Haemophagocytic lymphohistiocytosis has variable time to onset following CD19 chimeric antigen receptor T cell therapy

Two CD19 chimeric antigen receptor (CAR) T cell therapies are now approved by the US Food and Drug Administration for the treatment of relapsed or refractory (R/R) diffuse large B-cell lymphoma (DLBCL) and histological variants: axicabtagene ciloleucel (axi-cel) and tisagenlecleucel (tisa-cel) (Neelapu et al, 2017; Schuster et al, 2019). Common toxicities of CAR T therapy include cytokine release syndrome (CRS) and neurological toxicity. CRS is initiated by cytokine release from activated T cells and is reminiscent of haemophagocytic lymphohistiocytosis/ macrophage activation syndrome (HLH/MAS) by elevated levels of interleukin 6 (IL6), interferon-γ (IFN-γ) and macrophage activation. However, CRS is almost always reversible, and some have suggested a clinical distinction of fulminant HLH where, in 1% of cases, CRS is refractory to IL6 receptor blockade (i.e. tocilizumab) and steroids (Neelapu et al, 2018a). Traditional diagnostic criteria for HLH are non-specific and are often met during CAR T therapy with CRS that is not treatment-refractory, fulminant or associated with haemophagocytosis. Therefore, it is unclear how to incorporate the diagnostic criteria of HLH into CAR T toxicity management. Here, we describe two cases of CAR T toxicity with features of HLH, including biopsy-proven haemophagocytosis. We propose that the timing of HLH onset is an important distinguishing feature for CAR T cell-related CRS and HLH and may direct different management strategies. Patient 1 was a 64-year-old male diagnosed with stage IVBE DLBCL via liver biopsy, who had primary refractory disease following 8 cycles of R-CHOP (rituximab, cyclophosphamide, doxorubicin, vincristine, prednisolone). After failing to respond to second line treatment with R-ICE (rituximab, ifosfamide, carboplatin, etoposide), he proceeded to apheresis for axi-cel manufacture. He received lymphodepleting cyclophosphamide and fludarabine on days −5 to −3. On the day of CAR T infusion, the patient had ferritin and C-reactive protein (CRP) levels of 2744 µg/l and 111 mg/l, respectively (Fig 1). Continuous fevers, attributed to CRS, occurred 16 h post-infusion and tocilizumab plus dexamethasone 10 mg IV were given on day +2, after which he became afebrile. On day +5 the patient became obtunded and dexamethasone was resumed. Ferritin increased from 3558 µg/l on day +5 to 8667 µg/l on day +7. On day +6 soluble IL2R was highly elevated at 59 600 pg/ml and other HLH criteria were positive, e.g., triglycerides 3·63 mmol/l, white blood cell count 0·16 × 109/l, platelet count 27 × 109/l. Day +6 bone marrow biopsy showed hypocellularity with evidence of haemophagocytosis (Fig 2A,B). No additional therapy was given for haemophagocytosis given because he maintained stable organ function. Dexamethasone was tapered slowly starting on day +13 and the patient showed dramatic improvement in mental status on day +26. On day +136, positron emission tomography/computed tomography showed a complete metabolic response. Patient 2 was a 64-year-old female who presented with stage IIA DLBCL of the breast. She received 6 cycles of R–CHOP but relapsed 2 months later. She proceeded through multiple lines of therapy with no response, including R-GDP (rituximab, gemcitabine, dexamethasone, cisplatin), CC-122 (avadomide) and rituximab, lenalidomide and rituximab and, finally, R-ICE. She proceeded to apheresis for axi-cel manufacture. On day – 1 after lymphodepleting fludarabine and cyclophosphamide, her baseline ferritin was 1364 µg/l and CRP was 73·6 mg/l (Fig 1). On Day +2 she developed CRS grade 1 that resolved within 3 days with supportive management. Grade 1 neurotoxicity was noted beginning on day +6 and progressed over time to grade 4, which improved to grade 1 after pulse doses of methylprednisolone. She was subsequently noted to have rising ferritin, increased liver enzymes, pancytopenia developed and she became persistently obtunded from day +25. On day +29, bone marrow biopsy showed increased histiocytes with haemophagocytosis (Fig 2C,D). Soluble IL2R was within normal limits at 989 pg/ml, ferritin was elevated at 14 000 µg/l, and triglyceride level was normal at 1·06 mmol/l. Given the association of HLH with neurotoxicity and pancytopenia, she was started on etoposide 50 mg/m2 q3 days and dexamethasone 20 mg IV twice a day with little improvement. The patient became asystolic on day +37. Autopsy demonstrated a severely hypocellular bone marrow with stromal histiocytes containing ingested erythrocyte debris, consistent with the previously demonstrated HLH. There was no evidence of residual lymphoma. The macrophage activation syndrome in both CRS and HLH have similar characteristics (Lee et al, 2014). Most severe CRS responds to anti-IL6 (i.e. tocilizumab) and steroids. In the ZUMA-1 trial of axi-cel in DLBCL, 2 of 108 patients died of CRS and only one was described as a case of CAR T-associated HLH (Neelapu et al, 2017). In B-cell acute lymphoblastic leukaemia (B-ALL), HLH features have been described with severe CRS, with symptoms beginning early after CAR T infusion (median day 3), and typically recover with steroids, tocilizumab and time (Maude et al, 2018). Clinically, it remains unclear as to when the HLH features of CRS are significant enough to consider treatment with therapies used for non-CAR T-associated HLH, such as etoposide or methotrexate (Neelapu et al, 2018b). Recent consensus guidelines have removed HLH/MAS from the definition of CRS given the uncertainty on how to separate the two entities from a treatment perspective (Lee et al, 2019). Our first case, although associated with a high soluble IL2R level and biopsy-proven haemophagocytosis, occurred early after CAR T and was successfully treated with steroids and tocilizumab. Importantly, it is unknown how many patients with severe CRS, either in lymphoma or in B-ALL, would have demonstrable haemophagocytosis on bone marrow if done early after therapy. The second case had a later onset after resolution of CRS, low soluble IL2R levels and haemophagocytosis on biopsy, and demonstrated a second delayed IL6 peak after the initial CRS improved. This patient died of HLH while in lymphoma remission despite treatment with tocilizumab, steroids and etoposide. While CRS typically occurs early after CAR T therapy, CAR T-associated neurotoxicity may be delayed in timing and is associated with macrophage activation (Gust et al, 2017). It is possible that delayed CAR T-associated HLH represents an extreme case of CAR T-associated neurotoxicity, although HLH features, such as a late rising ferritin levels and second IL6 peak have not been observed in neurotoxicity case series outside of this case (Gust et al, 2017; Faramand et al, 2018). Our case series suggests that the timing of onset of CAR T-associated HLH may differ between patients, with concern when delayed or recurrent HLH symptoms occur with a rising ferritin level, which is when HLH-directed therapies should be considered. M.D.J. and F.L.L. designed the research and analysed results. H.K. and M.L.D. designed and performed cytokine assays. M.D.J., G.K., M.H. and J.S. created the figures. C.B., J.C.C. and T.N. provided patient information. H.H., C.B., F.L.L. and M.D. J. wrote the manuscript. Hamza Hashmi: none. Christina Bachmeier: Advisory for Kite/Gilead. Julio C. Chavez: Advisory for Kite/Gilead, Novartis, Bayer, Genetech; Speaker Bureau for Genetech. Jinming Song: none. Mohammad Hussaini: none. Gabriel Krivenko: none. Taiga Nishihori: none. Hiroshi Kotani: none. Marco L Davila: Research funding from Celgene, Novartis, Atara; Other financial support from Novartis, Precision Biosciences, Celyad, Bellicum, GlaxoSmithKline; Stock options from Precision Biosciences, Adaptive Biotechnologies, Anixa Biosciences. Frederick L. Locke: Consultant for Cellular Biomedicine Group, Inc.; Scientific Advisor for Kite/Gilead, Novartis. Michael D. Jain: Consultant for Kite/Gilead.