Complement in bullous pemphigoid: results from a large observational study

Dear Editor, Bullous pemphigoid (BP) is a subepidermal autoimmune blistering disease characterized by autoantibodies directed against the hemidesmosomal proteins 180‐kDa bullous pemphigoid antigen 2 (BP180, BPAG2) and/or 230‐kDa bullous pemphigoid antigen 1 (BP230, BPAG1).1 It is thought that complement activation plays an important role in the pathogenesis of BP.1 However, the role of complement in the formation of subepidermal blisters has been disputed, and complement‐independent mechanisms of blister formation have been described.2,3,4,5 However, interpretation of these data is limited, as almost all studies were based on experimental pemphigoid models of mice2,3 and human keratinocytes.4,5 Studies assessing the role of complement in a large population of patients with BP are lacking. Using a large database comprising a period of 13 years (2002–2015) we had the unique opportunity to evaluate the relationship of complement with various clinical and diagnostic parameters in a large cohort of patients with BP. This was a single‐centre retrospective observational study conducted at the Department of Pathology and Dermatology of the University Medical Centre Groningen. The study was performed according to local legislation. Patients diagnosed with BP between 2002 and 2015 were enrolled. Diagnosis was based on clinical features and positive direct immunofluorescence showing antibodies and/or C3c in an n‐serrated pattern along the basement membrane zone (BMZ) of patient skin and, if performed, indirect immunofluorescence on salt‐split skin (IF SSS), immunoblot and NC16A enzyme‐linked immunosorbent assay (ELISA). Patients with epidermolysis bullosa acquisita, antilaminin‐332 pemphigoid, anti‐p200 pemphigoid or mucous membrane pemphigoid were excluded. Statistical analyses were performed using SPSS 22.0 (IBM, Armonk, NY, U.S.A.). Associations between variables were tested with the χ2‐test, and means were compared with the independent‐samples t‐test. Univariate and multivariate analysis was performed with logistic regression, with C3c deposition as the dependent variable. Effect modification was tested using interaction variables. Outcomes with a P < 0·05 were considered significant. In total 301 patients with BP were enrolled – 138 men and 163 women – with a mean age of 73·3 ± 15·5 years. Overall 250 patients (83·1%) showed C3c deposition along the epidermal BMZ in their skin biopsy. Univariate analysis showed an association between C3c deposition and clinical and serological activity (IF SSS, BP180 + BP230 immunoblot, NC16A ELISA). In patients with BP the mean NC16A titre was significantly higher in the group of patients with C3c (78·0 vs. 40·8). Multivariate analysis showed an association between C3c and blisters at the moment of biopsy. There was no significant difference between C3c, erythema, itch, mucosal involvement or the presence of inflammatory cells assessed on haematoxylin and eosin slides from skin biopsies taken for routine histology. Patients receiving systemic treatment (21·3%) at the time of biopsy were equally distributed between both groups (Table 1). Laboratory results for patients with and without C3 Data are shown as the number of patients positive/total number of patients (%). BP, bullous pemphigoid; DIF, direct immunofluorescence; ELISA, enzyme‐linked immunosorbent assay; IF SSS, immunofluorescence on salt‐split skin. aUnivariate analysis calculated with Fisher's exact test or χ2‐test. bP = 0·030 on multivariate analysis (calculated with multivariate logistic regression analysis, χ2 = 14·10, five degrees of freedom). cP = 0·001 with t‐test. Laboratory results for patients with and without C3 Data are shown as the number of patients positive/total number of patients (%). BP, bullous pemphigoid; DIF, direct immunofluorescence; ELISA, enzyme‐linked immunosorbent assay; IF SSS, immunofluorescence on salt‐split skin. aUnivariate analysis calculated with Fisher's exact test or χ2‐test. bP = 0·030 on multivariate analysis (calculated with multivariate logistic regression analysis, χ2 = 14·10, five degrees of freedom). cP = 0·001 with t‐test. A subgroup of 28 patients in whom blisters never occurred (pruritic nonbullous pemphigoid, NBP)6 was analysed separately. Patients with pemphigoid without blisters had statistically significantly less C3c deposition than those with blisters (39% vs. 83%, P < 0·001). In NBP, C3c deposition was associated with positive IF SSS (P = 0·009). When comparing lesional, perilesional and uninvolved skin for C3c deposition in all patients with BP, uninvolved skin was significantly less often C3c positive (68·9%) than lesional (83%) and perilesional (84·8%) skin (Table 2). Immunoglobulin deposition was not statistically different between biopsy sites (Table 2). Immune depositions along the epidermal basement membrane zone in bullous pemphigoid (BP) according to biopsy Data are shown as the number of biopsies (%). In total 407 biopsies were taken from 301 patients with BP. For 15 biopsies the skin type was ‘missing’ (total 392). aSignificant difference in C3c deposition between lesional and uninvolved skin (P = 0·043). bSignificant difference in C3c deposition between perilesional and uninvolved skin (P = 0·001). Immune depositions along the epidermal basement membrane zone in bullous pemphigoid (BP) according to biopsy Data are shown as the number of biopsies (%). In total 407 biopsies were taken from 301 patients with BP. For 15 biopsies the skin type was ‘missing’ (total 392). aSignificant difference in C3c deposition between lesional and uninvolved skin (P = 0·043). bSignificant difference in C3c deposition between perilesional and uninvolved skin (P = 0·001). This study, comprising more than 300 patients with BP, showed that in the majority of skin biopsies, deposition of complement could be observed, which was related to clinical and serological disease activity and biopsy site, strengthening the importance of complement activation in the pathophysiology of blister formation in BP. Complement deposition was associated with NC16A ELISA titre, and the latter was shown by others to correlate with disease activity.7 However, in a considerable proportion of patients, no epidermal BMZ complement deposition was present, pointing out that it is possible to develop BP without the local presence of complement. In this respect, it should be noted that only one antibody against complement was used and that use of different antibodies might increase sensitivity. Various studies have shown that activation of the complement system is crucial in the development of BP. Passive transfer of rabbit antimurine IgG against the NC16A domain of murine BP180 initiated blister formation in C5‐sufficient, but not in C5‐deficient mice. Moreover, passive transfer of F(ab′)2 fragments, unable to activate the complement system, did not initiate blister formation.8 Other crucial complement‐dependent mechanisms described are mast cell activation and degranulation,9 and neutrophil attraction and activation, secreting neutrophil elastase and matrix metalloproteinase (MMP)‐9 cleaving murine BP180.10 That BP could not develop without activation of the complement system has been questioned by the report of a C4‐deficient patient with BP;11 the finding that in two cases of complement‐negative BP, IgG4 antibodies with limited complement‐fixing ability were the dominant antibody;12 and the fact that IgG4 antibodies have been reported to predominate in BP,13 especially in complement‐negative patients with BP.14 Experimental studies have described complement‐independent mechanisms that could play a role in the pathophysiology of BP. Autoantibodies from patients with BP have been shown to deplete directly cultured human keratinocytes of BP180,4 and internalization of the IgG–BP180 complex through endocytosis has been shown.5 In an experimental BP180 humanized mouse model, skin fragility was induced by passive transfer of F(ab′)2 fragments, and in C3‐deficient mice, blister formation could be induced by passive transfer of anti‐BP180 antibodies from patients with BP.2,3 Another complement‐independent mechanism of disruption of the hemidesmosome has been described in pancreatic ductal adenocarcinoma, in which cleavage of the extracellular domain of BP180 by MMP‐9 has been shown to be dependent on the phosphoinositide 3‐kinase pathway.15 It has been hypothesized that these contradictory results can be explained by the difference in the target epitopes on BP180,2 the applied dose of antibodies, the affinity of antibodies to the target molecules and the clonality of antibodies in each experimental system.3 Thus, the association found in our study between complement, NC16A titre, positive IF SSS and immunoblot could therefore be caused by higher levels of circulating IgG and subsequently more clinically active disease. The fact that we did not find an association between complement and blister cellularity could be explained by the observation that in cultured human keratinocytes, binding of anti‐BP180 IgG induces the secretion of interleukin‐8, a cytokine capable of neutrophil attraction in a complement‐independent manner. Other explanations for cellular blisters in complement‐negative skin could be direct activation of local inflammatory cells by Fcγ receptors16 or a direct effect of pathogenic IgE antibodies, which are capable of activating mast cells, attracting eosinophils and producing inflammatory cytokines in a complement‐independent manner.17 Negative direct immunofluorescence in patients with complete remission after systemic treatment has been described,18 and could be a confounder in our study. However, in both complement‐positive and ‐negative patients, the same percentage of patients (21%) had received systemic therapy at the moment of biopsy. In conclusion, our study supports the view that complement is an important factor in the pathogenesis of BP and is related to more active disease. However, this study also shows that complement deposition is not required, and that other, complement‐independent mechanisms may be involved in BP. Funding sources: none. Conflicts of interest: none declared.