Pilot observational study of rotational thromboelastometry (ROTEM) to predict bleeding risk in patients with hematological malignancies (ATHENA study)
Estcourt LJ., Stanworth SJ., Mumford AD., Harrison P., Powter G., Harding K., Dyer C., Howgate S., Murphy MF.
Background: Previous studies have shown that the platelet count is a poor predictor of bleeding in severely thrombocytopenic patients with hematological malignancies. Rotational thromboelastometry (ROTEM) detects thrombocytopenia but is also sensitive to functional alterations of platelets and soluble coagulation proteins which could contribute to bleeding in severe thrombocytopenia. Aims: The aim of this study was to evaluate whether ROTEM parameters of extrinsically activated coagulation [EXTEM (clotting time (CT), clot formation time (CFT), maximum clot firmness (MCF), maximum lysis (ML))], intrinsically activated coagulation [INTEM (CT, CFT, MCF, ML] and the fibrinogen contribution to coagulation [FIBTEM (test based on EXTEM but contains cytochalasin D to inhibit platelets) (MCF, ML)] were better at predicting bleeding than the platelet count. Methods: Prospective cohort study of adults with a hematological disorder undergoing intensive chemotherapy or stem cell transplant (ISRCTN81226121) at two UK centers (Bristol and Oxford) between September 2010 and September 2012. The study inclusion criterion was development of thrombocytopenia (platelet count ≤50x109/L). From the start of thrombocytopenia, the participants underwent daily formalized bleeding assessments until platelet count recovery, hospital discharge, death, or for up to 30 days. Bleeding was graded by the WHO score. Alongside routine tests, additional venous blood samples were collected into EDTA and citrate collection tubes three times a week for detailed analysis of platelet and ROTEM parameters. The data were analyzed using a generalized linear model with binomial family and logit link function clustered on patient identity to account for repeated measures. Results:All 50 participants were followed up until study completion. The baseline characteristics were: mean age 51.0 years; male (33/50); diagnosis (leukemia 16/50; lymphoma 14/50; myeloma 9/50; other 11/50); treatment (chemotherapy 4/50; allograft 33/50; autograft 13/50). Bleeding symptom data were available for 99.7% of study days. The participants had a median 3 days of bleeding (any severity) (interquartile range (IQR) 0-6); and median 11 days (IQR 8-16) with platelet count ≤50x109/L. 30.6% of patients had at least one episode of WHO grade 2 or above bleeding during the study period. The unadjusted odds ratio (OR) for bleeding the following day when the platelet count was ≤50x109/L was 0.98 (95% confidence interval (CI) 0.97-1.00; P=0.03). The unadjusted OR for bleeding based on the ROTEM parameters (EXTEM MCF and INTEM MCF) were close to statistical significance. However, this effect was lost when the results were adjusted for the platelet count (Table 1). The unadjusted OR for WHO grade 2 or above bleeding the following day when the platelet count was ≤50x109/L was 0.98 (95% CI 0.96 to 1.00; P=0.055). The unadjusted OR for bleeding based on the ROTEM parameters (EXTEM MCF and INTEM MCF) were statistically significant but this effect was lost when the results were adjusted for the platelet count (Table 1). The unadjusted OR for any bleeding or WHO grade 2 or above bleeding based on FIBTEM MCF were not significant. Summary / Conclusion: This preliminary study does not suggest that ROTEM has any greater predictive value than the platelet count for any bleeding or WHO grade 2 or above bleeding. The statistically significant OR for some of the ROTEM parameters are likely to be explained by a direct effect of platelet count on the ROTEM assay endpoints.