Hemophagocytic lymphohistiocytosis: Diagnosis, pathophysiology, treatment, and future perspectives

Abstract

Hemophagocytic lymphohistiocytosis (HLH) is a rare life‐threatening disease in which the immune system becomes overactive due to its inability to effectively respond to infections and/or shut down the immune response to such infections. The discovery of genetic defects in the secretory pathway of natural killer (NK) cells and cytotoxic T cells in some patients with this disease has raised important questions of the role of cytotoxic cells in the control of infections and in immune regulation. This review will give a brief overview of the clinical presentation and accepted treatment of HLH. Furthermore, it will give an in‐depth review into the known genetic defects and current knowledge of the pathophysiology of this disorder, and will highlight recent evidence suggesting that cytotoxic defects in CD4⁺ T regulatory cells may contribute to the pathogenesis of HLH.

• Granzyme
• HLH
• perforin
• suppressor
• T regulatory cells

Abbreviations
IL	=	interleukin
IFN	=	interferon
TNF	=	tumor necrosis virus
ESR	=	erythrocyte sedimentation rate
CMV	=	cytomegalovirus
HHV	=	human herpes virus
VSV	=	Varicella zoster virus
TGF	=	transforming growth factor
GTP	=	guanosine triphosphate
Ras	=	rat sarcoma oncogene
PBMC	=	peripheral blood mononuclear cell
HSV	=	herpes sinplex virus

Clinical presentation

Hemophagocytic lymphohistiocytosis (HLH) is a highly fatal disease of children with an estimated incidence of approximately 1.2 cases/million per year 1. The typical clinical presentation of HLH includes persistent high‐grade fever with splenomegaly, often in the setting of an acute infection 2–4. Other signs include lymphadenopathy, skin rash, jaundice, edema, and central nervous system symptoms. Abnormal laboratory findings include cytopenias affecting at least two lineages, liver dysfunction (e.g. hypertriglyceridemia, hypofibrinogenemia, elevated serum transaminases, hyperbilirubinemia, elevated prothrombin time, elevated partial thromboplastin time, and elevated thrombin time), hyperferritinemia, cerebral spinal fluid (CSF) pleocytosis, hyponatremia, and hypoproteinemia 4,5. In addition, histopathological evaluation of the bone marrow usually demonstrates significant histiocytosis with evidence of hemophagocytic activity. Hemophagocytosis can also be observed in other affected tissues, including the spleen, lymph nodes, CSF, and liver, with the latter demonstrating histological features similar to chronic hepatitis 2–4. One of the most consistent laboratory findings is low or absent natural killer (NK) cell cytotoxicity 6–8, which correlates with the known genetic defects associated with HLH (see below).

In 1991, the Histiocyte Society proposed diagnostic guidelines for HLH based on the common clinical and laboratory findings listed above 3. The Histiocyte Society HLH study group recently revised these guidelines in 2004 (Table I) 9. It is important to realize that not all diagnostic criteria noted in Table I need to be fulfilled before the institution of life‐saving therapies since many subjects fail to meet full diagnostic criteria until late in the course of their disease 3,9,10. Such early therapeutic interventions often need to be implemented based on strong clinical suspicion for HLH to prevent irreversible end‐organ damage.

Table I. Diagnostic Guidelines for HLH.

The diagnosis of HLH is established by fulfilling 1 or 2 of the following criteria:
1.	A molecular diagnosis consistent with HLH (e.g. PRF mutations, SAP mutations, etc.)
OR
2.	Having 5 out of 8 of the following:
	a. Fever
	b. Splenomegaly
	c. Cytopenia (affecting⩾2 cell lineages; hemoglobin⩽9 g/dL (or⩽10 g/dL for infants<4 weeks of age), platelets<100,000/µL, neutrophils<1,000/µL)
	d. Hypertriglyceridemia (⩾265 mg/dL) and/or hypofibrinogenemia (⩽150 mg/dL)
	e. Hemophagocytosis in the bone marrow, spleen, or lymph nodes without evidence of malignancy
	f. Low or absent NK cell cytotoxicity
	g. Hyperferritinemia (⩾500 ng/mL)
	h. Elevated soluble CD25 (IL‐2Rα chain; ⩾2,400 U/mL)

(Note: Adapted from Treatment Protocol of the 2nd International HLH Study, 2004)

Key messages

• Hemophagocytic lymphohistiocytosis (HLH) is a life‐threatening disorder resulting from uncontrolled immune activation.

• Genetic defects in patients with HLH and related disorders consist of mutations in genes required for cytotoxic granule release or function.

• Cytotoxicity is not only important for clearance of infectious organisms, but also for the regulation of the immune system.

The hemophagocytic spectrum

HLH represents the final pathologic diagnosis of a spectrum of disorders with similar clinical and pathologic findings (Table II). Attempts have been made to categorize HLH into two different conditions, ‘primary or familial HLH’ and ‘secondary HLH.’ Historically these attempts were made to help distinguish the cases of HLH that present during infancy and carry a higher mortality, from those due to ‘other causes’ found in secondary HLH that present later in life and have a better prognosis 3,9. Primary HLH has been used to describe patients with a known family history of HLH or documented genetic mutations that are associated with the development of HLH (see below). Secondary HLH encompasses a number of categories that include designations such as virus‐associated hemophagocytic syndrome, infection‐associated hemophagocytic syndrome, macrophage activation syndrome, and malignancy‐associated hemophagocytic syndrome. Unfortunately, distinguishing between primary and secondary HLH is nearly impossible in many circumstances for several reasons. Primary HLH can occur at any age, not just during infancy or early childhood 11–13. Since many of the known genetic mutations associated with the development of HLH are inherited in an autosomal recessive manner, there is often a negative family history of HLH. In addition, underlying genetic mutations account for only ∼40% of all primary HLH subjects. Finally many of the secondary causes are associated with a high mortality, and both primary HLH and secondary HLH can present in patients with a history of a preceding infection 2,4,14–16. These observations suggest the strong need for improved diagnostic criteria that help to distinguish patients based on their clinical severity and prognosis.

Table II. Spectrum of diseases characterized by hemophagocytosis.

Primary HLH

Familial HLH

Chediak Higashi syndrome

Griscelli's syndrome

X‐linked lymphoproliferative (XLP) disease

Wiskott‐Aldrich syndrome (WAS)

Secondary HLH

Virus associated HLH

Herpes virus infection (EBV, CMV, HHV‐6, HHV‐8, VZV, HSV)

HIV

Parvovirus, adenovirus, hepatitis virus

Infection associated HLH ^a

Numerous bacterial, spirocetal, and fungal associated infections

Malignancy Associated HLH

Macrophage activation syndrome (MAS) associated with autoimmune disease

Systemic onset juvenile rheumatoid arthritis (SOJRA)

Others (systemic lupus erythematosus, enthesitis related arthritis)

^a An extensive list of organism thought to be associated with HLH is available at www.cdc.gov/ncidod/eid/vol6no6/fisman.htm110

Abbreviations: CMV = cytomegalovirus; EBV = Epstein Barr virus; HHV = human herpes virus; VZV = Varicella zoster virus

Genetic mutations associated with HLH

Genetic analysis of several families with children presenting with HLH before six months of age demonstrated tight linkage with the chromosomal region 10q21‐22 5. Stepp et al. subsequently verified that four of these patients had homozygous nonsense mutations in the perforin gene found in this chromosomal region, resulting in the lack of protein expression and NK cell cytotoxicity 17. Since this initial discovery seven other genetic mutations have been associated with the development of HLH in humans (Figure 1) (see below). Interestingly, all seven proteins are integral components of the granule exocytosis pathway, and are critical for either the functional secretion of cytotoxic granules, or for the delivery of proteases required for apoptosis of target cells.

Figure 1. Graphical representation of the genes involved in cytotoxic granule formation, secretion, and target cell death. The first step in cellular cytotoxicity is the development of functional granules containing effector molecules(e.g. perforin, granzymes) and regulatory molecules (e.g. DPP1, serglycin, and calreticulin). Once an effector cell makes a functional secretory synapse through the interactions of adhesion molecules (e.g. LFA‐1/ICAM‐1), signal transduction events (e.g. SAP/SLAM) are initiated resulting in cell polarization by polymerization of actin/microtubules towards the secretory synapse (e.g. WASp). The preformed cytotoxic granules migrate to the secretory synapse using myosin proteins (e.g. Myo5a) and are released from the actin filaments (e.g. Rab27a) at the plasma membrane. Fusion of the granules with the plasma membrane occurs (e.g. Lyst, Munc13‐4) and effector molecules are released into the synapse. Perforin then polymerizes into pore‐forming structures in the target cell membrane that ultimately allow for the entry of granzymes into the cytoplasm of the target cells through a yet undefined mechanism. The granzymes subsequently induce apoptosis of the target cell through the cleavage/activation of multiple pro‐apoptotic target molecules in the cytoplasm, mitochondria, and nucleus. LFA‐1 = lymphocyte function associated antigen 1; ICAM‐1 = intercellular adhesion molecule 1; Lyst = lysosomal trafficking regulator gene.

Perforin mutations in primary HLH

Perforin is a hydrophobic protein that is packaged into specialized cytotoxic granules of human natural killer (NK) cells (CD56^loCD8⁻), natural killer T (NKT) cells (CD56^loCD8^+lo), and activated CD8⁺ T lymphocytes 18. Perforin is packaged with other granule regulatory proteins, including calreticulin, serglycin, and dipeptidyl peptidase I (DPPI) 19–21. Perforin is related to the complement component C9 22, and polymerizes into a pore‐forming structure in the plasma membrane of target cells upon release by cytotoxic effector cells (Figure 1) 19,20. The critical role of perforin appears to be the introduction of cytotoxic effector molecules (e.g. granzymes) into the cytoplasm of target cells 21,23. Granzymes are a highly conserved family of serine proteases that cleave and activate multiple cytoplasmic, mitochondrial, and nuclear targets, leading to the induction of apoptosis in the target cell 23,24. Perforin or granzymes alone do not appear to induce target cell death, rather the induction of target cell death requires the combination of both perforin and granzymes. How granzymes enter the cytoplasm of the target cells is controversial, and purported mechanisms include direct introduction through the perforin pore, an endocytic receptor (e.g. mannose‐6‐phosphate receptor), or some other endocytic pathway 19,20. Mutations in the perforin gene have been estimated to account for approximately 20%–40% percent of HLH subjects, with significant variations in the incidence of perforin gene mutations between different ethnic groups and types of HLH 25,26. Although the majority of patients with perforin gene mutations present with HLH in early childhood, there are rare reports of teenage and adult presentations 11–13,27,28. To date no other cytotoxic granule protein besides perforin has been associated with the development of HLH, possibly indicating functional redundancy among the various effector granule molecules (e.g. granzymes).

UNC13D/MUNC13‐4 mutations in primary HLH

Feldman et al. demonstrated strong linkage to chromosomal region 17q25 in several HLH patients that had normal perforin expression 29, and showed that all of these patients had mutations in the MUNC13‐4 gene (mammalian homolog of uncoordinated, a C. elegans mutant). This gene was shown to have homology to a family of proteins found in the brain that are critical for the exocytosis of synaptic neurotransmitter vesicles. It was subsequently demonstrated that these MUNC13‐4 patient mutations failed to allow the normal release of granules from cytotoxic T cells by preventing the fusion of these cytotoxic granules to the plasma membrane (Figure 1) 29–31. Mutations in MUNC13‐4 have been found in up to 20% of primary HLH patient cohorts 12.

SH2D1A/SAP mutations in XLP

X‐linked lymphoproliferative (XLP) disease is an immunodeficiency that is associated with the development of lymphoma and hypogammaglobulinemia 32. It has been recognized that XLP patients can present in a similar fashion to HLH patients when infected with Epstein Barr virus (EBV), in that they develop life‐threatening T lymphocyte and monocyte dysregulation 32,33. Arico et al. subsequently demonstrated germline mutations in the XLP‐associated gene, SAP/SH2D1A (see below), in four young males with the clinical diagnosis of HLH 34. Two of the patients were reported to have a family history that was consistent with XLP. Based on the findings of SAP mutations in young male patients with HLH, it has been recommended that all male HLH patients undergo genetic screening for SAP mutations. Proper genetic counseling and screening should also be performed for the families of subjects identified with SAP mutations.

SAP (SLAM‐associated protein), also known as SH2D1A (src homology 2 domain protein 1A), is a small SH2 domain‐containing protein that is a critical adapter molecule between the SLAM (signaling lymphocyte activation molecule) family of immune‐cell receptors and their cognate intracellular signaling molecules (Figure 1) 35–37. Studies in SAP‐deficient mice and humans have shown that SAP is crucial for SLAM‐induced phosphorylation and recruitment of downstream signaling molecules that are important in cytokine production and cell‐mediated cytotoxicity 37–39.

RAB27A mutations in Griscelli's syndrome

Griscelli's syndrome (GS) is rare autosomal recessive disorder characterized by albinism and neurological defects 40–44. It was noted that a subset of patients with GS had variable degrees of cellular immunodeficiency and developed clinical findings consistent with HLH that can lead to death without a stem cell transplantation 40–44. The first gene found to be mutated in GS patients was the MYO5A gene, which encodes a myosin heavy chain that is capable of binding vesicles and moving them along actin filaments 45. Interestingly, the GS patients that had an associated cellular immunodeficiency were found not to have mutations in the MYO5A gene, but rather in a closely linked gene on chromosome 15q21 named RAB27A46. To date, all GS patients that have been reported to develop HLH have been found to have mutations in the RAB27A gene and not the MYO5A gene 43,46,47.

RAB27A encodes for a Ras‐like GTPase that has both Myo5a‐dependent and independent roles in the movement of vesicles (Figure 1). RAB27A deficient patients (GS Type 2) have severely impaired cytotoxic activity 43,48,49, whereas MYO5A deficient patients (GS Type 1) have normal NK and CD8 cytotoxicity 43,48,49. GS Type 2 patients (RAB27A defects), as well as the naturally occurring mutant mouse counterpart (Ashen mice), have been shown to have a defect in the release of secretory granules from microtubules (Figure 1) 43,50.

CHS1/LYST mutations in Chediak‐Higashi syndrome

Chediak‐Higashi syndrome (CHS) is a human disease associated with the clinical findings of albinism, bleeding tendency, neurological symptoms, and frequent pyogenic infections 51. Approximately 85% of CHS patients will enter an ‘accelerated phase’, which is manifested with symptoms that are indistinguishable from HLH patients, including hepatosplenomegaly, fever, liver dysfunction, pancytopenia, lymphadenopathy, coagulopathy, and diffuse lymphocytic and hemophagocytic infiltration of the bone marrow, liver, central nervous system, spleen, and lymph nodes 52,53. These patients have a substantial mortality unless they undergo stem cell transplantation 52,54.

CHS patients display abnormally large granules in a number of cell types, including phagocytes, NK cells, and cytotoxic lymphocytes 51,55. CHS patients have been demonstrated to have mutations in the CHS1/LYST gene (Chediak‐Higashi syndrome 1 gene/lysosomal trafficking regulator gene) 56,57, and, similar to the naturally occurring mouse counterpart (Beige mice), have impaired NK and T cell cytotoxicity 55,58–60. Similar to the Munc13‐4 protein, it appears that the LYST protein is involved in the fusion of granules to the plasma membrane 43,61.

Miscellaneous mutations associated with HLH

Wiskott‐Aldrich syndrome (WAS) is a disease that is characterized by eczema, microthrombocytopenia, and recurrent bacterial, viral and fungal infections 62,63. Although the clinical spectrum is heterogeneous, WAS patients typically display defects in both cellular and humoral immune responses, and are at high risk for the development of malignancies (up to 20%) and autoimmune disorders (40%–70%) 62,63. The WAS protein (WASp) has been shown to be an actin‐modulatory protein that is important in actin cytoskeletal reorganization after activation of various types of receptors (Figure 1) 64,65. WASp is not only crucial for cell trafficking, but also for the development of functional secretory synapses. Cells from WAS patients demonstrate defects in cell motility, homing, phagocytosis, cell adherence, and NK cell cytotoxicity. Interestingly, Pasic et al. reported a case of EBV‐associated HLH in an infant with WAS, supporting the idea that genetic mutations that fail to form a functional secretory synapse can lead to HLH 66.

Other immunodeficiencies that have been associated with the development of HLH include T‐B+NK+ severe combined immune deficiency (SCID) 67, IL‐2Rγ common chain (X‐linked T‐B+NK‐ SCID) 68, Omenn's syndrome (T+B‐NK+ SCID) 69, and DiGeorge syndrome (22q11.2 deletion) 70,71. These cases demonstrate the ability to develop severe HLH in the face of selective deficiencies in the cellular immune system. In addition, they underscore the importance of a thorough immunological surveillance in HLH patients, and the consideration of other immunological defects other than those that previously associated with HLH.

Pathophysiology of HLH

The pathophysiology of HLH and related disorders appears to involve the excessive activation of lymphocytes and macrophages resulting in cytokine production by these cells. Numerous cytokines are reported to be elevated in HLH, including IFN‐γ, IL‐6, IL‐10, IL‐12, IL‐16, IL‐18, and TNF‐α 72–76. This profound cytokine activation results in the organ dysfunction characteristic of the disorder, including hyperferritinemia and hyperlipidemia, as well as macrophage activation. The resulting organ infiltration and excessive phagocytosis in the bone marrow, liver, and other organs result in the clinical features of the syndrome. Similar observations are noted in the closely related disorder macrophage activation syndrome (MAS), which occurs in patients with systemic onset juvenile rheumatoid arthritis (SOJRA) and other autoimmune diseases. Liver histology from patients with SOJRA shows the presence of macrophages and CD8⁺ lymphocytes, and significant staining for TNF‐α and IFN‐γ 77.

The close relationship between HLH, GS, and CHS is understandable since all of the genetic mutations are involved in granule exocytosis and cytotoxicity, but why secondary HLH occurs in response to infectious organisms, malignancy, or autoimmune disease is not clear. It is possible that these associations are not coincidental, but rather represent cytotoxic defects that are yet to be characterized. Alternatively, these may represent secondary phenomenon reflecting immune dysregulation and acquired cytotoxic defects. In support of this, Fillopovich et al. studied patients with familial HLH with or without perforin defects and patients with EBV associated HLH 25. In familial HLH patients with or without perforin defects, they found that NK cell numbers were normal yet they had absent NK cytotoxicity, but this has not been a consistent finding in other patients with reported perforin defects 78. In EBV‐associated HLH, both NK numbers and cytotoxicity were reduced, suggesting that an acquired NK cell deficiency may be responsible for this clinical syndrome. Similarly, in SOJRA‐associated MAS, which is close if not identical to HLH, reports have shown defects in NK cell numbers and NK cytotoxicity 79. Interestingly, decreases in perforin expression have been noted in SOJRA and the levels return to normal values after autologous stem cell transplant, arguing that the defects are acquired and can be reversed 80. We have also observed decreased cytotoxicity in two patients (one with secondary HLH and one with SOJRA) which reversed upon treatment, also arguing that these disorders may represent acquired defects in cytotoxicity (unpublished observations).

The genetic defects in cytotoxic granule release has led to the belief that HLH is caused by the inability of cytotoxic T lymphocytes or NK cells to clear intracellular infections. Consistent with this notion, perforin deficient mice exhibit increased susceptibility to lymphocyte choriomeningitis virus (LCMV) 81,82, ectromelia 83, Herpes simplex virus 84, Theiler's virus 85, Listeria monocytogenes86, Toxoplasma gondi87, and Histoplasma capsulatum88. These observations would argue that the failure to clear infections results in prolonged antigen presentation, excessive lymphocyte activation, and the immune pathology associated with HLH. This is supported by studies of mice exposed to LCMV showing that viral burdens were higher in perforin deficient animals and that this correlated with higher IFN‐γ levels 89. Furthermore, LCMV neutralizing antibodies were able to decrease the IFN‐γ levels and decrease mortality in these animals.

The failure to clear infections likely contributes to the pathology of HLH, although this cannot be the only explanation. Several studies have suggested that the lack of perforin results in a failure to regulate or terminate an immune response. Perforin deficient animals infected with LCMV displayed excessive expansion of activated CD8⁺ lymphocytes with significant mortality that was reversed by depletion of CD8⁺ cells 90,91. Similar studies showed that CD8⁺ depletion, but not NK cell or CD4⁺ lymphocyte depletion, rescued LCMV infected perforin deficient animals from mortality 89,91. These studies also demonstrated that neutralizing antibodies to IFN‐γ prevented mortality, while antibodies to IL‐10, TNF‐α, IL‐12, IL‐18, macrophage colony stimulating factor, and granulocyte macrophage colony stimulating factor had no effect on mortality 89. Despite the increased mortality of perforin deficient mice infected with LCMV, viral titers were the same between perforin deficient and wild type animals 90. Additional studies have demonstrated that inoculation of perforin deficient animals with staphylococcus superantigen results in increased expansion of CD8⁺ cells, and peptide‐specific activation of perforin deficient CD8⁺ T cells results in increased numbers of CD8⁺ T cells 92. These studies show that perforin contributes to the control of CD8⁺ expansion independent of the failure to clear infectious organisms, and other studies support this notion. Perforin and Fas double‐deficient animals show a more aggressive autoimmune phenotype than Fas deficient animals 93,94. Perforin deficient animals exhibit a chronic relapsing course of experimental allergic encephalomyelitis compared with Fas deficient animals which exhibit a monophasic course 95. Perforin deficient animals have also been shown to exhibit exaggerated antibody responses in response to influenza antigens, again arguing that perforin may perform additional regulatory functions in immune responses 96.

Role of regulatory T cells in HLH

There is another possible explanation for the observed immune regulatory role of perforin/cytotoxicity not mentioned yet which involves CD4⁺ T regulatory cells (Tregs). Tregs have emerged as active regulators of immune responses in both humans and mice. Tregs have been grouped into two general categories, adaptive Tregs and natural Tregs, based on how they are generated, their molecular expression patterns, and their suppressive properties. Natural Tregs constitutively express the high affinity IL‐2 receptor (CD25) and the protein forkhead box P3 (FoxP3), are generated in the thymus, and suppress naïve T cell proliferation in a contact dependent manner. Adaptive Tregs are generated in the periphery to a variety of stimuli and suppress T cells responses through the secretion of immunosuppressive cytokines, such as IL‐10 and TGF‐β. When human CD4⁺ cells are activated with antibodies to the T cell co‐receptor (CD3) and the complement regulatory receptor CD46 (membrane cofactor protein) in the presence of IL‐2, they proliferate vigorously and express very high levels of IL‐10. Supernatants from these cells suppress the proliferation of naïve T lymphocytes in an IL‐10 dependent manner, and thus the phenotype of these cells is consistent with adaptive Tregs 97. These cells also express high levels of granzyme B and perforin 98,99. In addition, when human natural Tregs are isolated from human peripheral blood mononuclear cells (PBMCs) by flow cytometry, activated with antibodies to CD3 and CD46, and tested for granzyme expression, these cells express granzyme A but not granzyme B 98. When both of these cell types are used in cytotoxicity assays they exhibit killing of autologous targets, which is dependent on perforin, but independent of Fas/Fas ligand interactions. Interestingly, only activated CD4⁺ and CD8⁺ autologous targets are killed, while resting T cells are resistant to killing. Finally, CD14 positive monocytes and dendritic cells can also be killed. Importantly dendritic cells, CD8⁺ cells, and monocytes are the most susceptible to cytotoxicity mediated by both types of T regulatory cells, and these target cell types are central to the immune pathology of HLH 98,99. In addition, Janus Kinase 3 deficient animals exhibit profound myeloid expansion as they age, and these mice lack Tregs, which suggests that Tregs are necessary to prevent myeloid expansion 100. Furthermore, FoxP3 deficient animals, which lack natural Tregs, show an expansion of myeloid cells with myeloid infiltration of bone marrow, liver and other organs and organ pathology which is reminiscent of HLH (manuscript in preparation). No studies, however, have directly tested regulatory T cell function in patients with HLH or related disorders. Regardless, these data suggest that CD4⁺ lymphocytes may also exhibit perforin dependent cytotoxicity that is required for immune regulation, and the loss of this function of Tregs in perforin deficient animals and patients may contribute to the pathophysiology of HLH.

The fact that humans and mice with defects in cytotoxicity exhibit findings consistent with HLH argues strongly that cytotoxicity is important in immune regulation. We would propose a model where early in a course of an immune response, antigen presentation to CD4⁺ and CD8⁺ cells results in priming of the immune response and expansion of antigen specific cells (Figure 2). These cells are then present in sufficient numbers as effector cells to eradicate viruses and other pathogens, which may be dependent on the cytotoxic function of these cells. Once the antigen load has decreased, the expanded immune system must contract to keep the final number of immune cells constant, which also appears to be dependent on functional perforin expression. Which cells are involved in the contraction phase of the immune response is not clear and may involve multiple cytotoxic cell types, such as NK cells, NKT cells, activated CD8⁺ cells, and/or CD4⁺ Tregs. The end result, however, is the effective clearance of the pathogens and contraction of the immune response. When this process is defective, the findings of excessive immune activation and HLH become apparent (Figure 2).

Figure 2. Hemophagocytic lymphohistiocytosis(HLH) results from a failure of immune cell contraction due to defects in cytotoxicity. Early in the immune response, CD4 and CD8 lymphocytes are activated by antigen presenting cells and expand to be able to clear infectious pathogens. Once an effective immune response is established and the pathogens are eradicated, the expanded immune repertoire must contract to keep the relative number of lymphocytes constant. This contraction phase is dependent on cytotoxicity of lymphocytes. This may be mediated by CD8 lymphocytes, natural killer cells, or possibly by the cytotoxic T regulatory cells. Without this contraction phase, the immune response proceeds unchecked as is seen in hemophagocytic lymphohistiocytosis, resulting in the demise of the affected patients without treatment. APC: antigen presenting cell.

Therapeutic management of HLH

Untreated HLH subjects develop end‐organ damage from lymphocyte infiltration and macrophage activation. Prior to the institution of aggressive chemotherapy and immunosuppression, children with HLH had a very poor prognosis with only ∼5% of patients surviving one year from diagnosis 1. The first prolongation of survival was reported when these regimens were modified to include etoposide and glucocorticoids 101,102. Later, bone marrow transplantation became the mainstay of treatment for familial HLH, which resulted in prolonged survival and even a cure in some patients 103. With the introduction of the HLH‐94 protocol, which includes dexamethasone, cyclosporin A, VP‐16, and intrathecal methotrexate combined with stem cell transplantation (SCT), the three‐year overall survival rate has approached ∼50%–70% 104,105.

In 2004 a revised HLH treatment protocol (HLH‐2004) was developed based on the success of the HLH‐94 protocol. Similar to the HLH‐94 protocol, patients with clinical and laboratory findings consistent with HLH are aggressively treated with dexamethasone, etoposide, cyclosporin A, and in some cases intrathecal methotrexate (HLH‐2004 protocol). In addition, the primary disorder should be treated appropriately in cases of infection‐associated HLH, malignancy‐associated HLH, or other secondary causes of HLH. During this period it is important to investigate for genetic defects in perforin and SAP (if male). Assays for genetic defects in MUNC13‐4 are not yet clinically available, nor are there clinical assays for genetic defects observed with Griscelli syndrome or Chediak‐Higashi syndrome. It is important to consider functional assays of cytotoxicity (i.e. NK cytotoxicity assays) to detect defects in cytotoxicity not yet explained by available genetic tests. After eight weeks of therapy, all children with a history of familial disease, with a genetic diagnosis, or with persistent, reactivated or unresponsive disease should go immediately to allogeneic stem cell transplantation if an acceptable donor is available. Donors related to the patient should be screened for cytotoxic defects or genetic mutations to prevent transplanting a patient with a similar genetically affected bone marrow, since the timing of HLH disease presentation can vary with similar genetic mutations (see above). For those patients with secondary HLH responsive to the initial therapy, treatment should be halted to prevent the morbidity/mortality associated with stem cell transplantation. It is critical at this time, however, to closely monitor patients with secondary HLH for any signs/symptoms of relapse, and if relapse is apparent, to proceed immediately to stem cell transplantation. Potentially, cytotoxicity assays could be used to determine if the defects of cytotoxicity are reversible or persistent in patients responsive to the HLH protocol, which could help decide what type of treatment would be best suited for these patients.

Emerging therapeutic modalities

As we learn more about the pathophysiology and molecular mechanisms that lead to the clinical symptoms of HLH, additional therapeutic interventions will undoubtedly become available for treatment of patients. As noted above, two cytokines that are negatively associated with the survival of HLH subjects include high levels of IFN‐γ and TNF‐α 106–108, and neutralizing antibodies against IFN‐γ significantly improved the overall survival in the perforin knockout mouse model of HLH 89,91. Given the availability of anti‐cytokine medications for humans, it would seem logical that such immune modulating drugs may be beneficial in the treatment of patients with HLH. In fact, we have treated two cases of refractory HLH with anti‐TNF‐α (Infliximab, 10 mg/kg once per week) with marked clinical and laboratory improvements (Figure 3) (unpublished observations, D.B. Wilson and W.J. Grossman, Washington University, 2000). In addition, there may be a role for IL‐1 neutralizing therapies in the treatment of HLH. Recently the IL‐1 receptor antagonist Anakinra has emerged as an effective therapy for SOJRA. As noted above there appear to be cytotoxic defects in patients with SOJRA, and SOJRA is associated with HLH. We have treated a patient with SOJRA and MAS with the IL‐1 receptor antagonist Anakinra with dramatic results 109. Although the patient was already receiving the HLH‐94 treatment protocol and was improving, it was not until Anakinra was started that his ESR, ferritin levels, and anemia normalized. He has remained symptom free, off all other therapies for almost two years on maintenance doses of Anakinra 109. Although this data is difficult to interpret given the multiple therapies this patient was receiving, it is a reasonable and safe treatment option to consider for patients resistant to current therapies. As new biologics become available, such as anti‐IFN‐γ, anti‐IL‐6 and anti‐IL‐18, there will certainly be other therapies in the arsenal for the treatment of HLH. There have been no studies to date which evaluate the effectiveness of these therapies, however, and these therapies should not replace the current protocols. The anecdotal evidence above was presented to raise the possibility of these therapies in the future or in patients resistant to standard therapies. The ultimate cure for patients with refractory HLH remains stem cell transplantation.

Figure 3. Effective treatment of a refractory hemophagocytic lymphohistiocytosis patient with Infliximab(anti‐TNF‐α). Briefly, the 16‐year‐old male subject had EBV‐associated hemophagocytic lymphohistiocytosis (HLH) that was refractory to IVIG, steroids, and VP‐16, leading to pulmonary failure and mechanical ventilation. Upon institution of Infliximab therapy his clinical status improved significantly within 24–48 hours, represented by a marked decrease in his serum ferritin levels and improvement of all clinical parameters including removal from mechanical ventilation.

The study of patients with HLH over the past several decades has led not only to substantial improvements in the care and survival of these patients, but also to significant new insights into the molecular defects and pathophysiology of a once fatal disease. This success story could not have been possible without the generation of an HLH registry, an active international collaborative effort, or parent organizations which helped to raise awareness and funding opportunities for this once universally fatal disease (http://histio.org/society/). With further study of the role of cytotoxicity in immune regulation, not only will the care for affected patients continue to improve, but also the scientific community will continue to gain knowledge of immune regulation, information that could prove vital to the treatment of other diseases in the future.

Acknowledgements

This work was supported by the Advancing a Healthier Wisconsin program (WJG), the Hope Street Kids (WJG), and the Abbott Scholars Award in Rheumatology Research (JWV). We thank David Wilson for his critical review of this manuscript.