Paraneoplastic pemphigus: the role of tumours and drugs

It has now been 10 years since paraneoplastic pemphigus (PNP) was first described.1 The originally proposed features of the syndrome have proven to be quite accurate, although clarification of some points has occurred. The disease is associated with a small number of lymphoproliferative disorders. Approximately 80% of cases are linked to just three neoplasms: Non-Hodgkin’s lymphoma (NHL), chronic lymphocytic leukaemia (CLL), and Castleman’s disease.2 Some cases are less commonly associated with retroperitoneal sarcomas, thymoma and Waldenstrom’s disease. The most constant clinical feature is the presence of intractable stomatitis. The autoantibody response is characterized by antibodies against desmoglein 13 and 3, as well as antibodies against proteins of the plakin family of desmosomal proteins. Of the various plakin antigens recognized by individual patients, the most constant serological findings are autoantibodies against periplakin and envoplakin.4 Some patients with CLL might respond to immunosuppressive treatment,5 but virtually all patients with NHL die from their disease or the complications of treatment. Many die from pulmonary involvement, which is a unique feature of PNP.6 The three cases described by Gooptu et al.7 clearly demonstrate many of these findings. The authors were concerned about the temporal relationship between administration of fludaribine, and the development of PNP. They also correctly note that cases of PNP have developed in temporal association with radiation therapy and treatment with interferon-alfa. Over the years, I have participated in the characterization of more than 140 cases of the disease, and no consistent pattern has arisen regarding chemotherapeutic treatment and development of the autoimmune disease. One must also recognize that about one-third of cases of PNP arise in patients with occult neoplasms, so that no therapeutic agent can be implicated. In the early 1990s, I suspected that treatment with cyclophosphamide, then a commonly used cytotoxic and immunomodulatory drug, may have initiated autoimmunity in PNP patients. This association did not bear out. When fludaribine was introduced as a new agent, it then became suspect. I feel that any newly introduced drug that is useful in the treatment of NHL and CLL will have the risk of being implicated in inducing the development of PNP. There is no registry for comprehensive data acquisition about this disease, so adverse events occurring with the use of new drugs will be affected by reporting bias. As new biological agents such as monoclonal antibodies are introduced for treatment of these neoplasms, the same questions may be raised as cases of PNP are identified in patients receiving these agents. Without clear data showing the risk of PNP developing in patients treated with older regimens vs. those treated with newer chemotherapeutic or biological agents, the association is only speculative. Also in this issue, Schaeppi et al.8 report the presence of suprabasilar acantholysis surrounding a primary melanoma. The authors suggest that the term ‘localized paraneoplastic pemphigus’ could be applied to this phenomenon, but there are reasons why this would not be appropriate. All forms of pemphigus are associated with antidesmoglein antibodies, and all are systemic autoimmune diseases. Absent in this case were the clinical signs of PNP: stomatitis, disseminated skin lesions, and evidence of antidesmoglein autoantibodies in skin and serum. The patient had autoantibodies that identified desmoplakin 1 and plakoglobin by immunochemical techniques, but antibodies against desmoplakin I are present in many disorders, including pemphigus vulgaris and erythema multiforme.9 In PNP, antibodies against envoplakin/periplakin are the most unique markers, and these also were absent in this case. Additionally, a paraneoplastic syndrome is defined as a syndrome caused by a remote effect of a neoplasm, not attributable to invasion by the tumour or by tumour metastases. It is not clear that this patient had such a syndrome, for the phenomenon was localized to the tissue immediately surrounding the primary tumour, and was not present distally, even in the face of recurrent metastatic disease. What is intriguing is the presence of such clear acantholysis in the absence of desmoglein autoantibodies, and the suggestion that antibodies against plakoglobin or desmoplakin can induce cell detachment, perhaps in the presence of locally produced factors, derived from the melanoma. So, why do patients with these specific neoplasms develop PNP? One must recognize that autoimmunity in patients with NHL, CLL and Castleman’s disease is well recognized by physicians in haematology/oncology, and knowledge of this predated the description of PNP. More commonly, patients develop autoimmune cytopenias, with autoantibodies against platelets, neutrophils or red cells.10 In these cases, the autoimmune disease is not influenced by the tumour burden, does not respond well to immunosuppressive therapy, and frequently is the cause of death. In some patients, it seems that these tumours induce an autoimmune response against epidermal cell proteins, not haematopoietic cells, and these patients present with PNP. This does not always occur in isolation, for patients with PNP may also develop additional autoimmune cytopenias, as documented in case no. 2 by Gooptu et al. This would indicate a complex and unpredictable effect of the tumour on the immune system, with potentially fatal autoimmunity as the end result. This disease remains a complex and challenging disorder. It is our expectation that the next decade will provide further insights into its pathogenesis, and provide more hope for affected patients.