Determination of high‐sensitivity cardiac troponin T upper reference limits under the improved selection criteria in a Chinese population

Abstract Background There is no common consensus on how to define the reference population for determination of high‐sensitivity cardiac troponin (hs‐cTn) upper reference limit (URL). This study aimed to establish 99th percentile URLs of hs‐cTnT under both 2018 AACC/IFCC criteria and improved selection criteria for further judging whether two URLs are different. Methods Applying the stratified cluster sampling protocol, this study took 1848 apparently healthy subjects in communities of Shenyang China as the screening objects. We first followed 2018 AACC/IFCC criteria using surrogate biomarker for diabetes, myocardial dysfunction, renal dysfunction, and electrocardiogram. Then, we followed improved selection criteria to exclude hypertension, overweight and obesity, and dyslipidemia by physical examination and laboratory screening. Accordingly, 99th percentile URLs of hs‐cTnT were established. Results If the 2018 AACC/IFCC criteria were applied, 99th percentile URLs (90% confidence interval) of hs‐cTnT male, female, and total were 19 (17‐20) ng/L, 16 (15‐17) ng/L, and 18 (16‐19) ng/L, respectively. If added a single supplementary selection criteria, 99th percentile URLs of hs‐cTnT total reduced to 16 ng/L, 17 ng/L, and 16 ng/L, respectively. If the improved selection criteria were applied, 99th percentile URLs (90% confidence interval) of hs‐cTnT male, female, and total were 18 (14‐24) ng/L, 13 (11‐16) ng/L, and 16 (13‐17) ng/L, respectively. The 99th percentile URLs of hs‐cTnT male were higher than those of female in every age group. Conclusions Improved selection criteria through questionnaire survey, physical examination, and laboratory screening to further exclude hypertension, overweight and obesity, and dyslipidemia can avoid overestimation of the 99th percentile URL of hs‐cTnT.


| INTRODUC TI ON
In light of the fourth universal definition of myocardial infarction (MI) announced in 2018 and 2015 European Society of Cardiology (ESC) guidelines for the management of acute coronary syndromes (ACS) in patients presenting without persistent ST-segment elevation, cardiac troponin (cTn), especially high-sensitivity cardiac troponin (hs-cTn), is the preferred biomarker for the diagnosis of MI. 1,2 It is important to establish 99th percentile upper reference limit (URL) of cTn, which is widely acknowledged in the universal definitions of myocardial injury and different types of MI. 1 For myocardial injury, cTn should elevate at least one value above 99th percentile URL. For types 1, 2 and 3 MI, cTn should rise and/or fall at least one value above 99th percentile URL. For type 4a MI (percutaneous coronary intervention-related MI) and type 5 MI (coronary artery bypass grafting-related MI), cTn should elevate more than 5 times and 10 times of 99th percentile URL, respectively.
However, up to now there is no consensus on how to define reference population. The CLSI document Reference Intervals in the Clinical Laboratory (EP28-A3c) 3 recommended a questionnaire-based approach for determining laboratory test reference intervals. The present 99th percentile URL of hs-cTnT (14 ng/L) was established based on 616 "apparently healthy" volunteers and blood donors, but little information was reported about population selection, 4 and then, it was confirmed by Saenger et al 5 based on 533 individuals with a standardized questionnaire from the United States and Europe. But Collinson et al, 6 Koerbin et al, 7  The chief objective of screening was to exclude cardiovascular disease and related disease, especially the ones which might influence hs-cTn values. There was still a possibility that 2018 AACC/ IFCC criteria, through questionnaires, surrogate biomarkers, and imaging, might omit these important cardiovascular risk factors, 13 such as hypertension, overweight and obesity, and dyslipidemia.
Without blood pressure measurements, hypertension was likely to be missed only through questionnaire. For example, in China, the prevalence and awareness of hypertension were 37.2% and 36.0%, respectively, 14 and the awareness rate still exited in the United States 15 and other countries. [16][17][18] For obesity and overweight, it was reported that it affected about two-thirds of the Americans, 19 and previous study showed that hs-cTnI was dependent on BMI in their multiple linear regression model. 20 24 ). According to this procedure, 932 individuals enrolled in our study under improved selection criteria. Figure 1 summarized specific screening process and the exact exclusion numbers during each step.

| Laboratory analysis
All subjects maintained normal working and dietary habits 1 week before blood collection, avoided strenuous exercise and physical labor, avoided heavy drinking, prohibited drinking 1 day before blood collection, and avoided overeating at night and high-fat and high-protein diet; fasting time was 8 hours to 14 hours after dinner (including drinks, milk, fruits, etc), and water consumption was not more than 200 mL before blood collection. Smoking and vigorous exercise are prohibited for one hour. Blood samples were collected from the elbow vein and allocated to 5 mL SST tube with gel (Becton Dickinson, USA). All blood samples were centrifuged with 1200 g for 10 minutes within 2 hours after collection, and serum was analyzed within 2 hours after separation.
We measured concentrations of FBG, TG, TC, HDL-C, LDL-C, and creatinine (Cr) by using Roche Modular automatic biochemical analyzer (Roche Diagnostics). We measured concentrations of NT-proBNP and hs-cTnT on a Cobas E170 Analyzer (Roche Diagnostics).

| Statistical analysis
The statistical analysis of this study was performed using SPSS 22.0.
The Mann-Whitney test was used for comparison between two subgroups. Outliers were identified in the total population and in male and female separately using the methods of Dixon 3 and Reed. 25 After excluding the outliers, the 99th percentile URLs and 90% confidence interval were calculated using nonparametric method. 26

| Distribution of hs-cTnT values by age and gender under improved selection criteria
Distribution of hs-cTnT values by age and gender under improved selection criteria is presented in Table 3. The 25th percentile, median, and 75th percentile of hs-cTnT male were higher than those of female in every age group (P = .004 in ≥60 years old and P < .001 in other age groups). The 99th percentile URLs of hs-cTnT male were higher than those of female in every age group. The 99th percentile URLs of hs-cTnT increased with age in both genders except for 18-29 years old and 30-to 39-year-old men. The difference in the 99th percentile URLs of hs-cTnT between women aged 50-59 and over 60 was very small (18 ng/L vs 19 ng/L).

| D ISCUSS I ON
Being a single-center study, our research programs are more representative such as stratified cluster sampling protocol but not volunteers nor blood donors, which avoided selection bias. Gender is an important factor affecting the 99th percentile URLs; therefore, when using hs-cTn, gender-specific 99th percentile URLs are recommended. 10,11 Together with other two studies of 99th percentile URLs of hs-cTnT for Italian 27 and Dutch 28 under 2018 AACC/ IFCC criteria, we all found 99th percentile URL of hs-cTnT male was higher than that of female. This attributed to intrinsic differences between genders that the concentration of cTn was correlated with left ventricular mass, which was higher in men than in women. [30][31][32] The physiological difference between genders was one thing but whether or to what extent it would affect clinical management and improve outcomes still needs further studies. There were two studies using gender-specific URLs of hs-cTnT (15.5 ng/L for men and 9 ng/L for women), but reclassification has little effect on diagnosis and prognosis. Mueller-Hennessen et al 33  This study has several limitations. Firstly, participants of male were 345 (37.0%) and participants more than 60 years old were less than other age group. But we still included more than 300 individuals of each gender, which met the requirements of 2018 AACC/IFCC criteria. 11 For the age characteristics of the elderly, exclusion of comorbidities was a hard task. The stricter the screening criteria were, the fewer individuals would be included. We used stricter exclusion criteria than previous studies; in consequence, the number of elderly was rare. Secondly, we used FBG instead of HbA1c as the surrogate biomarker for diabetes. Our strategy was the same as Franzini et al 27 that diabetes was defined as FBG more than 7 mmol/L or the use of any hypoglycemic agent. Thirdly except for ECG, we did not perform echocardiogram or MRI. Although ECG is definitely not a substitute for echocardiogram or MRI in diagnosis of CVD, we think ECG together with NT-proBNP can somehow play an alternative role in determination of hs-cTn 99th percentile URL. 28,36 Meanwhile, hs-cTnT was associated with some diseases such as atrial fibrillation, 45

CO N FLI C T O F I NTE R E S T
There are no conflicts of interest.