sex specific cut points did not result in an overall net gain in classification (NRI=−1.5%, 95% CI −4.0% to 1.1%). There was no reclassification improvement for those with 1-year MACE (NRIevents=−1.5% (95% CI −4.0% to 1.1%), or those without 1-year MACE (NRInon-events −0.04%, 95% CI −0.5% to 0.4%).

Table 3

Number of males and females in each troponin category by 1-year MACE

Online supplementary web appendix 2 provides details of patients with 1-year MACE with overall and sex-specific cut points.

If an overall cut point of 16 ng/L were used (instead of 26 ng/L) for all patients, 343 (12.1%) patients would be considered to have an elevated troponin, an increase of 79 patients. Of these 79 patients, 25 (NRI events=7.4% (4.5% to 10.3%) had a 1-year MACE and 54 (NRInon-events −2.2% (−2.7% to −1.6%) did not have a MACE. Online supplementary web appendix 3 provides details of the serial troponin values used in clinical care and disposition of such patients. Adopting a cut point of >16 ng/L for females only and leaving the cut point at 26 ng/L for males would result in 289 patients (10.2%) classified as having an elevated troponin, an increase of 25 patients. Of these 25 patients, seven had MACE (NRIevents=2.1% (0.5 to 3.6%)) and 18 did not have a MACE (NRInon-events=−0.7% (−1.1% to −0.4%)). Figure 3 provides the proportion of individuals within each hs-cTnI category who were diagnosed with 1-year MACE.

Figure 3

Proportion of patients within each hs-cTnI category with 1-year major adverse cardiac events. Embedded Image Male. Embedded Image Female.

Logistic regression was conducted including age, gender, troponin category, hospital and the interaction between troponin category and gender. ORs are reported separately for males and females. The odds of MACE increased with increasing troponin values across both sexes (table 4). The odds of MACE for females were lower than males for troponin values in the range of 3–16 ng/L and >34 ng/L. There was no difference in the odds of 1-year MACE for males and females with troponin values in the range of 17–26 ng/L or 27–34 ng/L. This model provided a good fit to the data (p=0.33).

Table 4

Logistic regression on the entire cohort

Limitations

Our study has several limitations. Only patients presenting with a chest pain syndrome were enrolled. Patients with AMI may present atypically with shortness of breath, nausea or syncope without chest pain. These patients were not included, and thus our findings are not applicable to all patients with an eventual diagnosis of AMI. The finding of an elevated troponin value alone is not indicative of AMI, hence we assessed reclassification of troponin values rather than new diagnosis of AMI using different cut-off values. We used both sensitive and highly sensitive troponin assays for the adjudication of the outcomes. It is possible that some patients with concentrations of troponin close to the cut point were incorrectly diagnosed with or without AMI due to imprecision of sensitive assays. We used overall cut points for the adjudication of events. This may have led to misclassification of index events. We used the time of presentation as the first sample, regardless of when the onset of symptoms occurred. Early and late presenters were represented equally in our cohort without a significant difference in the performance of the sex-specific cut points between these groups. Cardiac troponin I levels were determined at study inclusion (0 h). This result is of use for ED risk stratification however may not reflect the peak levels that may be associated with cardiac outcomes. Patients with suspected AMI would go onto have serial troponin measurements and it may be that peak troponin levels do not differ between males and females.16

Discussion

This is the first large multi-centred, international study of ED patients that reports the effect of sex-specific cut points for prediction of MACE in comparison to the use of overall cut points. In a high risk cohort, that is ED patients with possible ACS, our study investigates the prevalence of ischaemic heart disease over 1 year from presentation. It confirms that the use of sex-specific cut points improves the identification of women at high risk for cardiovascular events, within 1 year of presentation to the ED with chest discomfort, even when based on the single presenting ED value. This finding is in line with a prospective cohort study using the Abbott high sensitivity troponin I assay at a single UK site,4 and with research conducting using patients with confirmed non-ST elevation acute coronary syndromes (NSTEACS).5 ,17 This is the first study conducted across multiple EDs in patients with symptoms of possible ACS. In a recent study by Shah et al4 sex-specific cut points of a high-sensitivity cardiac troponin assay were shown to markedly increase the rate of diagnosis of AMI in women from 11% to 22%, but without decreasing the rate in men in ED patients. Additionally, in all-comers to the ED, the use of sex-specific cut points with a highly sensitive troponin assay for the prediction of hospital death have suggested to be of benefit for female patients;10 however other studies have reported contrasting findings. For example, in ED patients with acute chest pain, the differences in the use of sex-specific cut points for the diagnosis of AMI are reportedly small.11 In patients diagnosed with NSTEACS, the use of sex-specific cut points did not improve prognostic accuracy for 30-day AMI or 1-year mortality5 and the use of sex-specific cut points in patients with known ACS have not been supported.17

The choice of which sex-specific cut points are used needs to be carefully considered. Lowering the cut points will increase the number of patients deemed to have elevated troponin values. The benefit of improved identification of patients at long-term risk of MACE in whom treatment may be offered, may be offset by the larger volume of patients needing assessment for elevated troponin. If the cut point is lowered to the sex specific value for females only (16 ng/L) for every seven additional females who go on to have a MACE, 18 other female patients will be classified as having elevated troponin values requiring additional investigation. In our opinion, this is an acceptable increase in the identification of patients with cardiac injury not associated with AMI for the additional benefits.

Given that the patients in this study were clinically managed according to the use of troponin results using the overall cut point, our findings are important. While the net effect of the use of sex-specific cut points of 16 ng/L and 34 ng/L for females and males, respectively, across the whole ED population with chest pain is minimal, there may be an advantage to females and a disadvantage to males in terms of the potential for missed identification of males at high risk for events. Most studies on sex-specific cut points are limited by the lack of use of high sensitivity assays for clinical care17 or the use of overall cut point for both clinical care and the adjudication of AMI.4 ,5 ,11 These factors may limit the identification of possible early benefits of using sex-specific cut points for index AMI diagnosis. The magnitude of the effect in using sex-specific cut points to identify patients at long-term risk of MACE described in our study requires additional testing in a prospective study in a well-described ED cohort in which sex-specific cut points are in clinical use; however, we believe some male and female patients with troponin values >16 and ≤26 ng/L would benefit by being correctly identified during the index presentation.

The study adds to the body of evidence that questions current ideas about differences in event rates and the pathophysiology of AMI between males and females.4 ,5 ,17 The odds of a 1-year MACE events are similar for women and men where troponin values are greater than the sex-specific 99th percentile but below the overall cut point. We should question whether our current expectation of fewer cardiac events in females in comparison to males is simply related to our lack of ability to show sex-specific differences due to previous (less sensitive) troponin assays inability to reliably detect low abnormal concentrations of troponin.

Conclusion

Sex-specific cut points improve the identification of women at risk for 1-year MACE but the net effect across the whole ED population with symptoms of possible cardiac chest pain is minimal. The potential bias associated with using overall cut points to adjudicate endpoints in this study must be considered, and a prospective trial reporting outcomes with the clinical use of sex-specific cut points is needed. Lowering the clinical cut point for both sexes may be appropriate for prognostic purposes.

Key messages

What is already known on this subject?

  • Improvements in the analytical performance of troponin assays have identified sex-specific normal ranges in healthy populations, with the 99th percentile of females in healthy reference populations being lower than that of males.

What might this study add?

  • This study, the first large multi-centred, international study of emergency department (ED) patients, reports the effect of sex-specific cut points for prognosis of major adverse cardiac events (MACE). Sex-specific cut points improve the identification of women but not men at risk for 1-year MACE, and the net effect across the whole ED population with symptoms of possible cardiac chest pain is minimal.

How might this impact on clinical practice?

  • Current practice and potential bias with the clinical use of overall cut points must be considered. When troponin values are used for prognosis, it may be appropriate to lower the cut point for the Abbott ARCHITECT hsTnI assay.

Acknowledgments

RB acknowledges the support of UK National Institute for Health Research.

References

View Abstract

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

Footnotes

  • Twitter Follow Richard Body at @richardbody

  • Contributors LC: study design, data interpretation, literature search, writing, reviewing, editing. JHG: study design, data collection, data interpretation, writing, reviewing, editing. EC: study design, data interpretation, reviewing, editing. MT: study design, data interpretation, reviewing, editing. JWP: data interpretation, reviewing, editing. AH: literature search, reviewing, editing. KG: study design, reviewing, editing. SLB: reviewing, editing. RB: reviewing, editing. KR: reviewing, editing. WAP: study design, data interpretation, writing, reviewing, editing.

  • Funding The study has been supported by grants from Abbott Diagnostics, who played no role in the design and conduct of the study, the study analyses, the drafting and editing of the manuscript or its final content. The Queensland Emergency Medicine Research Foundation provided research funding (QEMRF-PROJ-2008-002) and supports LC through Fellowship funding. The TRUST study was supported by grants from the UK Royal College of Emergency Medicine and Bournemouth University.

  • Competing interests LC and WAP have received research grants and honorarium from Abbott Diagnostics, Siemens, Radiometer Pacific, Roche and Alere. RB has accepted reagents donated without charge for the purposes of research from Roche, Abbott, Siemens, Alere and Randox; and has accepted travel and accommodation to present research findings from Roche and Randox. JWP has consulted for AM Pharma. EC has received funding from Abbott in support for related research.

  • Patient consent Obtained.

  • Ethics approval The Royal Brisbane and Women's Hospital Human Research Ethics Committee, Upper South A Regional Ethics Committee, Frenchay Research Ethics Committee.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles