Bimodal distribution of soluble endothelial protein C receptor levels in healthy populations
Article first published online: 21 MAR 2003
Journal of Thrombosis and Haemostasis
Volume 1, Issue 4, pages 855–856, April 2003
How to Cite
Stearns-Kurosawa, D. J., Burgin, C., Parker, D., Comp, P. and Kurosawa, S. (2003), Bimodal distribution of soluble endothelial protein C receptor levels in healthy populations. Journal of Thrombosis and Haemostasis, 1: 855–856. doi: 10.1046/j.1538-7836.2003.t01-4-00115.x
- Issue published online: 21 MAR 2003
- Article first published online: 21 MAR 2003
- Received 8 November 2002, accepted 8 November 2002
Procoagulant and inflammatory processes that give rise to thrombin production contribute to development of coronary artery disease, stroke and other thrombotic disorders. The endothelial protein C receptor downmodulates these processes by augmenting protein C activation on endothelial cells [1,2]. A soluble, ligand-binding form of the endothelial protein C receptor (sEPCR) circulates in plasma and sEPCR levels increase significantly in patients with systemic inflammatory diseases [3,4]. In vitro, sEPCR is released from the endothelium due to events triggered by thrombin [5,6]. This potential mechanism and our observation that sEPCR levels and F1 + 2 levels decline in patients and healthy volunteers upon warfarin treatment  led us to re-examine the distribution of sEPCR levels in two apparently healthy adult populations. Citrated plasma samples were obtained from a French health care check-up center (IPC, Paris, France; n= 200; 100 male, 100 female) and from healthy Italian adult volunteers (n = 49; 20 male, 29 female; contributed by Dr Armando D'Angelo, Scientific Institute HS, Raffaele, Milano, Italy). All samples were obtained with informed consent and stored at −70 °C until assay. sEPCR levels were analyzed by the same individual using a prototype ELISA kit  and the same lot of kit. All samples were randomly selected and were from apparently healthy donors with no known medication. For analysis of sEPCR distribution, samples were grouped by gender and nationality (Italian males, Italian females, French males and French females) and tested for normality using the Shapiro–Wilk test. Since all showed significant departure from normality, the Kruskal–Wallis test was used to examine differences among the four populations. Since histograms suggested subpopulations within each group, each group was partitioned into low and high sEPCR populations by first ordering all values from smallest to largest. Next, the point where the two populations were apparently separated by at least 10 U was used as an initial partition point. If the resulting lower population contained a high outlier (as defined by box plot criterion), the partition point was lowered by one value until no outlier was observed. The resultant populations were tested for normality. The four lower populations were deemed suitable for a 2 × 2 (gender by nationality) analysis of variance. The four higher populations were deemed not suitable for parametric analysis and were examined using the Kruskal–Wallis test, a non-parametric analog of analysis of variance.
Analyses of plasma sEPCR levels in the four groups (males and females from Italy or France) demonstrated that each was significantly different from a normal distribution with a bimodal distribution (Fig. 1). The Kruskal–Wallis test indicated a significant difference (P < 0.0001) among the four groups. The partitioning algorithm yielded different cut points for high/low division for each group, but surprisingly similar proportions in the high sEPCR groups. For the French population, the cut point and proportion higher was 155 ng mL−1 and 22% for females, 176 ng mL−1 and 25% for males. For the Italian population, the cut point and proportion higher was 190 ng mL−1 and 24% for females, 200 ng mL−1 and 25% for males. Males had higher sEPCR levels and this gender difference was similar regardless of the country of origin.
Together with the observations that thrombin increases sEPCR production in vitro and anticoagulant therapy reduces sEPCR levels in vivo, it is our hypothesis that the high sEPCR levels in about 25% of the population reflect thrombin production and a possible hypercoagulable state. In this regard, it is interesting that thrombin–antithrombin (TAT) levels in a healthy adult population from northern Italy also show a skewed distribution and a subpopulation with high values . While we are not aware of differences in rates of thrombotic events between France and Italy, there are geographic differences in rates of coronary events . Interpretation of plasma sEPCR levels with respect to these differences must await additional studies addressing the physiological basis for the bimodal distribution of sEPCR values. A genetic basis is one possibility, but lifestyle (e.g. smoking) or undiagnosed pathological factors may also contribute because sEPCR levels increase in patients with systemic inflammatory diseases. Prospective studies in high risk patient groups will address this possibility. These preliminary data suggest that definition of a normal population for determination of plasma sEPCR values should take into consideration gender and geographic location. Additional studies are necessary to evaluate the influence of other variables, such as smoking or age, on the bimodal distribution of sEPCR levels, as well as the relationship of sEPCR levels with other hemostatic markers.