- Top of page
- Design and methods
The plasma metalloprotease ADAMTS13 (A Disintegrin And Metalloprotease with ThromboSpondin type 1 motif 13) cleaves prothrombotic ultralarge multimers of the platelet-adhesive protein von Willebrand factor (ULVWF) into less active multimers that promote haemostasis in injured blood vessels. When the enzyme is dysfunctional or undetectable, circulating ULVWF may cause massive intravascular aggregation of platelets and thrombotic thrombocytopenic purpura. This study compared ADAMTS13 antigen and activity in a large set of plasmas collected from subjects with various conditions of health and disease, most of which were associated with an increased thrombotic tendency. Pathological conditions were liver cirrhosis (n = 90), inflammatory bowel disease (n = 44) and cardiac surgery (n = 30). Healthy conditions were pregnancy (n = 42), oral contraceptive intake (n = 33) and the neonatal state (n = 41). Normal individuals of different ages were taken as controls (n = 132). The antigen assay showed less variability than the collagen binding-based activity assay. Antigen values correlated well with activity in normal individuals, but were discrepant to various degrees in neonates, pregnancies of later maternal age and cardiac surgery. No discrepancies were noted in liver cirrhosis and inflammatory bowel disease, which were both associated with low-plasma levels of ADAMTS13. The parallel measurement of ADAMTS13 activity and antigen provides a new tool for understanding the behaviour of the VWF cleaving protease in health and disease.
Thrombotic thrombocytopenic purpura (TTP) is a rare but life-threatening disease involving at least three determinants; von Willebrand factor (VWF), blood platelets and A Disintegrin And Metalloprotease with ThromboSpondin type 1 motif 13 (ADAMTS13). VWF is a multimeric plasma glycoprotein that bridges exposed subendothelium to flowing platelets when a blood vessel is damaged. It is released from endothelial cells in an ultra large form (ULVWF), which distinguishes itself not only by molecular weight but also by the ability to aggregate platelets in conditions of high shear stress (Arya et al, 2002). Proteolysis of the VWF 1605Y-1606M peptide bond by ADAMTS13 produces the naturally occurring multimeric VWF ladder in plasma. When ADAMTS13 is absent or dysfunctional, TTP may occur because ULVWF can circulate freely in plasma and potentially cause intravascular platelet aggregation and disseminated microvascular thrombosis (Lammle et al, 2005). This mechanism fits with pathological observations of VWF-rich thrombi and with the observed presence of ULVWF in TTP plasma. However, a number of TTP patients have detectable and even normal levels of plasma ADAMTS13 activity in vitro (Veyradier et al, 2001) and so, whether or not severe ADAMTS13 deficiency is specific for TTP is still a matter of debate (Remuzzi, 2003; Tsai, 2003).
ADAMTS13 activity was previously measured in a large set of plasmas obtained from donors in well-defined physiological or pathological states, most of which were associated with an increased thrombotic tendency (Mannucci et al, 2001). Significantly reduced levels of ADAMTS13 activity were found in physiological states, such as pregnancy and the neonatal period, as well as in several pathological conditions, in spite of the absence of clinically overt thrombotic events. At that time no tools were available to measure the immunoreactive protease in these conditions. In this study, a monoclonal antibody-based ADAMTS13 antigen assay (Feys et al, 2006) was employed to compare the immunoreactive protein with the corresponding functional activity in a freshly collected set of samples comparable to those of the previous study (Mannucci et al, 2001).
- Top of page
- Design and methods
A large series of plasmas was previously investigated to explore the behaviour of VWF cleaving activity in physiological or pathological conditions (Mannucci et al, 2001). At that time no antibodies for the determination of immunoreactive protease were available. Measuring antigen levels and comparing them with activity levels should help to understand whether or not the protease is fully enzymatically active, a situation expressed by values of activity to antigen ratios equal or close to 1. ADAMTS13 antigen levels were previously measured in a large number of TTP patients and higher values than those of activity were often found, indicating that a fraction of ADAMTS13 is non-functional in these patients (Peyvandi et al, 2006).
The present study involved a new group of healthy individuals and patients in whom ADAMTS13 activity and ADAMTS13 antigen levels were measured in parallel. VWF, the protease's substrate, was also measured immunologically in the same samples. In general, the antigen assay appeared to be less variable than the activity assay, as indicated by smaller standard deviations and coefficients of variation. The limitations of the indirect functional assays of ADAMTS13, and particularly of those measuring residual collagen binding of the substrate, have been a matter of debate ever since their initial description (Mannucci, 2003). Even though the FRETS-VWF73 assay (Kokame et al, 2005) offered a more direct measurement of ADAMTS13 through the assessment of the cleavage of a peptide containing the VWF cleavage sequence, the requirements for this assay in terms of equipment are quite demanding for the majority of unspecialised haemostasis laboratories. Hence the antigen assay used in this study may offer a good alternative, with acceptable coefficients of variation, classical setup (sandwich ELISA) and easy access.
The main finding of this study is that plasma levels of ADAMTS13 activity do not always coincide with those of the immunoreactive protease, as exemplified by the newborn group. Two reports claimed that ADAMTS13 activity was low in neonates (Mannucci et al, 2001; Takahashi et al, 2001) but another found normal levels (Tsai et al, 2002). In our study, neonatal plasma contained less ADAMTS13 antigen than that of young healthy adults taken as controls (between 17 and 35 years of age). Moreover, ADAMTS13 activity levels were even lower than antigen levels, indicating that a substantial amount of protease moieties were not active in vitro. Whether the metalloprotease is inactivated or consumed after VWF proteolysis at birth or whether there is an inhibition of its enzymatic activity remains to be understood. Large VWF multimers have been shown to be present in some newborn infants (Schmugge et al, 2004). Yet, previous studies have shown that high VWF levels do not affect the ADAMTS13 activity measurements by residual collagen binding (Mannucci et al, 2001).
We chose to measure ADAMTS13 in pregnancy because approximately two-thirds of all TTP cases occur in females, of which 12–25% occur during pregnancy (George, 2003). In agreement with our previous observations (Mannucci et al, 2001), Sanchez-Luceros et al (2004) reported a mean ADAMTS13 activity of 61·5% ± 14·5% (n = 71) in the late period of pregnancy. In our study, mean ADAMTS13 activity was only slightly reduced, with somewhat higher antigen values, particularly in a subset of older women aged above 35 years. Women taking oral contraceptives had normal and concordant levels of both ADAMTS13 activity and antigen, indicating that the hormonal changes induced by these drugs do not change ADAMTS13, at variance with pregnancy. Previous studies have shown that the high VWF:Ag levels, as observed in pregnancy, do not significantly alter the ADAMTS13 activity results when determined by residual collagen binding (Mannucci et al, 2001).
ADAMTS13 mRNA has been detected in many organs, but the stellate cell of the liver is a major site of production (Levy et al, 2001; Uemura et al, 2005). A model for injured liver in the rat indicated no significantly altered levels of ADAMTS13 in rat plasma, despite a substantial increase of protein transcription in the injured organ (Niiya et al, 2006). On the other hand, a very recent study performed in rats showed that, while the inhibition of hepatocyte function did not affect ADAMTS13 levels, specific inhibition of stellate cell function led to a decrease in ADAMTS13 activity (Kume et al, 2007). Whether this is also the case in human cirrhotic livers is poorly understood. Lisman et al (2006) reported that levels of ADAMTS13 activity were not significantly altered in the three Child-Pugh subclasses of liver cirrhosis Their data were very much spread, with some patients having over 300% of normal ADAMTS13 levels and others presenting with extremely low values. The present study involved 90 cirrhotics and although an important degree of dispersion was confirmed, none of the patients had activity nor antigen levels above 200%. At variance with Lisman et al (2006) a significant decrease of both enzymatically active and immunoreactive ADAMTS13 was found in Child classes B and C, with a more marked drop in those patients with more severe disease. The conflicting data between the different studies may be attributed to the use of different antigen assays. Lisman et al (2006) employed a commercial kit based on polyclonal anti-ADAMTS13 antibodies. This assay resulted in a lower correlation between activity and antigen (R = 0·39) compared to our study (R = 0·80), perhaps because of the overall lower variability in both our activity and antigen measurements. It has been suggested that high plasma levels of VWF, as those found in liver cirrhosis (Table I), may influence the collagen-binding assay of ADAMTS13 activity and give spuriously low levels. Our previous in vitro studies showed that this phenomenon is not prominent, even at high endogenous VWF levels similar to those found in liver cirrhosis (Mannucci et al, 2001). The reliability of the collagen-binding assay to measure ADAMTS13 activity in liver cirrhosis, even in the presence of high VWF in plasma, was indirectly confirmed by the close concordance between the low values of ADAMTS13 activity and antigen in this study, because the ADAMTS13 antigen assay is not influenced by VWF levels. On the whole, these data establish unequivocally that ADAMTS13 is lowered in severe liver cirrhosis (albeit not in all patients), consistent with the recent studies in rats (Kume et al, 2007).
ADAMTS13 activity was previously shown to be drastically reduced in patients with inflammatory diseases characterised by CRP levels higher than 50 mg/l (Mannucci et al, 2001). The current study involved only one such patient. Nevertheless, when comparing patients with moderate inflammation (CRP >10 mg/l) and those with normal CRP, significantly lower values for both activity and antigen were observed in the former. As in most other conditions, VWF:Ag was increased in the inflammation group, indicative for endothelial activation. Despite the presence of ADAMTS13 in endothelial cells (Shang et al, 2006; Turner et al, 2006), we found no specific rise of ADAMTS13 antigen nor activity. On the contrary, our observations of low activity and antigen values might imply that ADAMTS13 is not released from these cells or that it is rapidly cleared or consumed upon release.
In 2004, we showed that cardiac surgery drastically lowered ADAMTS13 activity (Mannucci et al, 2005). This is in contrast to recent findings of Lo et al (2007), who showed that ADAMTS13 activity towards the FRETS-VWF73 substrate was increased in off-pump cardiac surgery. Our current study extends our previous findings, showing that ADAMTS13 antigen levels were decreased after cardiac surgery (both off and on-pump cases were included), yet were higher than enzymatic activity at all investigated time-points, yielding a significantly discrepant ratio. The activation of coagulation during inflammation and cardiac surgery may result in inhibition of ADAMTS13 activity through proteolytic inactivation by thrombin (Crawley et al, 2005).
In conclusion, measuring both ADAMTS13 activity and antigen is preferred over measuring only one parameter, as exemplified by the novel information obtained in this study that investigated the same conditions reported by Mannucci et al (2001). Moreover, the antigen assay used here is less prone to artifacts than the most widely used activity assays. It remains to be explained and understood whether low levels of protease antigen found in newborns, after cardiac surgery and, to a lesser extent, during pregnancy are due to inhibition of enzymatic activity, consumption of the protease or increased clearance from plasma.