Hemophagocytic lymphohistiocytosis secondary to pembrolizumab treatment with insufficient response to high-dose steroids

Fulminant steroid-resistant haemophagocytic lymphohistiocytosis secondary to pembrolizumab treated successfully by addition of tacrolimus. Sir, Haemophagocytic lymphohistiocytosis (HLH), which in context of predisposing autoimmune disease is referred to as ‘macrophage activation syndrome’, is a rare but potentially fatal condition characterized by an acute overwhelming systemic inflammatory response [1]. The clinical hallmarks of this syndrome, namely high fever, splenomegaly, bi-cytopenia, hypertriglyceridaemia, high serum ferritin >3000 ng/ml and sIL-2-Receptor >2400 U/l, and proof of haemophagocytosis in bone marrow specimens, were included in the 2004 diagnostic criteria for primary HLH, which in contrast to secondary HLH, route back to monogenetic conditions and occur mainly in childhood and adolescence. Nevertheless, in principle, these criteria also appear valid for secondary HLH, as they basically aggregate the syndrome’s underlying pathophysiology: aberrant macrophage activation, polyclonal T cell activation and a consequent systemic cytokine storm [2, 3]. Herein, we report the case of a 68-year-old Caucasian male patient suffering from castration resistant prostatic cancer with bone-, lymph node- and lung metastases. He was started on the anti-PD-1 humanized monoclonal antibody pembrolizumab in combination with androgen deprivation therapy including gonadotropin-releasing hormone analoga and enzalutamide, when the patient presented with antibiotics-resistant fever, hepatosplenomegaly, progressive panzytopenia, hyperferritinaemia (73 662 ng/ml), elevated serum sIL-2-Receptor (20098 U/ml), fasting hypertriglyceridaemia (281 mg/dl), hypofibrinogenaemia (151 mg/dl) and elevated liver enzymes (Fig. 1A). Abdominal ultrasound reported splenomegaly (maximum dimension 13.2 cm). At the time of treatment initiation, the patient met 6/8 HLH criteria: the NK-cell-cytotoxicity assay was not available at our facility. The patient’s cumulative H-score was 288, which left us with a likelihood of 99.9% for presence of HLH [4]. In line with this, bone marrow biopsy reported haemophagocytosis next to massive polyclonal T cell infiltrates (Fig. 1B). Of note, the patient was negative for EBV-virus PCR and positive for EBNA-IgG. Extensive infectious work-up remained without positive results. Time course of the disease, laboratory parameters and treatment (A) Time course of the patient’s body temperature (red dots), total leucocyte count (black circle), haemoglobin (red circles), platelet counts (cyan coloured circles), serum glutamic oxaloacetic transaminase (black dots) and systemic tacrolimus levels (green squares); under the abscissa the immunosuppressive therapy regimes are shown in temporal relation to the latter; (B) Bone marrow biopsy findings, HE-, CD3-, CD68-staining (from top to bottom), before and after the treatment protocol (from left to right). BT: body temperature; GC: glucocorticoids; GOT: serum glutamic oxaloacetic transaminase; Hgb: haemoglobin; TAC: tacrolimus levels; TLC: total leucocyte count. Upon suspicion, the patient started on high dose steroids as well as daily plasmapheresis treatment, which temporarily controlled the fever and slightly improved the patient’s general condition, whereas pancytopenia worsened (Fig. 1A). Because addition of Ciclosporin A, from which insufficient trough levels were built up, had no beneficial effects, he received etoposide 250 mg i.v. twice. However, only the addition of tacrolimus (target level 7–10 ng/ml) induced a significant treatment success and improved pancytopenia along with resolution of haemophagocytosis and reduction of T cell infiltrates in serial bone marrow biopsy. In addition, this second biopsy reported reconstitution of haematopoietic precursors (Fig. 1B). The patient was discharged on tacrolimus monotherapy, which was tapered over the course of 7 months. At the time of discharge, the tumour assessment by CT and bone scan revealed a complete response of all metastatic sites. Finally, nine months after termination of pembrolizumab treatment, the patient remains free of relapse and in good general health condition. From an oncologic point of view, a complete response in castration resistant prostate cancer patients with visceral metastases is rare under standard androgen deprivation therapy, assuming the checkpoint-inhibitor therapy with pembrolizumab has triggered a high antitumor activity. The timing, clinical course and the polyclonal T cell expansion observed in bone specimens are highly suggestive for assuming another case of pembrolizumab triggered secondary HLH in a cancer patient [5]. It is not the first HLH secondary to PD-1-check-point-inhibitor-therapy or other ‘T cell-reenergizing’ monoclonal antibodies such as anti-CTLA-4 [5–7]. With the increasing use of these potent immunomodulators in cancer therapy, it will certainly not be the last and will prompt diagnosis plus immunosuppression upon suspicion remaining key to treatment-success. Whereas a previous report of secondary HLH associated with PD-1-check-point inhibitor therapy hypothesized a generally good steroid response [6, 8], our experience indicates insufficiency of sole high dose steroid and plasmapheresis treatment. In contrast to another pembrolizumab-associated HLH reports [7], our case occurred independent of EBV-co-infection, and only the addition of T-cell-targeted, tacrolimus-based immunosuppression provided for short-term success in treatment protocol. High-dose steroid and plasmapheresis treatment were able to dampen the ‘systemic cytokine storm’ and to temporarily control fever flares, while pancytopenia and haemophagocytosis progressed. A genetically particularly unfavourable constellation, or the hormone ablative co-medication could have been partially responsible for this steroid-refractory course [1, 8]. The patient agreed to the scientific evaluation and publication of his medical history and his clinical data. Funding: No specific funding was received to carry out the work described in this manuscript. Disclosure statement: The authors have declared no conflicts of interest.