Diffuse skin necrosis in a patient with an anti-endothelial cell protein C receptor autoantibody which blocks protein C activation


Jean-Christophe Gris, Laboratoire d'Hématologie, Centre Hospitalier Universitaire, Groupe Hospitalo-Universitaire Caremeau, Place du Pr. Robert Debré, F-30029 Nîmes Cédex 9, France.
Tel.: +33 4 66 68 32 11; fax: +33 4 66 68 36 48; e-mail: jcgris@chu-nimes.fr

The endothelial cell protein C receptor (EPCR) is a transmembrane protein which is closely associated to a phospholipid. It binds protein C and activated protein C (APC) and increases protein C activation by the thrombin–thrombomodulin complex in vitro. During meningococcemia, skin biopsies have shown a strong downregulation of EPCR expression on the dermal microcirculation, which is linked to the impairment of protein C activation associated with cutaneous necrotic lesions [1]. Autoantibodies against EPCR have been described recently in thrombotic antiphospholipid syndrome patients and in women with fetal death [2].

We report the case of a 74-year-old woman who was admitted to the Department of Dermatology because of an extensive skin necrosis. Three weeks earlier, she had developed a superficial thrombosis of the lower limb and received nadroparin sodium for 10 days at home. Several days after this treatment nadroparin was renewed, followed by oral anticoagulation (fluindione, 20 mg per day). Three days later, fluindione treatment was stopped due to macroscopic hematuria. Nine days later, an apyretic acute extensive skin necrosis developed, involving crural regions, the abdominal region, the breast and one forearm. Blood cultures were sterile. Protein C, protein S levels and prothrombin time values were back to normal at the onset of necrosis.

On admission, a lupus anticoagulant (LA) plasma activity was identified (diluted Russell's viper-venom time). Three days later she was admitted to the intensive care unit due to an extension of skin necrosis. The skin biopsy showed a diffuse capillary leucocytoclastic angiitis and thrombosed vessels with the presence of hyaline microthrombi (dermal–hypodermal junction). Treatment of danaparoid sodium, 200 IU h−1 and methylprednisolone, 80 mg day−1 were started; although no new skin lesion appeared, the patient died of a massive acute ischemic stroke.

All the tested patient's plasma samples were obtained when she was still in the intensive care unit before starting treatments, and plasmas obtained from 100 asymptomatic women, 65–75 years old, were used as controls.

The patient's antithrombin and protein S activities were within normal ranges. Plasma protein C activity and antigen levels were high (255 and 260%, respectively). The factor V Leiden mutation and the prothrombin 20210 A allele were negative. Anticardiolipin, antiβ2-glycoprotein I, antiannexin V, antiprothrombin, antiphosphatidylethanolamine, antitPA, antithrombomodulin, antitissue factor, antiprotein C and antiprotein S IgG and IgM antibodies, assayed as described previously [3], were all within normal values. Plasma concentrations of soluble thrombomodulin (14.5 ng mL−1) and soluble EPCR (325 ng mL−1) were within the highest quartile of normal values (Diagnostica Stago, Asnières, France). Heparin-induced thrombocytopenia was unlikely (normal platelet counts, no circulating heparin-dependent aggregating activity, negative antiplatelet factor 4/heparin antibodies). These results were controlled on four occasions during the sickness of the patient without any significant change.

APC plasma concentrations, assayed as described [4], were lower than 10% of the median control value obtained from the 100 normal control plasmas (median value: 41 pmol L−1 human APC; range: 23.8–61.3 pmol L−1): this was measured six times over the sickness of the patient without any consistent variation.

A highly positive anti-EPCR IgM autoantibody, assayed using a home-made enzyme-linked immunosorbent assay (ELISA) based on a commercially available recombinant extracellular portion of the human EPCR (Diagnostica Stago) and checked using the reference ELISA [2] was found in the patient's plasma (negative IgG). The patient's immunoglobulin fractions were prepared by affinity chromatography using antihuman γ-chain, µ-chain and α-chain antibodies coupled to CNBr-Sepharose 4B beds; specific anti-EPCR IgM were purified by affinity adsorption (0.92 mg from 40 mL plasma), this technique also being used to prepare the patient's immunoglobulin depleted of anti-EPCR IgM. When added to a series of control plasmas, the patient's immunopurified IgM had no LA activity, whereas a LA activity was obtained using the patient's immunoglobulins, and the patient's imunoglobulins depleted of their anti-EPCR IgM. When cultured human umbilical vein endothelial cells were incubated with human thrombin (Diagnostica Stago) and human protein C (ProtexelR, LFB, France), the specific IgM impaired the thrombin-induced APC generation, as monitored through the kinetics of S-2366 proteolysis (Chromogenix, Milan, Italy) by APC. This was not the case using the patient's imunoglobulins depleted of anti-EPCR IgM. A commercially available antibody, known to inhibit protein C binding on EPCR [5], and its immunogen synthetic peptide were used as controls (PolyFastTM ZMD.151, Zymed Laboratories, USA): a similar inhibition of the initial velocity of the reaction (Vo) was obtained using ZMD.151 and the patient's immunopurified IgM at 10 µg mL−1, this inhibition being abolished by prior incubation of both antibodies with the synthetic peptide corresponding to amino acids 60–100 of EPCR. Finally, the decrease in the thrombin-induced APC generation observed when the patient's anti-EPCR IgM was added to cells was due to an increase of the Km of the reaction rather than Vmax, which is compatible with an inhibition of the EPCR-mediated enhancement of protein C activation (Fig. 1a). This effect was not obtained using the patient's imunoglobulins depleted of the anti-EPCR IgM. We found thereafter that the patient's IgM fraction induced a dose-dependent inhibition of the APC binding to EPCR anchored to microplate wells (Fig. 1b), whereas the patient's imunoglobulins depleted of anti-EPCR IgM had no effect, thus demonstrating definitively that the inhibitory effect of the antibody on the APC generation was due to the inhibition of protein C binding to its receptor.

Figure 1.

(a) Effect of controls' IgM and patient's anti-EPCR immunopurified IgM on kinetic parameters of thrombin-induced APC generation by endothelial cells.Closed circles: controls' IgM; open squares: patient's anti-EPCR IgM; crosses: patient's immunoglobulins depleted of the anti-EPCR IgM. (b) Inhibition of APC binding to EPCR. Controls' IgM fraction (open circles), the patient's IgM fraction (solid circles) or the patient's immunoglobulins depleted of the anti-EPCR IgM (crosses) were incubated with well-anchored EPCR. APC capture by EPCR was measured by residual amidolytic APC activity. Results are expressed as percentage with respect to the residual APC activity in the absence of antibodies.

Extensive cutaneous necrosis has been described in patients with a protein C [6] or protein S [7] deficiency: homozygous [6,7] or heterozygous ones at the initiation of oral anticoagulants [8,9]. It is described in patients developing heparin-dependent, platelet-activating antibodies [10] in patients with antiphospholipid antibodies [11]—sometimes associated with an acquired protein S deficiency [12]—and during severe sepsis, such as meningococcemia, known to induce a severe protein C consumption [13].

Our patient only had a lupus anticoagulant. The responsibility of oral anticoagulants was ruled out (initiation of necrosis 9 days after the end of fluindione therapy, protein C and protein S levels normal at the onset), heparin-dependent antibodies were negative. Antiphospholipid antibodies-associated skin necrosis is a rare clinical feature [14] and case reports of widespread necrosis are exceptional: classical antiphospholipid antibodies are unlikely to be consistent risk factors.

The anti-EPCR IgM found in our patient was a strong one. This autoantibody prevented the binding of protein C to EPCR and thus dramatically decreased the APC generation. This in vitro feature would explain the low circulating APC levels found in the patient and the autoantibody-induced acquired protein C functional deficiency might have led to extensive skin necrosis. The high levels of circulating sEPCR, which are common in case of increased thrombin generation, may have further enhanced the anti-EPCR induced impaired APC generation through competition with membrane-bound EPCR for protein C and APC binding.


We thank E. Mercier, E. Cardi and H. Bres for technical assistance, M. Manson for editorial assistance and K. Martinez for technical support. We thank our colleagues C. Arich, C. Gervais and C. Bengler, all from the intensive care unit, who trusted in our investigations. We warmly thank J. Hermida and R. Montes for their assistance, and for the kind gift of their recombinant soluble EPCR with myc epitope and six histidine residues at the C-terminal end.