Sources of variability in measurements of cardiac troponin T in a community-based sample: the atherosclerosis risk in communities study

Abstract

Background: Application of cardiac troponin T (cTnT) as a marker of myocyte damage requires knowledge of its measurement variability. Using a highly sensitive assay for measurement, we evaluated the long-term storage stability of plasma cTnT at -70 °C and the sources of cTnT variability.

Methods: Samples from the Atherosclerosis Risk in Communities study collected in 1996-1998 and 2005-2006 were assayed centrally to quantify variability in cTnT attributable to processing (replicates from same blood draw, n = 87), laboratory (replicates after freeze thaw, n = 29), short-term (n = 40) and long-term biological variation (repeat visit, n = 38), and degradation in frozen storage (n = 7677).

Results: Approximately 30% of this population-based cohort had cTnT concentrations below the detection limit (3 ng/L). Reliability coefficients for all paired comparisons exceeded 0.93 except for samples drawn 8 years apart (r = 0.36). Sources of cTnT variation (as CVs) were: laboratory, 2.1% and 11.2% in those with and without heart failure, respectively; processing, 18.3%; biological, 16.6% at 6 weeks and 48.4% at 8 years. The reference change value at 6 weeks (68.5%) indicated that 4 samples are needed to determine a homeostatic set point within ±25%. The estimated cTnT degradation rate over the first year in long-term frozen storage was 0.36 ng/L per year.

Conclusions: cTnT was detectable in approximately 70% of community-dwelling middle-aged study participants and stable in -70 °C storage. The variability in cTnT attributable to 1 freeze-thaw cycle is of small magnitude. The observed high laboratory and intraindividual (biological) reliability of cTnT support its use for population-based research, and in clinical settings that rely on classification and serial measurements.

Figure 1

Bland-Altman plot showing bias against the mean of cTnT measurements obtained six weeks and eight years apart with 95% levels of agreement (broken lines). The differences and means are expressed in microgram/L.

Figure 2

Influence of time since collection on cTnT concentration in n=7,677 samples collected over 1,093 days in 1996–98 and stored at −70 °C till measurement in 2009–2010 i.e. freezer time varying from 11 to 14 years.