Primary HLH was initially described in the mid-1900s in a series of case reports (9). Infants would develop fevers, cytopenias, hepatosplenomegaly, and neurological features, and within weeks, succumb to disease. Post-mortem bone marrow analysis showed the characteristic finding of ‘hemophagocytosis’ – macrophages (histiocytes) phagocytosing blood cells. This syndrome eventually became known as HLH. The Histiocyte Society initially developed criteria for the diagnosis of HLH in children based on clinical and laboratory criteria for the HLH-94 trial (10). These criteria were subsequently refined for the HLH-2004 trial to include molecular and/or clinical criteria (11). Most recently, revised criteria were proposed in which functional assays in the setting of consistent clinical features can also be used to diagnose HLH ( Table 1 ) (12). Additional laboratory tests may support the diagnosis of HLH and be useful for disease activity monitoring. These include detection of the IFN-γ induced protein CXCL9 (since IFN-γ itself is difficult to detect in the bloodstream) (8, 13) as well as HLADR+CD38^hi CD8+ T cells, although the latter is not available as a certified laboratory test.

Primary HLH results from inborn errors of immunity (IEIs) and is estimated to describe 12% of cases of HLH in pediatrics (14). In 1999, perforin (PRF1) mutations were found to cause HLH in a subset of patients (15). Perforin is key to the cytotoxic function of CD8+ T cells and NK cells, enabling them to appropriately contract an inflammatory response. Soon thereafter, mutations in other genes (STX11, STXBP2, UNC13D) with related functions in the cytotoxic response were implicated in HLH, resulting in a nomenclature of familial HLH (FHL), where ‘familial’ implied inherited, and a corresponding number implied a specific genetic defect in granule mediated cytotoxicity (e.g., FHL2 is HLH due to a perforin defect). As genetic variants driving HLH can be de novo, and there is now a larger range of pathways implicated, it is now more common to use the term “primary HLH” and to specify the gene mutation (if known) (16) ( Table 2 ). For example, several IEIs that do not affect cytotoxic function but instead regulate the inflammasome (e.g. NLRC4, CDC42) have HLH as a common manifestation (17–19). In these disorders, overactive inflammasome signaling drives activation of caspase-1 and its cleavage products IL-1β and IL-18, which can promote a hyperinflammatory cascade (20). Lastly, IEIs that predispose to poor control of Epstein Barr virus (EBV) infection may increase the risk for HLH and will be discussed in a later section. There may also be an overlap, as XIAP deficiency exhibits both altered inflammasome regulation and increased susceptibility to EBV infection (21).

Early attempts to treat classic ‘familial’ HLH included adrenocorticotropic hormone, splenectomy, exchange transfusion, and steroids (1). There was also use of chemotherapeutic agents. One agent in particular – etoposide – was selected due to its use in monocytic leukemias suggesting a potential ability to subdue macrophage activation. Indeed, early studies with etoposide were highly promising (22). Ironically, the mechanism of its benefit was different than predicted. Etoposide is a topoisomerase II inhibitor, blocking repair of double stranded DNA breaks, triggering apoptotic cell death. This was shown to be true of activated, cytotoxic CD8+ T cells in murine models of HLH (23). Thus, etoposide eliminates a hyperactive cell population in HLH that would have otherwise been eliminated had there not been a genetic defect in cytotoxic function. In addition to etoposide, intrathecal methotrexate (anti-metabolite) was added to regimens for central nervous system (CNS) coverage, as there is CNS involvement in 30-73% of patients with primary HLH (24). As dexamethasone has better CNS penetrance than other forms of corticosteroids, it emerged as the steroid of choice. Calcineurin inhibitors – particularly cyclosporine – were added to protocols to block NFAT-mediated production of cytokines such as IL-2, TNF-α, and IFN-γ. Finally, allogeneic HCT was determined to be necessary for cure.

This constellation of therapies was the foundation of the HLH-94 protocol, used in a large international pediatric trial. Etoposide, dexamethasone, and intrathecal methotrexate were used as an 8-week induction regimen, followed by intermittent etoposide, dexamethasone, and twice daily cyclosporine as a bridge to HCT (25). In a long-term follow up study, 29% of patients died before HCT, and for those that underwent HCT, the 5-year survival was 66%. There was a trend for improved HCT outcomes in patients with better disease control at the time of HCT. Early deaths were reported in patients who developed jaundice, edema, and renal dysfunction, suggestive of veno-occlusive disease (VOD). Even with the 20-30% mortality rate pre- and post-HCT, outcomes in this study were significantly improved from historic regimens. As a follow up, the HLH-2004 study started cyclosporine during induction, but this did not improve outcomes, and there were concerns regarding the side effects of cyclosporine (26).

In the modern era, etoposide and dexamethasone still serve as the foundation of pre-HCT induction therapy in primary HLH (27). Due to the risk of myelosuppression (and associated infection risk) as well as secondary malignancy (albeit rare) with etoposide, alternative agents have been explored. For example, emapalumab (IFN-γ antagonist) is now FDA approved for patients with primary HLH who have had a refractory, recurrent, or progressive course or intolerance to conventional therapy (28). Other options include alemtuzumab, a monoclonal antibody against CD52 (targets T cells, B cells, and macrophages) and ruxolitinib (janus kinase (JAK)1/2 inhibitor) (29). In practice, these agents are not mutually exclusive with etoposide and can be used sequentially or in conjunction. For example, as will be discussed below, one of the major barriers to HCT for HLH is the high rate of graft failure, particularly with reduced intensity conditioning (RIC) regimens. This may be due to a detrimental effect of IFN-γ on donor stem cells that cannot compete with residual recipient cells in the bone marrow that are ‘better acclimated’ to high-level IFN-γ environments (30). Bridging emapalumab in patients previously treated with etoposide may address this challenge and improve donor chimerism (30).

Identifying the best conditioning regimen for HCT in HLH has been historically difficult. Initial transplants with busulfan (Bu)-based myeloablative conditioning (MAC) had high rates of fatal VOD, likely due to the hepatotoxicity of Bu in patients with pre-existing liver dysfunction from HLH. This motivated lower dose Bu or RIC regimens such as fludarabine/melphalan/alemtuzumab (Flu/Mel/Alem). While the toxicity of RIC regimens was lower, they had high rates of mixed chimerism and graft failure, with chimerism levels of at least 20-30% thought to be necessary to protect against HLH recurrence (31). A large study using the Center for International Blood and Marrow Transplant Research registry compared four different commonly used conditioning regimens, adjusted for year of transplant (32). These included Bu/Cyclophosphamide (Cy), Bu/Flu, Flu/Mel/Alem, and Flu/Mel/Alem/Thiotepa (TT). The primary outcome was event free survival (EFS) – i.e., survival without primary or secondary graft failure. While overall survival (OS) was similar between regimens, rates of graft failure were substantially higher in the Flu/Mel/Alem regimen, corresponding to the lowest EFS of 44% at 5 years. Addition of thiotepa (an alkylator) to this regimen substantially reduced graft failure, improved EFS, and was associated with a low rate of VOD, establishing Flu/Mel/Alem/TT as the currently recommended regimen.

As previously mentioned, CNS involvement occurs in a substantial fraction of patients with primary HLH, and in some forms of secondary HLH. Clinical symptoms may include seizures, altered mental status, and/or encephalopathy, with delayed or insufficient treatment risking permanent motor and cognitive deficits. Thus, early identification is crucial. As not all patients have clinical symptoms at onset, screening MRI and lumbar puncture are recommended (24). In addition to CNS inflammation being part of HLH, there is an entity of CNS-restricted HLH, often driven by similar pathogenic variants but more difficult to diagnose given the lack of systemic features and inflammatory markers (33). Compared to the classic systemic form of primary HLH, the onset of CNS-restricted HLH occurs later – at a median age of 6.5 years in one report (33). The differential diagnosis includes CNS vasculitis, acute disseminated encephalomyelitis, acute necrotizing encephalopathy, or CNS involvement of an underlying condition. Diffuse white matter and/or cerebellar involvement is commonly seen on imaging, and pleocytosis, elevated protein, and elevated neopterin may be seen in the cerebrospinal fluid (CSF). There are no consensus guidelines on treatment of CNS-restricted HLH but dexamethasone with or without intrathecal methotrexate may be considered, as can etoposide and emapalumab. Regardless of medical treatment choice, the definitive treatment of CNS-restricted HLH remains HCT, ideally as soon as feasible to avoid irreversible neurologic damage (34, 35).

Therapeutic approaches for primary HLH, including first line and HCT regimens, are likely to continue to evolve, as will supportive care measures. Our group has recently considered pre-emptive (prior to HCT) tracheostomy for small patients considered to be at high risk for respiratory complications. While not part of any established guidelines, this has been beneficial for respiratory support through the acute transplant period. In parallel, ex vivo autologous hematopoietic stem cell-based gene therapy for genetically defined HLH is being developed. The first trial is opening in France for patients with UNC13D mutations (NCT06736080). While this approach has great potential, challenges include the harvesting of stem cells from critically ill patients, the requirement for therapy while the genetically manipulated product is manufactured, and uncertainty about which conditioning regimen will be required and tolerated to facilitate engraftment of the genetically manipulated product.

HLH secondary to rheumatic disease is referred to as macrophage activation syndrome (MAS), with an estimated 10% of cases of HLH/MAS attributable to this cause in children (14). By far, the rheumatic disease most associated with MAS in pediatrics is systemic juvenile idiopathic arthritis (sJIA), also known as Still’s disease. Systemic JIA presents uniquely from other subclasses of JIA with daily fevers, rash, lymphadenopathy, and potentially MAS [in up to 14% of cases (36)]. This systemic inflammation results from an autoinflammatory component to the disease pathophysiology, with disease-specific cytokine elevation, including IL-1 (37), IL-18 (38), and IL-6 (39). While arthritis is part of historic disease definitions (40), not all patients have arthritis at presentation, and with modern therapy, the development of arthritis may even be avoided (41). Patients are thus recognized as having sJIA based on other consistent disease features, including MAS, as well as sJIA-suggestive biomarkers described below.

The definition of MAS in sJIA is unique from primary HLH, with classification criteria shown in Table 3 (42). The high incidence of MAS is hypothesized to result from elevations in IL-18 during active disease (43), which augments IFN-γ-production by cycling lymphocytes (6). These include HLADR+CD38^hi CD8+ T cells as well as CD4+ cells and NK cells, which have been found to be highly glycolytic due to activation of mTORC1 signaling. mTORC1 is a metabolic sensor that can drive cell proliferation and determine function. This mTORC1 signaling was also found in activated monocytes in murine models of sJIA and MAS, potentially coordinating the inflammatory response (44). In addition to IL-18 and IFN-γ signaling in sJIA-MAS, an elevated type I interferon signature has been described that may further prime pathogenic T and NK cells and augment IL-18 production (6). As type I interferons are commonly induced by viral infection, this may explain infection as a trigger of sJIA-MAS. Lastly, it has also been observed that heterozygous variants in primary HLH related genes are enriched in patients with sJIA who develop MAS (45) ( Figure 1 ).

The American College of Rheumatology recommendation (46) for first line treatment of sJIA is IL-1 inhibition (anakinra, canakinumab) or IL-6 inhibition (tocilizumab) (47–50). Use of JAK inhibitors in more severe cases, is increasingly common (51, 52). As plasma IL-1 levels are not reflective of IL-1 secretion, due to greater production in the tissues and binding to a natural decoy receptor in the blood (53), IL-18 – a related cytokine – is instead monitored. In active sJIA, IL-18 levels usually range from 10⁴ to 10⁵ pg/mL, which can help identify sJIA from other febrile disorders (38). During episodes of MAS, IL-18 levels >10⁵ pg/mL are often observed, and IFN-γ related biomarkers (e.g. CXCL9) become significantly elevated (39).

Unlike in primary HLH, there are few dedicated trials or consensus guidelines for treatment of MAS (14). To create a framework for approaching MAS/HLH diagnosis and treatment, our center previously developed an Evidence Based Guideline (EBG) based on expert participation from diverse specialties at Boston Children’s Hospital, using a consensus-based process (54, 55). This led to treatment recommendations that include consideration of anakinra, intravenous immune globulin, calcineurin inhibitors, and high dose steroids for MAS. Since the implementation of this EBG, there has been updated evidence for use of JAK inhibitors (for other etiologies of HLH) (29) and emapalumab (recently FDA approved for MAS) (56), and the EBG has been updated accordingly. Use of etoposide, although described and beneficial for some patients with sJIA-MAS, is generally not recommended due to associated myelosuppression but could be considered in collaboration with Oncologists for patients with highly refractory disease (57, 58). A 2025 review on treatment evidence for MAS is provided (57). Importantly, while systemic reviews and single center guidelines are important reference tools, centers should adapt recommendations to local resources and patient specific needs. For example, general dosing guidelines may not be appropriate for patients with significant organ dysfunction or in the setting of potential drug-drug interactions (e.g. liver toxicity and/or antifungal prophylaxis with ‘azoles’ for ruxolitinib). Infection prophylaxis should be considered but is not detailed in this review. Lastly, with new advances in the field, guidelines will eventually become outdated.

Beyond treating individual episodes of MAS, there is also a history of HCT for sJIA, although transplant strategies and indications have evolved over time. Initially, in the late 1990s when there were no sJIA targeted biologics, autologous HCT was explored for patients with refractory arthritis. The premise was that high dose chemotherapy (usually ATG and Cy ± Flu or low-dose total body irradiation) could reset the immune system, and pre-collected hematopoietic stem cells would then be infused to restore hematopoiesis with subsequent production of naïve, clonally diverse T cells in the thymus. Intrinsic to this approach was ex vivo T cell depletion of the stem cell graft to avoid reinfusing dysregulated T cells. In a multicenter cohort of 34 patients with arthritis unresponsive to methotrexate and at least one other therapy, 53% of patients had a complete response, 18% a partial response, and 21% had no response by an average of 29 months of follow up (59). Prolonged depression of CD4+ T cells was common after transplant. Three patients developed fatal MAS triggered by infection, prompting an international consensus to reduce the degree of T cell depletion and improve disease control ahead of transplantation (60). Given the transplant related mortality of 9% and the emergence of IL-1 and IL-6 targeted biologics by the mid-2000s, autologous HCT became uncommon.

With new biologic agents curtailing the development of chronic arthritis, morbidity from sJIA transitioned from arthritis to recurrent MAS and potentially fatal lung disease (sJIA-LD), comprised of interstitial involvement and/or pulmonary hypertension (61). One of the earliest reports of sJIA-LD had a 5-year survival of only 42% (61). Due to increased screening, patients with less severe lung disease are now being identified, improving survival statistics (62). Risk factors for lung disease include young age of sJIA onset, HLA-DRB1*15 positivity, congenital heart disease, chronic IL-18 signaling, recurrent MAS, and trisomy 21 ( Figure 2 ) (63–66). The latter two factors have a causal explanation – an IFN-γ signature has been found in the bronchoalveolar lavage fluid of patients with lung disease suggesting a role in its pathogenesis (67). MAS is largely IFN-γ driven, and in trisomy 21, there is increased interferon signaling due to triplicate copies of a receptor for IFN-γ (IFNGR2) (68). Due to concern for disease relapse and MAS flare with autologous HCT, the field shifted to allogeneic HCT for recurrent MAS ± lung disease as the main transplant indications. There have been two retrospective cohort studies in the last 10 years. The first included 11 patients with treatment-refractory sJIA with 5/11 having a history of MAS (69). All patients received a reduced toxicity conditioning (RTC) regimen of Flu/Mel/Alem or Flu/Treosulfan/Alem. Viral infections after transplant were common and 1/11 patients died of fungemia in the setting of graft-vs-host-disease treatment. At last follow-up, 9/10 surviving patients were in complete remission (no symptoms, off immunosuppression), while one patient had a flare of arthritis and an MAS-like episode. The second report included 13 patients with sJIA-LD – 5 with an oxygen requirement prior to transplant (70). Patients received RIC or RTC regimens, most commonly Flu/Mel/Alem/TT (akin to that used in primary HLH). Four patients died – two from cytomegalovirus infection, one from intracranial hemorrhage during asymptomatic SAR-CoV2 infection, and one from progressive sJIA-LD. All 9 surviving patients were in complete remission off immune suppression at the time of last follow up, with previously oxygen-dependent patients coming off oxygen (70).

In summary, the pediatric rheumatic disease most associated with MAS is sJIA due to baseline activation of pathways that can trigger MAS-related inflammation. Medical management with small molecules and biologic agents may successfully control these episodes in some patients, but for those with recurrent MAS and/or lung disease, allogeneic HCT may be considered to halt a potentially fatal disease trajectory and achieve drug free remission. As transplant-related mortality remains 12.5-25% (higher for those with lung disease), careful patient selection is required, and there is currently no consensus on when this should be pursued.

Infections are the most common trigger of HLH in children, both in primary and secondary cases (57% of cases overall) (14). Viral, bacterial, parasitic, and fungal infections have been linked to HLH in both immune competent and immune compromised hosts. Work-up for suspected HLH should therefore include comprehensive evaluation in consultation with an infectious disease expert. As HLH is an uncommon complication of infection overall, a genetic evaluation should be strongly considered to look for predisposing pathogenic variants.

The most common trigger of infection associated HLH is EBV. EBV is a herpesvirus that primarily infects B cells through their CD21 and CD35 receptors and establishes latency. Initial EBV infection is asymptomatic in most individuals, and an estimated >90% of adults are latently infected (71). Clinical sequelae can range from ‘infectious mononucleosis’ (e.g., fever, sore throat, splenomegaly) to EBV-related malignancy or HLH.

A recent study by Liu et al. elucidates why EBV has a higher predilection for triggering HLH (72). In this study, serum and peripheral blood mononuclear cells were compared amongst pediatric patients with non-EBV viral illness, EBV infection, and known HLH. Relative to other viral illnesses, acute EBV infection led to increases in IL-18 and IL-27 (which contribute to Th1 cell polarization) as well as CXCL9. As with other etiologies of HLH, there was also expansion of HLADR+CD38^hi CD8+ T cells, which corresponded to the ‘atypical lymphocytes’ often observed in acute EBV infection (72). Overall, these data suggest that the response to acute EBV infection includes production of cytokines and populations of T cells that are hallmarks of HLH.

Treatment of infection-driven HLH involves targeting the underlying infection in addition to standard HLH directed therapies. For EBV driven HLH, addition of a B cell targeting agent like rituximab has been shown to facilitate elimination of the EBV reservoir (73). Anakinra is safe in the setting of infection (was initially investigated for patients with sepsis) (74, 75) but is not always sufficient to control EBV-HLH (76). In a single-arm phase II trial, 52 pediatric patients with a new diagnosis of HLH (diverse etiologies but malignancy excluded) received ruxolitinib as first-line therapy, with a complete response of 58.3% in the subset with EBV-HLH (29). Lastly, emapalumab may be used as an adjunctive agent to chemotherapy for cases that are particularly severe (77, 78). More specific guidelines for EBV-HLH will require prospective studies. However, one important consideration is whether a patient is at risk for recurrence.

Allogeneic HCT may be needed to resolve EBV viremia and its associated complications. For example, there are several IEIs (due to loss-of-function variants in XIAP, SAP, CD27, CD70, ITK, and MAGT1) associated with EBV-HLH and EBV-triggered lymphoma (79) that impair the ability of the immune system to clear EBV infection. There is also an entity known as Chronic Active EBV (CAEBV), which describes persistent EBV viremia without a known IEI. CAEBV has clear geographic associations and has mainly been described in East Asia and in Central America. As with some IEIs, a characteristic of CAEBV is that EBV infection is found in other lymphoid and myeloid lineages – thus not restricted to the B cell compartment ( Table 4 ). A 2024 study by Wang et al. provides a potential explanation (80). In this study, bone marrow and peripheral blood samples were obtained from patients with CAEBV, supporting infection of hematopoietic stem cells, which then differentiate to become EBV infected myeloid and lymphoid cells. While the circumstances under which they become infected with EBV are still unknown, this finding supports allogeneic HCT as the only curative option. There is no standardized treatment regimen for CAEBV, but a combination of immune targeted therapy (e.g. steroids, cyclosporine, PD1 inhibition) and/or chemotherapy is often used to control disease ahead of HCT (81, 82). In a large Japanese registry study of 102 patients with CAEBV that underwent allogeneic HCT, worse outcomes were associated with elevated soluble IL2 receptor at the time of transplant (suggesting poor control of inflammation) and use of radiotherapy in conditioning regimens (83). In a smaller study, MAC and RIC regimens were compared (84). The EFS at 3 years was 54.5 ± 15.0% in the MAC group and 85.0 ± 8.0% for the RIC group, supporting use of the latter.

HLH may be seen secondary to hematologic malignancies, which is a substantial challenge and major driver of HLH in adult patients (14) but observed in children more rarely (85). HLH may occur at diagnosis and be driven by the malignancy (M-HLH) or during chemotherapy (Ch-HLH). Ch-HLH or “on-therapy” HLH should raise high concern for a concomitant infection. Limited data on the frequency and management of malignancy associated HLH in pediatrics is available, although malignancy is estimated as the underlying etiology in approximately 5% of HLH cases (14). In one retrospective analysis of 29 pediatric and young adult patients, HLH was considered triggered by the malignancy in 21 patients with T- (n=12) or B-cell neoplasms (n=7), with EBV infection present as a co-trigger in 5 patients. The remaining 8 patients experienced HLH during chemotherapy (Ch-HLH) mainly for acute leukemias (86).

The diagnosis of M-HLH is frequently difficult to make, particularly when the underlying malignancy is yet undiagnosed. The diagnosis of M-HLH should be considered in a patient with a biopsy-proven malignancy who has clinical, and laboratory features consistent with HLH and no alternative cause of hyperinflammation. Different criteria and scoring systems have been used in adult patients, including the MD Anderson Criteria for M-HLH, H Score, and modification to the HLH-2004 criteria (87). Additionally, an Optimized HLH Inflammatory index (OHI) comprising soluble CD25 >3900 U/mL and Ferritin >1000ng/mL was developed to diagnose M-HLH more accurately and predict mortality in adult patients (88).

Current treatment practices for M-HLH are largely based on expert opinion, given the absence of randomized clinical trials and heterogeneity of patient populations. Treatment is centered on the initiation of malignancy-directed therapy as soon as safely feasible and utilizing anti-inflammatory therapies to control hyperinflammation. In a pediatric cohort, the median overall survival (OS) was 1.2 years in M-HLH and 0.9 years in Ch-HLH (86). Given this was a small retrospective series, the superiority of prioritizing malignancy-directed vs. HLH-directed therapies could not be determined for the M-HLH cohort. In the Ch-HLH cohort, interventions ranged from postponement of chemotherapy to the use of etoposide-containing regimens (86). Adult M-HLH may be managed with an etoposide-based regimen that closely resembles the HLH-94 protocol, consisting of 8 weeks of treatment with dexamethasone 10mg/m²/d, subsequently tapered, or a dose-reduced regimen in patients with advanced age or significant comorbidities (89). A single-arm Phase 2 trial evaluated the safety and efficacy of a dose adjusted EPOCH regimen (Etoposide, Prednisolone, Vincristine, Cyclophosphamide, Doxorubicin) with or without rituximab in patients with previously untreated Non-Hodgkin Lymphoma-associated HLH to simultaneously address the HLH and underlying lymphoma. While this regimen led to an overall response rate (ORR) of 80.7% and 5-year OS of 73.1% in patients with B-cell lymphomas, those with T/NK lymphomas had dismal outcomes, with ORR of 13.8% and 1 year OS of 3.4% (90). A two-tiered approach of using etoposide and corticosteroids and, if needed, a salvage regimen of doxorubicin, etoposide, and methylprednisolone, to control the HLH with subsequent initiation of appropriate chemotherapy is frequently favored. Allogeneic HCT or cellular therapy in patients with malignancy associated HLH are employed only if this is indicated for the underlying malignancy.

The earliest recognized hyperinflammatory complication of CAR T cell therapy was Cytokine Release Syndrome (CRS) (95). The hallmarks of CRS are fever, cardiopulmonary dysfunction, and other (hepatic, renal, gastrointestinal) organ dysfunction. Laboratory abnormalities include elevations in ferritin and IL-6, and coagulopathy, resembling those of HLH/MAS. CAR T cell activation leads to proliferation, secretion of proinflammatory cytokines, and activation of the host immune system including myeloid cells, thereby inducing an inflammatory feedback loop. In patients receiving 4-1BB costimulated CD19-CAR T cells, elevations in IFN-γ, IL-6, IL-8, sIL-2R, sgp130, MCP1, MIP1α, MIP1β, and GM-CSF as well as higher ferritin (>10,000) and CRP were significantly associated with severe (Grade 4-5) CRS compared to those with mild (Grade 0-3) CRS (96). Risk factors for the severity of CRS include cell dose, as well as recipient characteristics such as high tumor burden (97). Additionally, pre-infusion inflammation, peak expansion kinetics, and characteristics of the CAR itself are important (98) such as the transmembrane domain, costimulatory domain (e.g. CD28>41BB for CRS) (99) and affinity of the binder to target (less CRS with lower affinity binder) (100). Thus far, patients receiving CAR T cells for autoimmune indications have not been shown to be at increased risk for CRS, and rates and severity of CRS appear to be low (101). However, patients can experience localized inflammation in previously affected tissues, which is often mild and self resolves or resolves with steroid therapy (referred to as localized immune effector cell associated-toxicity syndrome or LICATS) (102).

The use of the monoclonal antibody tocilizumab revolutionized CRS management by interrupting the proinflammatory signaling cascade via IL-6 blockade without impacting CAR T cell efficacy (95, 103). While corticosteroids were initially avoided so as not to impair function and durability of CAR T cells, targeted use of corticosteroids during peak inflammation or prophylactically in high-risk patients has not been found to adversely impact efficacy (104).

A related hyperinflammatory syndrome, IEC-associated neurotoxicity syndrome (ICANS), was recognized as a distinct entity that could overlap with CRS or occur following its resolution, and thus distinct grading systems and management algorithms were developed (105). Symptoms may range from mild confusion, headaches, tremors, and word finding difficulties to focal deficits, seizures, coma, and cerebral edema. Particularly in patients with prior CNS involvement or pre-existing neurological deficits, documentation of a patient’s baseline neurological exam and, if indicated, baseline MRI/imaging prior to CAR T administration is critical. Management guidelines suggest expeditious workup with neuroimaging, lumbar puncture, and EEG, in consultation with a Neurology team. Preclinical studies have identified IL-1 and IL-6 as key mediators in CRS and ICANS. However, while IL-6 blockade ameliorates CRS, only IL-1 blockade abrogates both CRS and ICANS (106–108). Initially treated supportively, management of ICANS currently includes corticosteroids, tocilizumab (if co-occurring with CRS, although there are concerns about enhancing IL-6 levels in the CSF due to non-CSF permeability of tocilizumab) and, if refractory, anakinra, intrathecal chemotherapy, and CSF pressure relief (109–111).

Most recently, a separate entity, IEC-associated hemophagocytic lymphohistiocytosis-like syndrome (IEC-HS), was described following the experience with CD22-CAR T cell therapies (112). This syndrome has substantial overlap with the clinical and laboratory features of CRS and may be difficult to distinguish; however, it typically presents with delayed onset or following resolution of initial CRS and is not responsive to conventional CRS management with tocilizumab. Although the frequency was as high as 40.4% in a trial of 59 patients infused with CD22-CAR T cell products (112), other CD22-CAR T cell trials reported lower rates at 18.75% (113). A broad consensus definition ( Table 5 ), grading system, and management recommendations have since been developed by the American Society for Transplantation and Cellular Therapy (114). With increasing recognition, it has also been described with other CAR T cell products such as those targeting CD19 (115) and in investigational CAR T cell products for solid tumors (116). Onset with resolving/resolved CRS or a worsening inflammatory response after initial improvement with CRS therapy is one of the most common manifestations of IEC-HS. Pre-infusion NK cell lymphopenia and higher bone marrow T cell:NK cell ratio may predispose to IEC-HS at least in the setting of CD22-CAR T cell therapy, possibly leading to a deficiency in regulating CD8+ T cell hyperactivation and CAR T cell contraction (112). Free IL-18 was differentially elevated in patients with CRS and IEC-HS compared to those with CRS alone (117) and may warrant investigation of agents targeting this pathway such as tadekinig alfa, a human recombinant IL-18-binding protein, which is not currently FDA approved. High pre-CAR T cell disease burden, a high level of pre-infusion inflammation based on ferritin and C-reactive protein, as well as lower platelet and neutrophil counts were associated with the development of IEC-HS in the context of the commercial CD19-CAR T cell product tisagenlecleucel (115). Germline genetic variants associated with HLH (118) and clonal hematopoiesis (119) may also contribute. Murine studies with perforin-deficient CAR T cells aimed at providing mechanistic insight, recapitulated elevated IL-1β and IL-18 levels and biphasic late CAR T cell expansion, which is characteristic of IEC-HS in humans (120). However, whether the pathophysiology underlying IEC-HS is distinct from or on the spectrum of CRS, and the identification of unifying risk factors, remains an area of active investigation. Similarly, the optimal therapeutic approach remains to be defined. Anakinra is recommended as first-line therapy given its favorable side-effect profile and short half-life allowing dose titration. Corticosteroids should be incorporated for moderate or higher disease severity or maintained in patients already receiving them for CRS. Given the importance of the IFN-γ signaling pathway in IEC-HS, as in other HLH-like conditions, ruxolitinib or emapalumab (used more frequently in pediatrics) are being employed for high grade/progressive IEC-HS (114, 121). Although the effect on durable CAR T cell function remains to be clarified in clinical studies, preclinical models predict that in hematologic malignancies, IFN-γ blockade can reduce macrophage activation without compromising CAR T cell function (122, 123). In contrast, IFN-γR signaling was shown to play a critical role in the susceptibility of solid tumors such as glioblastoma to CAR T cell mediated killing (124). This highlights the need for judicious use of these agents and further investigation to elucidate their impact mechanistically. Given the high rate of morbidity and mortality associated with IEC-HS, prompt diagnosis and aggressive management should be instituted.