What are normal high-sensitivity troponin-T values in a large multi-ethnic Asian population?

Background: Cardiovascular disease is rising exponentially in Asian countries. High-sensitivity cardiac troponin-T (hs-TnT) is vital for evaluating acute coronary syndromes. Although hs-TnT has been available since 2010 and used widely in Asia-Pacific there is paucity of published literature on what constitutes expected hsTnT concentrations as well as the 99th percentile (99P) upper reference limits (URL) in Asians. Almost all the hs-TnT data has been derived on Western populations. Besides, very few studies encompass large enough reference cohorts to mitigate against outlier distortion of the URL. Methods: We determined the 99P URL for hs-TnT (Roche Diagnostics) in 1086 healthy Asian subjects (via questionnaire) aged 40-65 years (mean 52.2+7.24; 543 men) from 3 laboratories. There were at least 100 subjects in each 5-year age-group. Statistical analyses were performed on MedCalc 15.0 software (Ostend, Belgium). The hs-TnT assay has a limit of detection of 5 ng/L. Results: The inter-assay precision was 2.8% and 1.9% at mean hs-TnT concentrations of 23 and 1950 ng/L respectively. The hs-TnT concentrations corresponding to the inter-assay coefficient of variation of 20% and 10% was 7.5 and 11.5 ng/L respectively. Hs-TnT concentrations were detectable in 39.4% of participants; higher in men and individuals >55 y. However, 82.0% of women had undetectable hs-TnT compared to 39.2% of men. This was also seen in 98.5% younger women aged 20-39 (n=260). Detectable hs-TnT was greater in those over 55 years versus younger subjects in women (30.3% and 11%) and men (71% and 54.8%). The overall, male, and female 99th percentile URL were: 17.0,18.6 and 12.0 ng/L respectively. Conclusion: This hs-TnT assay can perform with high sensitivity and exhibits gender and age differences. The 99P URL is similar to some previous reports but not in others. Users need to establish/verify their own decision limits. Young women may not be suitable as reference subjects. Correspondence to: Dr. Tar-Choon Aw, Department of Laboratory Medicine, Changi General Hospital, 2 Simei Street 3, Singapore 529889, Singapore; Tel: 656850-4927; Fax: 65-6426-9507; E-mail: tarchoon@gmail.com


Introduction
The burden of cardiovascular disease is increasing globally [1] with low-and middle-income nations bearing the brunt of this scourge [2] including Asia and South-East Asia [3]. Hospitals here are seeing greater numbers of patients for the evaluation of chest pain and acute coronary syndrome (ACS). In this regard troponin, especially the high sensitivity (hs) variety, is the preferred biomarker for the evaluation of ACS [4][5][6][7][8]. A key metric for decision-making in ACS is the 99 th percentile (99P) upper reference limit (URL) for troponin in the healthy reference population. To derive the 99P URL a sufficiently large cohort (minimally 300 subjects of either gender, preferably 500) [5, [9][10][11] is required to mitigate against outlier distortion of this cutpoint. In Asia-Pacific hs-troponin-T (hs-TnT) has been available since 2010 and hs-troponin-I (hs-TnI) since 2012. Despite their wide usage there is paucity of published literature on what are expected hstroponin concentrations and the 99P URL in Asians. For hs-troponin reference range studies involving at least 600 Asian subjects, there is a solitary publication on hs-TnI (n=1120) [12], but none for hs-TnT. We determined the 99 percentile URL in a large national multi-center Asian population for the hs-TnT assay (Roche Diagnostics Asia-Pacific, Singapore).

Methods
The study was a prospective, cross-sectional national study from 3 College of American Pathologists (CAP) accredited hospital laboratories in Singapore (Changi, Alexandra, and Khoo Teck Puat). It was approved by the respective local institutional review boards. The initial hs-TnT assay verification from Changi General in 2010 is reported here. All 3 sites contributed the data they generated for the reference range study from 2013 to 2015. Their performance for the CAP troponin-T proficiency testing program was satisfactory during this period. at these URL concentrations were all <10%. Men had higher serum hs-TnT than women (p < 0.0001, Mann-Whitney U test). There were 658 samples comprising 445 women and 213 men which contained TnT concentrations below the LoD. Thus, TnT was detectable (> LoD) in 39.4% (428/1086) of all subjects (18.0% female, 60.8% male). The female preponderance of undetectable hs-TnT was confirmed (98.5%) in a further cohort of young women (n=260) aged 20-39.
There is a step-wise increase in hs-TnT values with age. Nonparametric analysis of variance shows differences in hs-TnT between the various age groups in both men and women (p < 0.000001, Kruskal-Wallis test). These differences remain significant (p < 0.05) even when multiple pair-wise comparisons were made between the different agegroups (Kruskal-Wallis post hoc analysis by the method of Conover) other than that between 2 age groups -females 60-65 y and males 40-49 y ( Table 2).

Discussion
Prior to its adoption at Changi General in 2011, we verified the performance of this hs-TnT assay and found the overall 99P URL to 52.2±7.07 y) were enrolled. Pregnant subjects and those with any acute illness or co-morbidities (diabetes, hypertension, heart, muscle or renal disease) were excluded.

hs-TnT analyses
Details of the Roche hs-TnT assay have been published [13,14] and will not be provided here. Blood was collected in serum vacutainer tubes with gel and clot activator (Becton-Dickinson); visibly hemolyzed or turbid samples were omitted. All samples were centrifuged at 3000g for 10 minutes and analyzed within an hour on the Cobas e601 analyzer (Roche). The assay limit of blank (LoB) and assay limit of detection (LoD) were verified as per CLSI document EP17-A [15]. We evaluated assay precision according to CLSI EP5-A2 [16] with imprecision profiling as per Sadler [17]. Between-day assay precision was assessed over 30 days using the 2 concentrations of control materials supplied which contained 20 and 2000 ng/L of hs-TnT. Between reagent lot comparisons was done according to CLSI guide EP26-A [18]. Measurable hs-TnT was considered to be any value above the LoD; values below the LoD were imputed at half the LoD. Consecutive samples that had been stored at 4 o C were re-tested within 24 hours to verify the stability of hs-TnT samples should an add test be required. Hs-TnT was analyzed in consecutive paired EDTA plasma-serum samples to assess suitability of plasma as an alternative test material. Statistical analyses were performed using MedCalc 15.0 software (Ostend, Belgium). The 99P URL was determined according to CLSI guide C28-A3 using 1-tailed non-parametric statistics [19].

Reference population
In our current study, serum hs-TnT concentrations (range 2.5-34.0 ng/L) were positively skewed (Kolmogorov-Smirnov test: p < 0.0001) (Figure 1). The median hs-TnT concentrations by age and gender are given in Table 1.
The overall, male and female 99P URL (90% CI) were 17.0 (14-19), 18.6 (17-25) and 12.0 ng/L (11)(12)(13)(14)(15)(16)(17) respectively. The hs-TnT assay CVs   be 15 ng/L [20]. The assay LoB and LoD is similar to that in previous publications [13,14]. The analytical precision of the assay and its measuring range of up to 10,000 ng/L are quite acceptable. Serum samples re-analyzed within 24 hours show stability when stored at 4 o C permitting us to allow doctors to request adding an hs-TnT test from a previously drawn blood sample. In addition, the close agreement between paired EDTA plasma and serum hs-TnT enables us to use EDTA plasma as an alternative sample type.
All the current literature on TnT has reported gender differences. In a 2012 study Apple [21] reported an overall 99P URL of 15 ng/L for 524 subjects (age 18-64) without biomarker pre-qualification. This value is akin to what we found in 2010 and close to that in our current study (17 ng/L); our 99P decision limit for men and women (18.6 and 12.0 ng/L) was quite similar to that quoted by Apple -20 and 13 ng/L. In one of the largest hs-TnT reference range investigations (n=1157) involving 565 US and 592 Vietnamese volunteers (mean age 41.2 y) screened by questionnaire, a higher Asian male-female hs-TnT cutpoints (25 and 19 ng/L) was found [22]. Giannitsis [13] reported an overall 99P cut-point of 13.5 ng/L for 616 Germans with a mean age of 44±13.4 y while that in men was 14.5 and women 10.0. In a multicenter trans-Atlantic study (n=533 by questionnaire, mean age 37 y), Saenger [14] cited the following 99P URL of 14 (overall), 15.5 (male) and 9.0 (female) ng/L that was quite close to the publication from Germany. The Saenger study formed the data set cited in the hs-TnT product insert. However, this hs-TnT reference limits differ from that quoted in the new hs-TnT product insert (Elecsys Troponin T Gen 5 Stat package insert, March 2017) recently launched in the US. The 99P URL were higher in 1301 American subjects aged 21-89 y -overall 19, men 22 and women 14 ng/L respectively.
We also noted age-related differences in serum hs-TnT in both sexes, but more pronounced in men. Troponin elevations with age, contributed in part by age-related co-morbidities, have been emphasized [23][24][25]. Sandoval [9] and Collinson [26] underscore the vital role of population selection and pre-screening on the 99P decision limit. However, the data supporting the need to pre-qualify reference subjects with more tests (biochemical, imaging and ECG) is far from clear. In the Collinson study, questionnaires are a vital first step in volunteer recruitment. Thereafter the influence of additional testing (BP, ECG, NTproBNP, echocardiography, and estimated GFR) on the 99P URL for hs-TnT versus questionnaire alone is underwhelming -21.5-23.8 versus 23.2 ng/L in men and 12.8-13.9 versus 13.6 in women. Besides the study is underpowered with less than 340 subjects. Thus, laboratories will need to verify the transferability of the 99 percentile URL to their intended populations. When they establish their own reference limits, institutions need to carefully choose whether additional tests are needed for the reference population recruits. Moreover, as noted previously [27] almost all younger women (20-39) have undetectable hs-TnT. Thus, guidance is needed from learned societies and experts regarding the advisability of including them as part of reference population studies. Besides, this group also has a low prevalence of acute coronary syndrome.
The current reference range study comprises a sufficiently large number of subjects (>500 per gender) encompassing equal numbers of men and women evenly distributed in their age groups. This should provide a solid basis for hsTnT cut-points in clinical decision-making. This study is unique in that it is the first and only large-scale hs-TnT reference range study in an Asian population. The sample size is adequately powered and secure sex-dependent reference limits for serum hs-TnT has been established. Our study population comprises mainly local residents (Chinese, Malay, and Indian origin) and some foreigners (from the Philippines, China, Malaysia, India, and Indonesia). Our data may not be applicable to other ethnic groups in Asia and studies in these populations are needed. Reference to recent literature in this regard [10] will be helpful.
We acknowledge some limitations in our study. Participants were not pre-screened with electrocardiograms, cardiac imaging or other serum biomarkers such as creatinine or NT-proBNP as they are costly. Subjects > 65 y were not included as it is difficult to recruit subjects without medical co-morbidities. While applying lower hs-TnT cut-offs in subjects over 65 y may misclassify risk, this may be mitigated by serial troponin testing [7]. We have not assessed the diagnostic efficiency of this hs-TnT assay in clinical situations such as chest pain, chronic kidney disease, diabetes or community screening. However, this assay has been widely studied with favorable performance in several clinical studies -diagnosis of acute myocardial infarction [28], early rule-in rule-out strategies for myocardial infarction [29], myocardial infarction after percutaneous coronary intervention [30], and heart failure [31].
In conclusion, the Roche hs-TnT assay exhibits hs performance (supporting a 99 percentile URL with a CV < 10% and > 50% detectable values especially in a large number of healthy male subjects). It qualifies for the proposed high-sensitivity troponin designation in the current literature. Gender differences in hs-TnT values are seen. Thus sex-specific URLs should be used just as is done for CK, CK-MB, natriuretic peptides, uric acid, creatinine, and blood cell counts. Our findings allow us to report a robust overall and gender-based reference ranges for hs-TnT in our multi-ethnic Asian cohort. The multi-center collaboration in this effort strengthens its wider utility nationally since it covers a broader spectrum of laboratories and patients. Age-related differences in hs-TnT were also found. Laboratories need to verify or establish their own hs-TnT reference values. Reference interval studies should be sufficiently large to accurately reflect the effects of gender and age. The impact of using gender-or age-specific cTn limits for the management of ACS is emerging and merits further studies. The use of group-specific URLs can only improve the diagnostic specificity of hs-TnT in the evaluation of ACS.