Soluble p-selectin should be measured in citrated plasma, not in serum.

Abstract

We read with interest the recent paper (Ritchie et al, 2002) regarding soluble P-selectin (CD62P) measurement in serum as a marker of platelet activation after nitric oxide inhibition. We wish to highlight the fact that, worldwide, virtually all research groups measure soluble P-selectin in citrated plasma and not in serum. Until recently, the origin of circulating, soluble P-selectin in the blood was in doubt. Many groups have supported our original hypothesis (Blann & Lip, 1997) that the great proportion, if not all, of soluble P-selectin arises from platelets, not the endothelium. Confusion may have arisen from reports that P-selectin is a component of the membrane of the alpha and dense granules of the platelet, and of the membrane of the Weibel–Palade body of the endothelium cell. P-selectin is not a component of the matrix of either type of organelle (unlike, for example, beta thromboglobulin and von Willebrand factor respectively). Evidence for this comes from many sources. For example, the endothelial cell stimulant DDAVP (desmopressin) induces raised levels of von Willebrand factor but not soluble P-selectin in vivo (Jilma et al, 1996). Furthermore, we have been unable to detect P-selectin, but plentiful von Willebrand factor, by enzyme-linked immunosorbent assay (ELISA), in human umbilical vein endothelial cell lysates or tissue culture supernatants (data not shown), whereas P-selectin is easily detected in platelet lysates and from the ‘supernatants’ of aged endothelial-cell-free platelet concentrates designed for transfusion (Kostelijk et al, 1996). However, more pertinent data arise from the measurement of soluble P-selectin in paired serum and plasma. The great disparity in serum/plasma P-selectin data can be highlighted when comparing similar studies. Kirk et al (1997), using a different ELISA kit to that of Ritchie et al (2002), reported a mean soluble P-selectin concentration in citrated plasma from 20 normal control subjects of 140 ng/ml, with mean levels in serum of 616 ng/ml. Ritchie et al (2002) themselves reported a median serum P-selectin concentration from 12 healthy volunteers of 128 ng/ml. Other much larger studies have reported mean serum P-selectin levels well in excess of 100 ng/ml (Malik et al, 2001). Using the same kit as Ritchie et al (2002), we measured the soluble P-selectin concentration in paired samples of serum and citrated plasma from 18 patients (mean age 60 years, 13 men) with a variety of cardiovascular diseases (e.g. acute myocardial infarction, stenotic coronary arteries, chronic heart failure) and in 18 age- and sex-matched healthy control subjects. The data (Table I) indicated that serum P-selectin was significantly higher in serum than in plasma in both the control subjects (by 9%) and in the patients (by 39%). However, the difference between the control subjects and the patients was significant only when measured in the citrated plasma sample, not in the serum sample. If soluble P-selectin does indeed arise from platelets, then we may predict that the products of activated platelets contain more P-selectin. Thus we suggest that the mechanism for increased P-selectin in serum is simply due to excess P-selectin released from platelets activated, or even ‘exhausted’, by the process of coagulation and clot formation. We conclude that the measurement of serum soluble P-selectin may not necessarily be a true reflection of physiological levels in vivo. If this is the case, measurement in serum may give an artefactually incorrect result, as we have found in our current data. This may also explain why a recent large study of serum P-selectin (Malik et al, 2001) failed to support other data reporting that raised plasma P-selectin predicts adverse cardiovascular events (Ridker et al, 2001; Hillis et al, 2002). We urge workers to measure this molecule in citrated plasma.