M. Cugno, Department of Internal Medicine, University of Milan, Via Pace 9, Milan 20122, Italy. E-mail: firstname.lastname@example.org
The main autoimmune blistering skin disorders are pemphigus vulgaris (PV) and bullous pemphigoid (BP). They differ in the inflammatory infiltrate, which is more intense in BP. Inflammation is known to activate coagulation in several disorders. Local and systemic activation of coagulation was evaluated in BP and PV. We studied 20 BP patients (10 active and 10 remittent), 23 PV patients (13 active and 10 remittent) and 10 healthy subjects. The coagulation markers prothrombin fragment F1+2 and D-dimer were measured by enzyme-immunoassays in plasma. The presence of tissue factor (TF), the main initiator of blood coagulation, was evaluated immunohistochemically in skin specimens from 10 patients with active PV, 10 patients with active BP and 10 controls. Plasma F1+2 and D-dimer levels were significantly high in active BP (P = 0·001), whereas in active PV the levels were normal. During remission, F1+2 and D-dimer plasma levels were normal in both BP and PV. TF immunoreactivity was found in active BP but neither in active PV nor in normal skin. TF reactivity scores were higher in active BP than in controls or active PV (P = 0·0001). No difference in TF scores was found between active PV and controls. BP is associated with coagulation activation, which is lacking in PV. This suggests that BP but not PV patients have an increased thrombotic risk. The observation that thrombotic complications occur more frequently in BP than in PV further supports this view.
Bullous pemphigoid (BP) and pemphigus vulgaris (PV) are the main autoimmune skin disorders characterized clinically by the presence of blisters and erosions, but they differ in the depth of bulla formation and in inflammation [1–3]. In fact, in PV the bulla is intraepidermal due to immunoglobulin G (IgG) autoantibodies directed against the desmosomal desmogleins 3 and 1, whereas in BP the bulla is dermoepidermal due to IgG autoantibodies against two hemidesmosomal antigens (BP 180 and BP 230). On the other hand, a crucial role of autoreactive T cells in the induction and regulation of autoantibody production has been suggested for both diseases [4–6]. Histologically, BP is characterized by a dense dermal inflammatory infiltrate composed mainly of lymphocytes and eosinophils; in contrast, in PV, the major variant of pemphigus, the dermal infiltrate is mild, containing only few eosinophils . Clinically, in BP, the blisters are ‘hot’, due to the presence of an erythematous-oedematous halo, while in PV they are ‘cold’, arising usually on normal skin. Extensive cross-talk between inflammation and coagulation has been demonstrated , and markers of coagulation activation have been found in several inflammatory systemic diseases of different pathogenesis such as rheumatoid arthritis [9,10], inflammatory bowel diseases  and sepsis [12,13]. In several diseases, the measurement of biomarkers of coagulation activation has been employed widely . Using these biomarkers, preliminary data indicate that the coagulation cascade is also activated in an inflammatory cutaneous disorder such as BP and that the eosinophils, present in the BP inflammatory infiltrate, strongly express tissue factor (TF), the main initiator of blood coagulation .
With this background, we evaluated the coagulation activation in BP and PV, two bullous skin disorders that share the autoimmune origin but differ in the intensity of inflammation.
Patients and methods
We enrolled 13 consecutive patients with previously untreated active PV (eight males and five females; mean age 50 years, range 23–82 years), 10 patients with PV in clinical remission (three males and seven females; mean age 50 years, range 17–70 years), 10 patients with active BP (five males and five females; mean age 77 years, range 54–99 years) and 10 patients with BP in clinical remission (seven males and three females; mean age 69 years, range 60–87 years) admitted to our dermatology department and 10 healthy controls (five males and five females; mean age 54 years, range 34–80 years). The diagnosis of PV was based on typical clinical findings, histopathological features of supra-basilar acantholysis, demonstration of intercellular IgG deposition with or without C3 by direct immunofluorescence of perilesional skin; demonstration of circulating IgG anti-epidermis antibodies by indirect immunofluorescence on monkey oesophagus substrates; and detection of circulating anti-desmoglein 3 (Dsg3) (with or without anti-Dsg1 antibodies) by enzyme-linked immunosorbent assay (ELISA). Active PV patients were defined as having blisters/erosions, whereas remittent PV patients were defined as having no blisters/erosions for at least 1 month. At the sampling time, active PV were newly diagnosed and were not consuming any drug, whereas remittent PV patients received a low dosage of corticosteroids (methylprednisolone 4 mg daily).
The diagnosis of BP was established on the basis of both clinical and immunopathological criteria. All the patients had a clinical picture of generalized BP without any mucous membrane involvement. Direct immunofluorescence examination of the perilesional skin revealed the linear deposition of IgG and/or C3 in the basement membrane zone (BMZ) in all patients; when present, circulating anti-BMZ autoantibodies were detected by means of indirect immunofluorescence on monkey oesophagus substrates. Circulating anti-BP180 autoantibodies were found by an ELISA. Active BP patients were defined as having blisters/erosions, whereas clinical remission was defined as the absence of any new BP lesions with complete healing of the previous lesions for a minimum of 1 month. At sampling time, active BP patients were newly diagnosed, consuming neither immunosuppressive/anti-inflammatory agents nor anti-coagulant drugs, while BP patients in clinical remission were under treatment with a low dosage of corticosteroids (methylprednisolone 4 mg daily). Concomitant neoplastic, inflammatory or coagulation pathologies were excluded on the basis of clinical, laboratory and instrumental examinations.
Sodium citrate anti-coagulated plasma samples taken from all patients and normal subjects were stored in plastic cones at −80°C until in vitro assay.
The protocol was approved by our Institutional Review Board, and all the subjects gave their written informed consent before participating in the study.
Prothrombin fragment 1+2 measurements
Prothrombin F1+2 levels were measured using a sandwich ELISA (Enzygnost F1+2; Behring Diagnostic GmbH, Frankfurt, Germany), with intra- and interassay coefficients of variation (CV) of 5% and 8%, respectively.
D-dimer levels were measured by means of ELISA (Zymutest D-dimer; Hyphen BioMed, Neuville-sur-Oise, France) in accordance with the manufacturer's instructions. The intra- and interassay CV were 10% and 15%, respectively.
Skin specimens were obtained from the early-appearing lesions (≤24 h) of all patients with active BP or PV. Controls were normal skin tissue specimens taken from 10 patients who underwent the excision of benign skin tumours. The tissue samples were fixed in buffered formalin, dehydrated, embedded in paraffin wax and sectioned; no antigen unmasking pretreatment was needed. After deparaffining and rehydrating, each tissue section was placed on a Dako automated immunostainer, and incubated with the specific monoclonal antibody (mouse anti-human tissue factor, 1 : 100; American Diagnostica Inc., Greenwich, CT, USA) at room temperature for 45 min, and then washed with Tris-buffered saline (TBS), pH 7·6, and incubated in biotinylated goat anti-mouse and anti-rabbit immunoglobulins (Dako REAL™, cod.K5005; Dako Cytomation, Glostrup, Denmark) at room temperature for 30 min. After incubation with the secondary antibody and another washing with TBS, pH 7·6, the sections were incubated with streptavidin conjugated to alkaline phosphatase (Dako REAL™, cod.K5005; Dako Cytomation) at room temperature for 30 min. A red chromogen solution was prepared as indicated by the Dako REAL™ datasheet and used as an enzyme substrate, followed by counterstaining with Mayer's haematoxylin. After air-drying, each section was coverslipped using the VectaMount™ mounting medium (Vector Laboratories, Burlingame, CA, USA). A negative control was performed using a pool of mouse immunoglobulins (IgG1, IgG2a, IgG2b and IgM) as primary antibody (negative control; Dako Cytomation). Two independent ‘blinded’ observers evaluated serial sections. Tissue factor immunoreactivity was scored according to the number of immunoreactive cells per field (200×): 0 = no immunoreactive cells, 1 = between 1 and 5, 2 = between 6 and 20 and 3 = more than 20.
Results are reported as mean ± standard error of the mean (s.e.m.) and were log-transformed before analysis because they were positively skewed. Student's t-test for unpaired values was used to assess the statistical significance of differences. Differences in the immunohistochemical scores were assessed using the Wilcoxon–Mann–Whitney non-parametric test. P < 0·05 was considered to indicate a statistically significant difference or correlation.
Prothrombin fragment 1+2
Figure 1 shows the prothrombin F1+2 measurements. Plasma F1+2 levels were significantly higher in active BP patients (730 ± 116 pmol/l) than in normal subjects (155 ± 13 pmol/l) (P = 0·001), whereas levels were normal in active PV (136 ± 23 pmol/l) During remission, F1+2 plasma levels were normal in both BP (157 ± 15 pmol/l) and PV (119 ± 15 pmol/l) patients. In BP, the plasma levels of F1+2 correlated significantly with those of anti-BP180 autoantibodies (r = 0·715, P = 0·003).
Figure 2 shows D-dimer measurements. Plasma D-dimer levels were also significantly higher in active BP patients (22·63 ± 5·99 nmol/l) than in normal subjects (1·63 ± 0·21 nmol/l) (P = 0·003) and resulted normal in active PV (1·60 ± 0·19 nmol/l). During remission, D-dimer plasma levels were normal in both BP (2·27 ± 0·17 nmol/l) and PV (1·35 ± 0·17 nmol/l) patients, as found for F1+2. In BP, the plasma levels of D-dimer correlated significantly with those of anti-BP180 autoantibodies (r = 0·744, P = 0·001).
All the skin samples from active BP patients, processed for immunohistochemistry, were characterized histologically by the presence of a dermal inflammatory infiltrate consisting mainly of lymphocytes and eosinophils (Fig. 3a and b). The immunohistochemical experiments revealed TF reactivity clearly in the specimens taken from the active BP patients (Fig. 3c and d), whereas the normal skin samples showed no TF reactivity at all (Fig. 3e and f).
The skin samples from active PV patients, processed for immunohistochemistry, were characterized histologically by a mild upper dermal inflammatory infiltrate consisting mainly of lymphocytes invading the epidermis (Fig. 4a and b). The immunohistochemical experiments revealed the absence of TF reactivity in the specimens taken from the active PV patients (Fig. 4c and d).
There was a statistically significant difference in TF reactivity scores between the skin samples from active BP patients and those of both normal controls and patients with active PV (P = 0·0001 for both). No significant difference in TF reactivity scores was found between active PV patients and normal controls (Fig. 5).
Our data show clear differences in activation of blood coagulation between BP and PV, both of which are autoantibody-mediated blistering skin disorders but which differ in the nature of inflammatory infiltrate [1–3,7]. In fact, BP shows signs of coagulation activation via the TF pathway, as demonstrated locally by the immunohistochemical expression of TF in the inflammatory infiltrate and systemically by high levels of prothrombotic markers such as F1+2 and D-dimer ; these signs are completely lacking in PV. Recently, using co-localization experiments, we demonstrated that the cells expressing TF in the inflammatory infiltrate of BP are eosinophils due to the co-expression of the eosinophil cationic protein (ECP) . Eosinophils are important components of the BP inflammatory infiltrate , which also consist of numerous T lymphocytes and few other inflammatory cells [17,18]. Notably, the T lymphocytes infiltrating lesional skin of BP exhibit a T helper type 2 (Th2) phenotype and produce cytokines and chemokines inducing recruitment and activation of eosinophils, in particular interleukin-5 and eotaxin [17–19]. In contrast, PV lesional skin is characterized by a mild inflammatory infiltrate consisting mainly of T lymphocytes and few other inflammatory cells . Literature data regarding the T lymphocyte phenotype in PV are controversial, with some authors reporting a Th1 phenotype , others a Th2 phenotype  and still others a mixed Th1–Th2 phenotype . However, what appears clear histologically is that the eosinophils in the inflammatory infiltrate of PV are few or absent, at variance with BP, in which they are largely represented . It has been demonstrated that the eosinophil is the blood cell producing the largest amount of TF and TF can also facilitate specifically the early transendothelial migration of the eosinophils ; this finding is in agreement with our data showing lack of TF expression in PV skin specimens which are usually devoid of infiltrating eosinophils. Our observation of a positive correlation between coagulation markers and levels of circulating anti-BP180 autoantibodies may support the involvement of coagulation activation in BP pathophysiology. On the other hand, the absence of TF expression by eosinophils in lesional skin and the low plasma levels of F1+2 and D-dimer indicate that blood coagulation is not activated in PV. Thus, in the present study, it is evident that PV and BP differ in coagulation activation and this fact leads to noteworthy consequences, both at local and systemic levels. Locally, in BP but not in PV, activation of the extrinsic coagulation pathway via TF generates thrombin, an enzyme which increases vascular permeability [24,25], favouring the transendothelial migration of inflammatory cells and their accumulation in the skin. Moreover, thrombin and activated coagulation factor VII and X are proinflammatory mediators inducing the production and release of various interleukins, adhesion molecules, selectins and growth factors, thus amplifying the inflammatory network . Intriguingly, the activation of coagulation observed in BP may contribute to influence both the nature of the inflammatory infiltrate, which is dense and composed of various cellular types, and the feature of the bulla, which is ‘hot’, due to the presence of an erythematous–oedematous halo. In contrast, PV, in which TF-mediated activation of coagulation does not occur, is characterized histologically by a scanty inflammatory infiltrate, consisting mainly of T lymphocytes, and clinically by a ‘cold’ bulla, with no or only mild inflammation. Also the pathogenic role that autoantibodies play in the two diseases may account for the differences observed. In fact, in PV, binding of autoantibodies itself is sufficient to induce blister formation , whereas in BP autoantibodies induce a very complex inflammatory response by their Fc portion , which activates complement. The pivotal role of complement is supported by a murine model of BP in which the IgG autoantibodies directed against the ectodomain of the hemidesmosomal protein BP180 trigger subepidermal blistering via complement activation; such an effect was not evident in C5-deficient mice or after depletion of complement with cobra venom factor  At systemic level, the most important clinical consequence of the hypercoagulable state related to coagulation activation in BP is an increased thrombotic risk, as we have highlighted in a recent study . Analysing retrospectively a series of 130 BP patients and 60 PV patients seen at our institute during the last 4 years, we found an incidence of venous thrombosis of 8% per year in BP  and of 0·2% per year in PV. Thus, the venous thrombosis incidence is increased markedly in BP patients, whereas in PV patients it is similar to that reported in general population (1–2 per 1000 per year) . These findings are in agreement with the aforementioned data of the present study reporting high plasma levels of prothrombotic markers F1+2 and D-dimer in active BP but not in active PV patients. Future studies should aim to evaluate the possible involvement of coagulation activation in other eosinophil-mediated disorders, such as Churg–Strauss syndrome and hypereosinophilic syndromes, in which an increased thrombotic risk has also been reported [31–33].
The study was supported by ‘Fondo Interno per la Ricerca Scientifica e Tecnologica’, University of Milan.
None of the authors have any conflict of interest with the subject matter or materials discussed in the manuscript.