Rare and fatal complication of immune checkpoint inhibition: a case report of haemophagocytic lymphohistiocytosis with severe lichenoid dermatitis
Sara Choi, Maggie Zhou, Eman Bahrani, Beth A. Martin, Kristen N. Ganjoo, Lisa C. Zaba
Abstract
Anti-programmed cell death 1 (PD-1) therapy is increasingly considered for therapy of refractory Kaposi sarcoma (KS), a human herpesvirus-8 (HHV-8)-associated neoplasm. Haemophagocytic lymphohistiocytosis (HLH) is a rare, life-threatening immune-related adverse event (irAE) of immune checkpoint inhibitors (ICI).1 HLH in adults usually develops in a stepwise fashion beginning with the presence of an underlying predisposition, usually lymphoid malignancy, with a subsequent pro-inflammatory trigger. Common infectious triggers include members of herpesviridae, Epstein–Barr virus (EBV) and HHV-8. ICI and other T-cell activating therapies are also increasingly recognised as HLH triggers.2 We report a case of lethal ICI- and HHV-8-associated HLH in a patient with KS who first presented with severe cutaneous irAE. An 85-year-old man, who was human immunodeficiency virus (HIV) negative, was treated with nine cycles of monthly anti-PD-1 therapy, nivolumab, for paclitaxel and everolimus-resistant, skin-limited classic KS. Approximately 8 months after nivolumab initiation, he developed a Common Terminology Criteria for Adverse Events (CTCAE) Grade 1 pruritic truncal rash (Fig 1A). Skin biopsy of a pink papule on the back revealed lichenoid dermatitis (Fig 2A). Direct immunofluorescence (DIF) pattern did not support autoimmune bullous disease. Topical steroids were prescribed for suspected ICI-induced lichenoid eruption. His clinical course is outlined in Figure S1. Over several weeks, his rash progressed to CTCAE Grade 2 with violaceous papules and erosions on the trunk, abdomen, extremities and genitals; active bullae were noted on his hands and lower extremities (Fig 1B). Oral prednisone (1 mg/kg), empiric doxycycline, and atovaquone prophylaxis were initiated. Nivolumab was stopped. After several weeks of improvement, the rash then flared as prednisone was tapered to 20 mg daily. Eosinophils were within the normal range. Skin biopsies revealed worsening lichenoid dermatitis with incipient vesiculation (Fig 2B). DIF did not support autoimmune blistering disease. At 1 week later, he developed jaundice and elevated transaminases [aspartate transaminase (AST) 249 u/l (normal range 10–50 u/l); alanine transaminase (ALT) 919 u/l (normal range 10–50 u/l; alkaline phosphatase 181 u/l (normal range 40–130 u/l); total bilirubin 12 mg/dl (normal value <1·2 mg/dl)]. Acute viral hepatitis, including hepatitis E, and quantitative polymerase chain reaction for EBV/cytomegalovirus (CMV) reactivations were negative. HHV-8 was elevated at 8059 copies/ml (normal value <1000 copies/ml). Bullae were present in the oropharynx, nasopharynx, and ears, along with CTCAE Grade 3–4 desquamation and dusky, diffuse violaceous erythema on the abdomen, back, thighs, arms, hands, feet, cutaneous lips, buccal mucosa and soft palate (Fig 1C,D). Oral lesions were negative for herpes simplex virus/varicella-zoster virus. Skin biopsy showed pauci-inflammatory, near-full-thickness epidermal necrosis with separation of the epidermis from the dermis, suggesting Steven–Johnson syndrome/toxic epidermal necrolysis (SJS/TEN)-like eruption secondary to immunotherapy (Fig 2C). Treatment with methylprednisolone sodium succinate (Solu-Medrol; 1 mg/kg/day then 1·25 mg/kg/day) and intravenous immunoglobulin (2 g/kg/day) was initiated on hospital day 4. Liver biopsy revealed multifocal, non-zonal hepatocyte necrosis with a prominent periportal inflammatory infiltrate, suggestive of drug reaction, including immunotherapy, and/or HLH. Immunohistochemistry for EBV, CMV, and HHV-8 were negative. Over his 3-week hospitalisation, his skin irAE improved to a Grade 2 rash with most erosions having re-epithelialised; however, sudden pancytopenia with prominent absolute neutropenia, unresponsive to filgrastim, developed. Bone marrow biopsy found extensive nucleated erythrophagocytosis without EBV. He was diagnosed with HLH based on: fever at time of elevated transaminases, tri-lineage cytopenias, hypertriglyceridaemia, haemophagocytosis, and hyperferritinaemia (H-Score: 218 points, 93–96% probability of HLH) (HLH-2004)3 (Table SI). Persistently neutropenic and febrile, blood cultures detected viridans streptococci, Enterococcus faecalis, Clostridium perfringens, and Streptococcous gordonii. Imaging revealed bilateral pulmonary nodules suggestive of angioinvasive fungus. Despite broad spectrum antibiotics with anti-fungal therapy, the patient precipitously declined and died on hospital day 22, which was 135 days after his first presentation with rash. We report a patient with KS who had multiple, fatal complications of anti-PD-1 therapy that presented as lichenoid dermatitis progressing to severe SJS/TEN-like eruption, with concurrent HLH. Notably, the increased HHV-8 viraemia from baseline in the setting of immunosuppression likely contributed to prominent hyperinflammation. Cutaneous reactions are the most common irAE, occurring in 17–40% of patients receiving PD-1/programmed death-ligand 1 (PD-L1) inhibitors;4 although rare, fatal cases of TEN from nivolumab have been reported.5, 6 PD-1/PD-L interactions are important for maintenance of peripheral tolerance. PD-L1 expression on keratinocytes in mouse models has been shown to regulate cluster of differentiation 8 (CD8)+ T-cell activity and suppress autoimmunity, while PD-1 knockout mice develop lupus-like autoimmune skin pathology.7, 8 Immunohistochemistry in patients receiving PD-1/PD-L1 inhibitors demonstrated increased CD8+ epidermal exocytosis and increased PD-L1 keratinocyte expression, similar to histopathology seen in patients with SJS/TEN, suggesting overlapping mechanisms of skin autoimmunity.6 Vivar et al.5 confirmed increased PD-L1 expression in their report of nivolumab-induced TEN. Mechanisms of ICI-related HLH and hepatotoxicity are also not well understood and may differ from PD-L1 upregulation seen in the skin. Haematological irAEs have a median onset of 16 weeks, and are more common with anti-PD1/PD-L1 (4·1%, 4·7%) than anti-cytotoxic T-lymphocyte antigen 4 (CTLA4; 0·5%) therapy.1 Overall, the haematological irAE mortality rate is 14%, with HLH being most life-threatening. Skin findings are not part of HLH criteria, but HLH presenting with desquamation (e.g. SJS/TEN),9 morbilliform eruption, and targetoid lesions are reported.10 An immunological link between HLH and SJS/TEN has been hypothesised given that both are mediated by interferon-γ and dysregulated CD8+ T-cell activation. HLH co-occurring with severe cutaneous irAE presents diagnostic challenges. Both HLH and SJS/TEN are triggered by drugs or infection, and occur in the setting of multi-organ failure. Additionally, in KS, important mimickers of HLH include KS herpesvirus (KSHV)-associated inflammatory cytokine syndrome (KICS) and multicentric Castleman disease from cytokine overproduction; these rare KS-related disorders are mostly reported in patients with HIV. In favour of HLH, our patient met HLH criteria, was HIV negative, and had skin-limited KS without lymphadenopathy, splenomegaly, or HHV-8 in the liver. Expanding indications for ICI use continue to reveal the spectrum of potentially fatal irAEs, including HLH and SJS/TEN-like eruptions. Importantly, in patients with KS being treated with ICI, potentially fatal reactions must be considered in cases of increasing HHV-8 viraemia. Oral steroids are first-line therapies for irAEs; however, there is a theoretical risk that immunosuppression may worsen HHV-8 viraemia, leading to catastrophic immune activation. Further investigation is needed to elucidate the risk of ICI in this patient population. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.