To the Editor:
In a recent article in Blood, Mesters et al1 have prospectively evaluated coagulation parameters in 62 patients with severe chemotherapy-induced neutropenia. Thirteen patients progressed to severe sepsis and another 13 to septic shock. Septic shock was associated with increased thrombin generation. Increased factor VIIa (FVIIa) and decreased antithrombin III measurements both were found to be sensitive markers of an unfavorable outcome. As markers of contact system activation, activated FXII (FXIIa) antigen was measured in these patients. At onset of fever, FXIIa levels were not different between the patient group with severe sepsis and the patients with progression to septic shock, respectively, and FXIIa increased only modestly during the course of septic shock. The investigators conclude that the contribution of the contact system to coagulation activation in septic shock is a minor one.
We are concerned about the results of FXIIa measurements. Plasma levels of FXIIa antigen were determined by a sandwich enzyme-linked immunosorbent assay (ELISA) technique. The investigators reference the assay as being in press in 1996 in the Journal of Immunological Methods. However, we are aware of a report most probably concerning this assay that has been published in the Journal of Immunoassay.2 From the assay procedure described in this report, we conclude that the assay is not suitable for measurement of activated FXII in plasma samples. This assay uses purified β-FXIIa as a standard. However, it was not possible to prepare a standard curve in plasma, because FXIIa lost its activity after addition to plasma due to complex formation with plasma protein inhibitors. Maybe the ELISA measures part of formed FXIIa-inhibitor complexes.
Using assays measuring FXIIa-C1-inhibitor, kallikrein-C1-inhibitor, and FXIa-inhibitor complexes,3,4 Nuijens et al were able to show FXII activation and kallikrein formation in patients with sepsis,5 and we found activation of FXII, prekallikrein, and FXI in children with septic shock due to meningococcal infection.6 These data clearly show the contact system to be activated in patients with severe sepsis and in patients with septic shock. However, we agree with the investigators that it remains unclear whether contact activation indeed mediates coagulation activation in these patients.
Response
In their letter, Wuillemin and Lämmle have apparently interpreted our data as indicating that there is a minor contribution of the contact system in septic shock.1-1 In our study, the nonsignificant increase of FXIIa, measured at the onset of fever in patients who subsequently developed septic shock, does not mean that the contact system is not involved in septic shock. This finding simply means that FXIIa has no prognostic value and that there are more sensitive markers of an unfavorable prognosis such as FVIIa and ATIII measurements. Incidentally, we wish to point out that we found a decrease of FVIIa1-1 and not an increase as understood by Wuillemin and Lämmle. We found a significant increase of FXIIa (P = .001) during the course of septic shock.1-1 These results are in agreement with those of both Nuijens et al,1-2 who, using a radioimmunoassay to measure FXIIa- and kallikrein-C1-inhibitor complexes, found a variable degree of contact system activation in sepsis and the same Wuillemin et al.1-3 In the latter study of 13 children with meningococcal septic shock, 4 had increased levels of FXIIa-C1-inhibitor complexes and 2 had increased levels of kallikrein-C1 inhibitor complexes on admission to the intensive care unit.
Wuillemin and Lämmle also questioned whether the FXIIa assay used by us actually measures FXIIa or FXIIa-inhibitor complexes, because an FXIIa standard when added to normal plasma loses its activity due to complex formation with inhibitors. We are aware that there are technical difficulties in preparing the ideal standard for the FXIIa assay. These difficulties are not unique for the FXIIa assay and apply to other markers of activation, including the widely used D-dimer assay. Supplying standards such as purified βFXIIa in buffer provides a stable and reproducible reference solution that can be used to compare different samples. Perhaps we should have given the results as XIIa equivalents rather than in nanograms per milliliter of XIIa.
With regard to what the assay recognizes in plasma, Esnouf detected no FXIIa-C1-inhibitor complexes when the complexes were added to plasma (Peter Esnouf, unpublished observations). In addition, using Western blot techniques, he also showed that the antibody detects αFXIIa in human plasma.
