Chimeric antigen receptor (CAR) T cell therapy has made an impact on the treatment of hematological malignancies. In the United States of America, since August 2017, six commercial CAR-T cell products were developed and approved by the Food and Drug Administration (FDA). The indications include acute lymphoblastic leukemia, several types of non-Hodgkin lymphomas and multiple myeloma. They come with a unique set of toxicities, of which, cytokine release syndrome (CRS) and immune effector cell-associated neurotoxicity syndrome (ICANS) are best known.¹ A lesser known but very serious side effect that is increasingly recognized is severe toxicity resembling hemophagocytic lymphohistiocytosis (HLH), termed as CAR-HLH. It is debated whether CAR-HLH is a separate entity or an extended or delayed inflammatory phase of CRS.^{2, 3} Limited data indicate that CAR-HLH may be refractory to cytokine neutralization with tocilizumab.⁴ Early studies reported an incidence of 1% for CAR-HLH⁵ while a survey from EBMT centers in 2020 indicated an incidence of 3.5%.³ Except for two case reports of patients developing HLH after the infusion of CAR-T cells, no additional data on this potentially lethal toxicity were published.^{6, 7} As far as experimental, non-FDA approved CAR T cells are concerned, a significantly higher incidence was recently observed with a CD22 product.⁸ Therefore, we decided to access the FDA and Vizient databases to get further insight into trends, incidence, risk factors, and potential outcomes of CAR-HLH.

The FDA Adverse Events Reporting System (FAERS) was last accessed on March 3, 2022, to collect data on the reported cases of CAR-HLH. The search terms included all currently approved CAR-T cell products including axicabtagene ciloleucel (axi-cel), brexucabtagene autoleucel (brexu-cel), lisocabtagene maraleucel (liso-cel), tisagenlecleucel (tisa-cel), idecabtagene vicleucel (ide-cel), and ciltacabtagene autoleucel (cilta-cel). Familial HLH cases were excluded. Four cases were eliminated because of likely duplicate reporting.

We identified and manually reviewed 121 cases of CAR-HLH. The data were analyzed with regards to the associated CAR-T cell product, country, year, and mortality. To further understand the incidence of CAR-HLH cases, we further searched for secondary HLH (listed as “Overall Adverse Events” in FAERS over the last 10 years). Since the diagnostic criteria for CAR-HLH depended on each submitting institution, no further review was possible. The trend of secondary HLH cases is shown in Table 1. As evident from this table, secondary HLH cases are increasing likely due to improved awareness and more aggressive treatments for the underlying malignancy. CAR-HLH shows a similar trend as more CAR-T cell products are approved and the awareness for CAR-HLH has increased. World-wide, a total of 6034 cases were reported in FAERS (see Table 2). This results in an overall incidence of approximately 2% (varying between 1.6% and 2.8%). According to Fisher's exact test, no statistical difference exists between the different CAR T cell products (data not shown). In more than two thirds of cases, the outcome was death. Since there was no chart review of the FDA database, it is uncertain to what extent HLH contributed to the death of patients. Another subset of HLH is an overlapping life-threatening clinical entity called macrophage activation syndrome (MAS) which is difficult to differentiate from CRS. We found seven cases between 2017 and 2021 (not included in the present CAR-HLH study). Since FAERS is a voluntary data base, it may suffer from both underreporting or over-reporting (if not all criteria for CAR-HLH are met).

In order to get an independent verification of CAR-HLH, we queried the Vizient database. The Vizient® Clinical Data Base is a comparative database that contains discharge and line-item, patient-level detail data from more than 800 Vizient members, including over 97% of United States academic medical centers. Data were compiled for 3771 CAR-T infusion encounters in 81 hospitals across the United States from October 2017 through December 2021. CAR-T infusion encounters were identified using ICD-10-PCS, CPT, and charge codes, and HLH encounters using ICD-10-CM codes. Only the first infusion for all patients was included. Aggregations were grouped by product as defined primarily by charge code and procedure codes where no charge code was available. HLH cases were defined as a diagnosis of HLH during the initial encounter or during a subsequent encounter that occurred within 60 days of the infusion. A data sharing agreement using anonymized data was signed between the University of Kentucky and Vizient. As shown in Table 3, both the number of patients treated with CAR-T cells and diagnosis of CAR-HLH increased between 2017 and 2021; however, the relative incidence remained between 1.0% and 1.55%. As far as individual products are concerned, the risk with axi-cel is slightly higher than with tisa-cel (1.57% vs. 1.06%, not statistically different according to Fisher's exact test, p = .22). The total number of CAR-HLH events in the Vizient database is lower than that reported in FAERS which may be related to HLH not being coded in discharge diagnoses (69 vs. 48 cases). Taken together, we show here from two independent databases that the incidence of CAR-HLH is somewhere between 1% and 2%.

HLH is a rare life-threatening emergency first identified in 1928. This clinical condition is characterized by immune dysregulation, lymphocyte, and macrophage hyperactivation leading to infiltration of lymphohistiocytic cells. It is known to occur either as familial or sporadic forms.⁹ Common clinical findings according to the HLH-2004 study are: fever, bicytopenia, splenomegaly, and hyperferritinemia. As mentioned, similar clinical manifestations have been seen in patients after CAR-T cell therapy and were designated as CAR-HLH. However, data on CAR-HLH is still evolving. With our series from the FDA database, we describe the largest series of cases of HLH world-wide (121 cases) and in the United States (69 cases). A strength of our study is that we show real-world data. A weakness is that no universally accepted or validated consensus criteria were used. The Vizient database relies on discharge diagnostic codes and captured only 48 cases. Overall, the usage of CAR-T cell products in the United States is in line with data published by the CIBMTR (website accessed May 29, 2022). We hope to increase the awareness of CAR-HLH as a potentially fatal complication of a potentially curative treatment for B cell leukemias and lymphomas. Two recently approved CAR-T cell products for multiple myeloma (ide-cel and cilta-cel) have a black box warning for CAR-HLH. The ASCO guideline for managing immune-related adverse events mentions CAR-HLH but does not offer grading or specific treatments.¹⁰ With better patient selection, delineation of risk factors and early treatment with steroids and other agents, the risk and mortality of CAR-HLH may decrease.