A Short Thrombin Generation Time-to-Peak/Lag Time Ratio Is an Independent Risk Factor for Mortality in the General Population

Background Thrombin generation (TG) is used to detect small changes in coagulation indicating a bleeding or prothrombotic phenotype. Until now, large prospective studies are missing to study the association of TG parameters with the occurrence of future cardiovascular events and mortality. In this study, we investigated the prospective value of TG measured in baseline plasma samples of a large cohort (n=20,991) with overall mortality (1,098; 5.3%) with a median follow-up of 8.2 years.

Methods Thrombin generation was measured by calibrated automated thrombinography (CAT) using PPP Reagent low. The lag time (LT), endogenous thrombin potential (ETP), peak height, time-to-peak (TTP) and velocity index (VI) were quantified. Subjects on anticoagulant therapy were excluded from analysis, because this affects the measurement of TG. Cox regression was used to analyze differences in survival in subjects according to their variations in the TG parameters. The reported results are obtained from a crude model considering only single TG parameters standardized for their standard deviation, and from a final multivariable model further adjusted for sex, age, BMI, smoking, contraceptive use, antiplatelet medication and medical history (hypertension, hypercholesterolemia, diabetes, history of cardiovascular diseases or cancer).

Results The TG lag time at 1 pM tissue factor was 4.1 min ± 1.0 on average, and the peak height was 363 nM ± 89. The time-to-peak was 6.6 min ± 1.6 on average, the velocity index was 163 nM/min ± 63 and the endogenous thrombin potential was 1717 nM·min ± 415. The SD-standardized lag time was significantly associated with mortality risk in the crude model (HR _LT=1.09; CI:1.03-1.16; p=0.002) and after adjustment for all covariates, including medical history (HR _LT=1.09; CI: 1.03-1.15; p=0.002). In contrast, a short time-to-peak was significantly associated with reduced mortality risk (HR _TTP=0.92; CI: 0.86-0.98; p=0.012). As the opposite associations of LT and TTP with mortality could have a synergistic effect on the mortality risk, we combined the variables into the novel TTP/LT-ratio. The mean TTP/LT ratio was 1.61 ± 0.14. The hazard ratio for mortality decreased by 40% for each standard deviation increment of the TTP/LT-ratio in the crude model (HR _ratio=0.606; CI:0.562-0.654; p<0.001), and by 18% in the fully adjusted model (HR _ratio=0.824; CI: 0.766-0.887; p<0.001).

The TTP/LT-ratio variable was divided in quintiles for further analysis: Quintile 1 included subjects up to a TTP/LT-ratio of 1.50, quintile 2 ranged from 1.50 to 1.58, quintile 3 ranged from 1.58 to 1.62, quintile 4 ranged from 1.62 to 1.70, and quintile 5 consisted of subjects with a TTP/LT-ratio above 1.70. Subjects in TTP/LT-ratio quintile 1 had an increased mortality risk (HR _{ratio_Q1vsQ5} =2.96; CI: 2.42-3.63; p<0.001), even after full adjusting (HR _{ratio_Q1vsQ5} =1.46; CI: 1.18-1.79; p<0.001; Figures 1 and 2). Moreover, subjects in TTP/LT-ratio quintile 2, have a similar statistically significant, although less pronounced increase risk for mortality in the crude model (HR _{ratio_Q1vsQ5} =2.17; CI: 1.18-2.68; p<0.001) and the fully adjusted model (HR _{ratio_T2vsT5} =1.29; CI: 1.04-1.60; p=0.019; Figure 1).

Conclusion A prolonged TG lag time and a shortened time-to-peak in the general population and within the normal ranges are risk factors for mortality. The combination of a long LT and a short TTP results in a low TTP/LT-ratio, which was identified as an independent risk factor for increased mortality risk in the general population.