Paraneoplastic pemphigus: Diagnostic mimics, confounders and management challenges in a series of five long‐term survivors from a single centre

Abstract We present a series of five cases who presented to our institution with treatment‐refractory mucosal ulceration, all of whom were subsequently diagnosed with paraneoplastic pemphigus (PNP). This case series highlights the diagnostic and treatment considerations for PNP – in particular, the steroid‐dependent, recalcitrant, polymorphic manifestations; the combination of histopathological and clinical findings that may overlap with clinically similar diseases, for example, pemphigus vulgaris and lichen planus; the importance of immunopathological findings for its diagnosis, and the need for surveillance and management of life‐threatening bronchiolitis obliterans.


| INTRODUCTION
Paraneoplastic pemphigus (PNP) is a rare autoantibodydriven mucocutaneous disease that not only poses a diagnostic challenge, but is also characterised by a poor prognosis and high mortality rate. 1 It presents similarly to other autoimmune bullous diseases (e.g., pemphigus vulgaris (PV) or bullous pemphigoid) and inflammatory dermatoses (e.g., lichen planus (LP)), which may delay diagnosis and appropriate treatment. PNP may, however, be distinguished by recalcitrance to conventional immunosuppression, an invariable association with malignancy (which may be subclinical), and its potential to cause life-threatening pulmonary complications.
We outline our experience of five cases of PNP managed at the Oral Immunology Clinic at Westmead Hospital, including two cases previously reported in 2015, 2 which highlight the clinical features and investigations that may lead to its early recognition and minimisation of morbidity (Table 1).

of 7 -
malignancy, none of which were clinically apparent at the time of the mucocutaneous manifestations and were only detected upon screening for a neoplastic process -multicentric Castleman disease (MCD, n = 1), unicentric Castleman disease (UCD, n = 2) and B-cell lymphoma (BCL, n = 2). Four cases received an alternate diagnosis prior to their PNP diagnosis in our institute, with all cases having a mean symptom duration of 9.8 months prior to PNP confirmation. In all cases, the diagnosis of PNP was based on clinical, serological and tissue-based investigation results, including persistent, treatment-refractory oral/extra-oral mucocutaneous ulceration, positive serological findings by indirect immunofluorescence (IIF) using rat bladder substrate, and characteristic findings by direct immunofluorescence (DIF) of tissue specimens. Of note, histopathology for each case was non-specific; this partly may have contributed to delayed PNP diagnosis, particularly for those with the initial diagnosis of LP.
Following appropriate therapy of the associated malignancy and PNP, Patients 1-4 (  remains on 1 mg of prednisolone and IVIG for secondary hypogammaglobulinemia.

| Red flags for PNP
Our cases presented similarly with a distinctive patten of treatment-refractory oral mucosal inflammation and ulceration localised to the latero-dorsal, tongue and lips ( Figure 1), with or without extra-oral mucocutaneous involvement. The majority had received an alternate diagnosis prior to our tertiary care. Although clinical features may overlap with autoimmune bullous diseases and inflammatory dermatoses, therapy resistance (characteristic in all five cases) and polymorphic cutaneous eruptions (seen in Patient 4) are features that may portend PNP, heightening the index of suspicion in our cases. 3 The pathogenic autoantibodies that underpin PNP are typically directed against plakin family proteins, as well as α2-macroglobulin-like antigen-I, BP180, p200 protein, desmogleins 1 and 3, and desmocollins. 4 IIF assays utilizing rat bladder substrate, which is rich in plakin proteins, demonstrates staining of the transitional epithelium (Figure 2a) and has a sensitivity of 86% and specificity of 98% for anti-plakin antibody-associated PNP; these were positive in all five of our cases. 5 Immunoblot assays utilizing epidermal extracts of envoplakin and periplakin have demonstrated similarly high sensitivities and specificities, however performance characteristics of commercial blots may be laboratorydependent. 5 ELISAs for anti-envoplakin and antiperiplakin antibodies have demonstrated lower sensitivity and specificity than IIF and immunoblot assays. 6 These factors should be considered during investigation.
Histopathology varies with lesion morphologyblisters or erosions may show suprabasal acantholysis and detachment, and inflammatory maculopapular lesions may show dyskeratotic or necrotic keratinocytes, vacuolar interface dermatitis or lichenoid infiltrates. 5 DIF findings are key to the diagnosis. There is often overlap with PV and LP, with the combination of IgG and C3 complement deposition limited to the basal epidermal intercellular surface (ICS), and fibrinogen deposition along the basement membrane zone (BMZ) (Figure 2b-d).

| Malignancy screening & treatment
Malignancies may be occult and tend to be haematological, mostly B-cell neoplasms or Castleman's disease, as was in our five cases. 7 Therefore, findings consistent with PNP should prompt screening with a CT and flow cytometry of peripheral blood as a starting point. A minority of patients may not have an underlying malignancy, however should be closely monitored for its subsequent development. 7 Treatment of benign tumours or localised Castleman's disease can lead to disease remission following tumour resection. 8 Whilst all our patients achieved PNP remission following treatment of their malignancy, this may not always result. A combination of treatments that target humoral and cellular immunity may be appropriate, although this may depend on the underlying malignancy.

| Bronchiolitis obliterans
PNP is associated with a poor prognosis, which is not only a reflection of malignancy or immunosuppression, but also BO, which may occur in up to 30% of PNP patients, 1 has a mortality rate of at least 40%, 7 and therefore must be screened for in all patients. It may present insidiously with dyspnoea or cough, an obstructive defect on pulmonary function testing and air trapping on CT, 1 and its irreversible course may progress independent of the malignancy or PNP. Although its pathogenesis is not well defined, it is thought that the autoantibodies target certain plakin proteins expressed in respiratory epithelia, 9 which was likely the case in Patient 5, who was seropositive for anti-envoplakin and anti-periplakin antibodies. She was managed initially with plasma exchange -for the putative role of the autoantibodies in the disease pathophysiology -systemic corticosteroids and maintenance immunosuppression. Although BO confers a poor prognosis and is resistant to immunosuppressive medications, 8,10 our case experienced stabilization over a prolonged duration following treatment, which may have been the outcome of diagnostic vigilance and early aggressive management.

| CONCLUSION
Good clinicopathologic correlation is key to the timely diagnosis of PNP (Figure 3), supported by the detection of PNP antibodies on testing that includes IIF on rat F I G U R E 3 Approach to paraneoplastic pemphigus EMERSON ET AL. bladder substrate in preference to other serological assays. 11 Tissue histopathology and DIF definitively establishes the diagnosis of PNP, and should prompt further investigation for an underlying neoplasm, even in the absence of suggestive symptoms. BO is a lifethreatening complication of PNP, and follows a course independent of the malignancy and mucocutaneous disease; judicious disease monitoring for its occurrence should be considered even in the absence of symptoms. Treatment of the malignancy may lead to remission of the PNP without the need for persistent immunosuppression.
As demonstrated in this small series, early clinical suspicion, extensive imaging to investigate for often clinically silent malignancy, intensive multidisciplinary care, awareness for and management of BO, if present, resulted in impressive long-term survival when compared with recent data of less than 50% survival at 6 years 12