Evaluation of haemolysis in emergency department samples requesting high sensitivity troponin T measurement

There have been several recent reports in Annals of Clinical Biochemistry concerning a new and highly sensitive troponin T (hsTnT) assay. One study compared the hsTnT assay to conventional troponin assays for the early detection of myocardial infarction in an emergency department. ¹ This study demonstrated better performance of the hsTnT assay compared with the fourth-generation TnT assay and that diagnostic accuracy was improved when a follow-up sample was tested. In a second report, analytical evaluation of the hsTnT assay showed that haemolysis impacts the assay results with haemoglobin concentrations above 0.25 g/dL, reducing troponin T concentrations by 20%. ² These results confirmed previous studies by Bais ³ and Florkowski et al., ⁴ who showed that negative interference could be as high as 60% in heavily haemolysed samples. Moreover, the assay manufacturer (Roche Diagnostics, Mannheim, Germany) claims that haemoglobin concentrations above 0.1 g/dL may cause interference and loss of assay precision.

Having used the hsTnT assay for several months in our laboratory, we sought to determine the extent to which results might be affected by haemolysis in our emergency department. To accomplish this, we evaluated the frequency of haemolysed blood samples in our emergency department and classified them according to their haemolysis index (HI; measured by spectrophotometry), which corresponds to 0.05, 0.1, 0.25, 0.5 and 1 g/dL of haemoglobin. We measured hsTnT using electrochemiluminescence on Cobas 8000 (e602) analysers and systematically incorporated the HI into the assay result. The HI correlated well with haemoglobin concentration (Figure 1). During a three-month period, hsTnT was measured in approximately 2500 samples from the emergency department. Haemolysed samples were frequently encountered, with more than 15% of samples having substantial haemolysis (HI above 50, the level at which the plasma potassium concentration is significantly affected). A similarly high frequency has been also reported by Berg et al., ⁵ who found a frequency of haemolysed samples of 10.7% with an index cut-off above 70. Using this cut-off, we observed the same frequency (10.9%). According to the hsTnT manufacturer's recommendations, haemolysis would impact up to 7% of our emergency department samples. However, this more conservative haemoglobin concentration limit of 0.1 g/dL causes troponin T concentrations to decrease by only 5–10%. Using a haemoglobin limit of 0.25 g/dL taken from the results of Chenevier-Gobeaux et al, ² haemolysis would have impacted troponin T concentrations in 2.6% of the samples. Haemolysis is an important parameter to consider when measuring troponin T with the hsTnT assay. We recommend that HI be systematically incorporated into troponin T measurement protocols and that hsTnT results be invalidated if the HI is above 220 (i.e. haemoglobin >0.25 g/dL). These higher haemoglobin concentrations cause troponin T concentrations to be underestimated by at least 20%; however, such levels of haemolysis were found in less than 3% of the samples in our emergency department.

Figure 1

Sample frequency according to haemolysis index

DECLARATIONS

Competing interest: NP and BC received consulting fees from Roche Diagnostics France.

Funding: None.

Ethical approval: Not required.

Guarantor: BC.

Contributorship: All authors were involved in the data analysis and contributed to both draft and final manuscripts.

Acknowledgements: None.