The diagnosis of BP is characteristically confirmed by direct immunofluorescence. Although it was performed for our patient’s cutaneous lesions, it was not obtained for the lymphoepithelial cyst of the pancreas. Once the diagnosis of the lymphoepithelial cyst of the pancreas was established, as direct immunofluorescence could not be performed in formalin-fixed tissue, immunoperoxidase staining with C3d was obtained. C3 has a well-established role in activation of complement and as a marker in BP. Deposition of C3d is a result of deactivation of C3b, a cleavage product of C3. In a study of 6 autoimmune blistering disorders that included 32 patients with BP, Pfaltz et al12 found positive immunoperoxidase staining for C3d in 31 of 32 patients, which translated to a sensitivity of 97%, a positive predictive value of 100%, and a negative predictive value of 98% among the blistering diseases being studied. Similarly, Magro and Dyrsen13 had positive staining of C3d in 17 of 17 (100%) patients with BP.
In theory, any process that involves deposition of C3 should be positive for C3d on immunoperoxidase staining. Other dermatologic inflammatory conditions stain positively with C3d, such as systemic lupus erythematosus, discoid lupus erythematosus, subacute cutaneous lupus erythematosus, and dermatomysositis.13 The staining for these diseases correlates with the site of the associated inflammatory component seen on hematoxylin and eosin staining. The staining of C3d along the basement membrane of stratified squamous epithelium in the lymphoepithelial cyst of the pancreas seen in our patient closely resembles the staining seen in cutaneous BP.
A proposed mechanism for BP in our patient would be exposure of BP-1 antigen in the pancreatic cyst leading to antibody recognition and C3 deposition along the basement membrane in the cyst, as evidenced by C3d immunoperoxidase staining. The IgG and C3 deposition along the cutaneous basement membrane would then represent a systemic response to the antigen exposure in the cyst. Thus, the lymphoepithelial cyst provided the immunologic stimulus for the development of the cutaneous BP. This theory is based on the observation of our patient’s rapid improvement without a change in his treatment regimen immediately after surgical excision of the cyst.
Despite the plausibility of our hypothesis, several questions remain regarding the validity of our assumptions. Although sensitive for C3 deposition, C3d immunoperoxidase staining is not specific for BP. If the proposed mechanism for causation is true, one might have expected that a subepithelial cleft along the basement membrane of the pancreatic cyst would be observed, which was not seen. A repeat BP antigen antibody was not obtained, which would have been helpful in determining if there was clearance of the antibody that would have correlated with the clinical resolution of the BP lesions.