Recent advances in the diagnosis and treatment of hemophagocytic lymphohistiocytosis

Hemophagocytic lymphohistiocytosis (HLH) is a rare life-threatening disease of severe hyperinflammation caused by uncontrolled proliferation of activated lymphocytes and macrophages secreting high amounts of inflammatory cytokines. It is a frequent manifestation in patients with predisposing genetic defects, but can occur secondary to various infectious, malignant, and autoimmune triggers in patients without a known genetic predisposition. Clinical hallmarks are prolonged fever, cytopenias, hepatosplenomegaly, and neurological symptoms, but atypical variants presenting with signs of chronic immunodeficiency are increasingly recognized. Impaired secretion of perforin is a key feature in several genetic forms of the disease, but not required for disease pathogenesis. Despite progress in diagnostics and therapy, mortality of patients with severe HLH is still above 40%. Reference treatment is an etoposide-based protocol, but new approaches are currently explored. Key for a favorable prognosis is the rapid identification of an underlying genetic cause, which has been facilitated by recent immunological and genetic advances. In patients with predisposing genetic disease, hematopoietic stem cell transplantation is performed increasingly with reduced intensity conditioning regimes. Current research aims at a better understanding of disease pathogenesis and evaluation of more targeted approaches to therapy, including anti-cytokine antibodies and gene therapy.


Introduction
Hemophagocytic lymphohistiocytosis (HLH) is a lifethreatening hyperinfl ammatory disease caused by an uncontrolled and dysfunctional immune response [1]. It is characterized by activation and massive proliferation of T cells and macrophages, leading to marked hypercytokinemia [2,3]. Low or absent natural killer (NK) cell and CD8+ cytotoxic T lymphocyte (CTL) cytotoxicity is one of the hallmarks of HLH and leads to impaired regulation of the immune response [4]. HLH is fatal when untreated but even with aggressive treatment still has a high mortality rate [1]. When fully developed, HLH has a characteristic phenotype, but diff erent path ways of pathogenesis can lead to this clinical picture. HLH frequently develops in patients with underlying genetic disease (primary or familial HLH), but can also occur secondary to infection, malignancy, metabolic or autoimmune diseases in patients with no known genetic predisposition ('secondary' or acquired HLH).

Primary HLH
Familial HLH (FHL) has an incidence of approximately 0.12 to 1 cases per 100,000 children per year, even though it may be more common in areas with high consanguinity due to the autosomal-recessive inheritance [5,6]. Five diff erent forms of FHL have so far been described and four genes, accounting for over 90% of familial cases, have been identifi ed (Table 1) [7][8][9][10][11][12][13]. Th ey encode the proteins perforin, MUNC13-4, syntaxin-11 and MUNC18-2, all of which play a key role in lymphocyte cytotoxicity ( Figure 1) [14]. Most patients with FHL develop HLH within the fi rst months of life. However, up to 20% of patients present at more than 2 years of age, and in rare cases, patients with FHL remain asymptomatic until adulthood [1,[15][16][17].
Syndromal immunodefi ciencies associated with albinism, including Chédiak Higashi syndrome (CHS), Griscelli syndrome type II (GSII), and Hermansky-Pudlak syndrome type II (HPSII) also predispose to HLH. Th ese patients show variable degrees of (partial) albinism, platelet dysfunction, and immunodefi ciency in addition to their risk of developing HLH [18,19]. Th e genes aff ected in CHS, GSII, and HPSII are also involved in granule-dependent lymphocyte cytotoxicity (Figure 1) [20,21]. Onset of HLH in patients with these diseases tends to be later than in patients with FHL.

Abstract
Hemophagocytic lymphohistiocytosis (HLH) is a rare life-threatening disease of severe hyperinfl ammation caused by uncontrolled proliferation of activated lymphocytes and macrophages secreting high amounts of infl ammatory cytokines. It is a frequent manifestation in patients with predisposing genetic defects, but can occur secondary to various infectious, malignant, and autoimmune triggers in patients without a known genetic predisposition. Clinical hallmarks are prolonged fever, cytopenias, hepatosplenomegaly, and neurological symptoms, but atypical variants presenting with signs of chronic immunodefi ciency are increasingly recognized. Impaired secretion of perforin is a key feature in several genetic forms of the disease, but not required for disease pathogenesis. Despite progress in diagnostics and therapy, mortality of patients with severe HLH is still above 40%. Reference treatment is an etoposide-based protocol, but new approaches are currently explored. Key for a favorable prognosis is the rapid identifi cation of an underlying genetic cause, which has been facilitated by recent immunological and genetic advances. In patients with predisposing genetic disease, hematopoietic stem cell transplantation is performed increasingly with reduced intensity conditioning regimes. Current research aims at a better understanding of disease pathogenesis and evaluation of more targeted approaches to therapy, including anti-cytokine antibodies and gene therapy.
All patients with primary HLH have a high risk of recurrence. Th erefore, hematopoietic stem cell trans planta tion (HSCT) is the only curative option [27].

'Secondary' HLH
Patients manifesting with HLH in the absence of a disease-causing mutation in the known genes and without strong indications for a genetic predisposition, such as familial disease or recurrent episodes of HLH, are currently classifi ed as suff ering from 'secondary' HLH. Th ere are sparse data on the incidence, but it is probably more frequent than primary HLH [28]. One source reports an incidence of 0.36 cases of malignancyassociated HLH per 100,000 adults per year [29]. It is highly likely that, due to some overlap with sepsis syndromes, HLH is under-diagnosed, in particular in the adult population. Many patients with 'secondary' HLH mani fest beyond infancy, but increasing recognition of late-onset FHL renders the age at onset a poor indicator of disease etiology. Moreover, the genetic basis of HLH remains to be fully defi ned. Th us, fi nal classifi cation of a patient as suff ering from 'secondary' HLH must remain preliminary.

Macrophage activation syndrome
Macrophage activation syndrome (MAS) is a potentially life-threatening complication of autoinfl ammatory and autoimmune diseases and may be classifi ed as a variant of 'secondary' HLH. When fully developed, the clinical features of MAS are indistinguishable from HLH. However, patients with MAS may show distinct fi ndings like neutrophilia or thrombocytosis in early disease stages, which are unusual in patients with primary HLH [38].
MAS is rare, but it is estimated that up to 7 to 30% of patients with active systemic juvenile idiopathic arthritis (Still's disease) experience some form of MAS, ranging from subclinical or mild to full-blown disease, with a mortality of up to 22% [38]. MAS is most frequent in systemic juvenile idiopathic arthritis, but has been observed in Kawasaki's disease, systemic lupus erythematodes, and other rheumatic diseases [38]. Th ere are some reports of MAS in patients after treatment with anti-TNF-α antibodies for rheumatological disease [39].
Specifi c diagnostic criteria for MAS have been suggested. Th ey include falling leukocyte and platelet counts, hyperferritinemia, hypofi brinogenemia, hemophagocytosis in bone marrow, elevated liver enzymes, elevated erythro cyte CTL or NK cell Perforin Granzymes sedimentation rate, and hypertriglyceridemia [40,41]. MAS can be the presenting feature of patients with autoinfl ammatory and autoimmune diseases and features of such diseases (such as arthritis or rash) should therefore be carefully looked for in HLH patients in the course of their disease.

FHL and immunodefi ciencies associated with albinism
One of the main immune defense mechanisms against infections with intracellular pathogens is contactdependent cytotoxicity mediated by CTLs and NK cells [42]. After recognition of infected cells and formation of an immunological synapse, cytotoxic granules, containing perforin and granzymes, are polarized to the contact site between eff ector and target cell and are released into the intercellular space, where they can mediate their cytolytic eff ector function [43]. Importantly, lymphocyte cytotoxicity is not only directed at infected cells, but also against antigen-presenting cells (APCs). Elimination of APCs provides an important negative feedback to limit T-cell-mediated immune responses. Th ere is evidence that CTLs are more important in the pathogenesis of HLH than NK cells, but immune regulation through killing of APCs has also been described for NK cells [44][45][46]. In the absence of eff ective cytotoxicity, APCs continue to stimulate CTL [47], leading to ongoing production of cytokines, in particular IFN-γ, which has a key role in macrophage activation. Activated T cells and macrophages infi ltrate tissues such as the liver, bone marrow and central nervous system, secrete cytokines and show excessive phagocytic activity. IFN-γ and TNF-α have toxic eff ects on hematopoietic cells, contributing to cytopenias. TNF-α also inhibits lipoprotein lipase, causing hypertriglyceridemia [1], and IL-1, IL-6, and TNF-α induce fever [1]. Activated macrophages secrete ferritin as well as plasminogen activator, while activated T and NK cells shed their IL-2 receptor, further contributing to the characteristic laboratory abnormalities of HLH. Data from animal models suggest that IFN-γ is the key cytokine involved in this infl ammatory cascade and IFN-γ blockade is an eff ective treatment for HLH in mice [45]. FHL and the immunodefi ciencies associated with albinism are associated with defective lymphocyte cytotoxicity. In FHL2, perforin itself is defective, while the other diseases aff ect proteins involved in the biogenesis, intracellular transport and exocytosis of perforincontaining lytic granules ( Figure 1 and Table 1) [14]. Th e complex syndromic nature of the albinism disorders can be explained by the similarities in the molecular machineries of vesicle traffi cking, including pigment transport in skin and hair or degranulation of platelets and mast cells.
It remains to be determined whether the above stated concept of pathogenesis is suffi cient to explain the immune dysregulation of HLH or whether an additional immune regulatory role of perforin -directed at immune cells other than APCs -has to be postulated. Another open question is whether an external triggering factor, such as an infection, is always required (but not always detected) or whether presentation of auto-antigens and APC activation by endogenous infl ammatory triggers may be suffi cient to trigger disease. In any case, the fulminant nature of this highly infl ammatory disease indicates a key role for lymphocyte cytotoxicity in limiting physiological immune reactions.

Other primary immunodefi ciencies predisposing to HLH
At least four other primary immunodefi ciencies predispose to HLH, the two X-linked disorders SAP and XIAP defi ciency and the autosomal-recessive ITK defi ciency and CD27 defi ciency [22,23,25,26]. In all four diseases, development of HLH is almost exclusively triggered by EBV infection [22,23,25,26]. SAP defi ciency is characterized by an impaired T cell-B cell interaction, which also involves an inability of cytotoxic T cells to lyse B cells, which are the main target cells of EBV [48]. ITK and CD27 defi ciency are characterized by a poor control of EBV infection, but as in XIAP defi ciency, the molecular mechanisms predisposing to HLH so far remain elusive. Of note, NK T-cell development, which may also be relevant for control of EBV infection, is impaired in all four diseases [24][25][26]49,50].

'Secondary' HLH
Th e pathogenesis of 'secondary' HLH is less well understood. Cytotoxic lymphocyte degranulation and cytotoxicity are not impaired in most cases [51]. Nevertheless, the balance between APC activation and CTL-mediated control may be disrupted through increased APC activation. Intracellular pathogens can activate APC directly -for example, via toll-like receptor (TLR) activation. TLRs could also be stimulated by anti-DNA antibodies in systemic lupus erythematosus [44,52]. Data from a mouse model indicate an important protective role for IL-10 in 'secondary' HLH induced by TLR9 stimulation [44].

Clinical picture
Th e full clinical picture of HLH is quite characteristic, but the initial presentation is non-specifi c. First manifestations of HLH can resemble common infections, malignancy, fever of unknown origin, or autoimmune and autoinfl ammatory disorders [1]. Typical clinical fi ndings in the further course are prolonged fever, mostly unresponsive to antibiotic therapy, and hepatosplenomegaly. In over 30% of cases, neurologic symptoms, including seizures, meningism, a decreased level of consciousness, irritability, hyper-/hypotonia or ataxia, are present [53,54]. Lymphadenopathy, diff erent kinds of rash, edema, and jaundice are less frequent.
Laboratory fi ndings include cytopenias, usually beginning with thrombocytopenia evolving into severe pancytopenia, hyperferritinemia, elevated transaminases, hypofi brinogenemia, hypertriglyceridemia, hypoalbumine mia, and hyponatremia [1]. Additional immunological fi ndings include elevated sCD25 and reduced NK cell cytotoxicity. Many patients with HLH show signs of disseminated intravascular coagulation. Hemophagocyto sis can be absent in the early stages of the disease [55], but repetitive bone marrow aspirations may reveal hemophagocytosis in the further course. It is important to note that the diagnosis of HLH does not depend on this morphological fi nding.
Th ere are an increasing number of descriptions of atypical and late-onset cases of primary HLH. Patients may present with diff erent clinical phenotypes, including aplastic anemia, autoimmune lymphoproliferative disease, recurrent fever, hypogammaglobulinemia and infection susceptibility (resembling common variable immunodefi ciency), granulomatous lung or liver disease, or sterile encephalitis [16,17,56,57]. In one series up to 14% of adult patients presenting with HLH showed hypomorphic mutations in PRF1, MUNC13-4, and STXBP2, leading to less severe cytotoxicity defects that cause milder and atypical forms of HLH presenting beyond infancy [15].

Prognostic factors for poor outcome
Th e extent of hyperbilirubinemia, thrombocytopenia, hyper ferritinemia and cerebrospinal fl uid pleocytosis seem to be key risk factors for early death in HLH, as are lack of improvement in hemoglobin or fi brinogen levels, persisting thrombocytopenia and persistent fever after the start of therapy [58]. In EBV-induced HLH, a high viral load is associated with a poor outcome [31].

Diagnosis
A timely diagnosis is essential in HLH to initiate lifesaving treatment. Th ere is no single clinical or laboratory parameter that allows establishing a defi nite diagnosis. A panel of eight diagnostic criteria proposed by the Histiocyte Society is the current standard for establishing the diagnosis (Table 2) [59].
Five of these eight criteria have to be fulfi lled in order to establish the clinical diagnosis. Th e criteria do not discriminate between primary and 'secondary' forms of the disease. Th ere is some evidence that simplifi ed criteria, including fever, splenomegaly, thrombocytopenia, and elevated ferritin, warrant more detailed immunological diagnostic evaluation. Ferritin >10,000 μg/L and high sCD25 seem to be more specifi c immuno logical markers [27,60]. Nonetheless, their specifi city and sensitivity is limited as well and neither normal ferritin nor normal sCD25 values exclude the diagnosis of HLH.
It is relevant to note that currently available additional immunological or genetic fi ndings are not helpful in answering the question of whether or not a patient suff ers from the acute clinical syndrome of HLH. Th ey are of key importance, however, to discriminate between the primary and 'secondary' forms of the disease. Moreover, they are relevant for deciding whether symptoms and fi ndings that do not strictly fulfi ll the criteria for HLH are related to an underlying genetic disease predisposing to HLH. Exclusion of a genetic disease predisposing to HLH is equally important for childhood and adult-onset forms of the disease. Impaired lymphocyte cytotoxicity is the hallmark of primary HLH and reduced NK cell cytotoxicity has been well established as a diagnostic criterion. However, the test has several practical limitations, in particular in patients with reduced NK cell numbers. Degranulation assays of NK cells or CTLs use the membrane expression of the lysosomal marker protein CD107a upon activation as a marker of exocytosis of lytic granules [61]. Th ese assays can rapidly diagnose patients with a degranulation defect [18,62,63] (Figure 2). In a recent large collaborative study involving 494 patients referred for evaluation of HLH, a resting NK degranulation assay below 5% could detect genetic degranulation defects (FHL3-5, GSII, CHS) with a sensitivity of 96% and a specifi city of 88% [51]. Blood fi lm and hair shaft microscopy (to identify CHS or GSII) and intracellular stainings for perforin, SAP, and XIAP provide additional simple tests allowing rapid identifi cation of a potential predisposing genetic disease. Figure 3 gives an overview of a proposed diagnostic approach in patients with suspected primary HLH, and Figure 4 shows additional microscopic fi ndings in HLH. Using these tests, the indication for hematopoietic stem cell transplantation can now be established within 48 hours and before results of genetic tests are known, which will improve the management and prognosis for patients with HLH. On the basis of these immunological tests, a priority list for targeted sequencing of the relevant genes can be established [51,64,65]. Of note, deep intronic mutations, genomic inversions or unusual splice site muta tions may also predispose to HLH such that conven tional exon sequencing is not suffi cient to exclude a genetic cause [66,67].
In parallel to the immunological investigations, a careful microbiological workup is required to identify infectious triggers of HLH, in particular EBV, cytomegalovirus or leishmania infection. Controlling infectious diseases is an important element of HLH therapy [31].

General principles of therapy
Untreated primary HLH is rapidly fatal within a few weeks [5]. Prompt and adequate treatment is of crucial importance for a positive outcome. Th erapy should be started in all cases with high suspicion after diagnostic tests have been initiated but regardless of whether the results of all examinations have been obtained. Th e initial therapy consists of immunosuppressive and/or chemothera peutic agents and aims at suppressing the hyperinfl ammatory component of the disease as well as eliminating activated cytotoxic lymphocytes and macrophages. Steroids inhibit infl ammation by attenuating cytokine responses and inhibiting diff erentiation of dendritic cells and also have cytotoxic eff ects on lymphocytes. Cyclosporin A directly aff ects CTL activation as well as macrophage function. Etoposide induces apoptosis in lymphocytes as well as in APCs. Anti-thymocyte globulin (ATG) directly targets T cells, whereas alemtuzu mab, an anti-CD52 antibody, targets lymphocytes in general and APCs. Intrathecal therapy with methotrexate and steroids is targeted at central nervous system disease.

Immunochemotherapy
Immunochemotherapy is widely used to induce and maintain remission until HSCT in primary HLH. Th e HLH-2004 protocol [59] consists of a two-week induction phase including etoposide, cyclosporin A, dexamethasone, intrathecal methotrexate, and intrathecal prednisone followed by a tapering phase of 6 weeks. If HSCT is planned, patients are placed on continuation therapy consisting of cyclosporine A and biweekly pulses of etoposide and dexamethasone [59]. Remission could be achieved in 78% of all patients treated with the HLH-94 protocol (which included cyclosporin A treat ment only in the continuation phase) [68].

Immunotherapy
A more targeted immunotherapeutic protocol has mainly been used by a single center with similar survival rates compared to the HLH-94 protocol [69]. Th is regimen consists of ATG (rabbit) and methylprednisolone. Th ereafter, cyclosporin A is given until HSCT, generally allowing tapering of methylprednisolone. Patients with central nervous system disease also receive intrathecal methotrexate and corticosteroids. First experiences with alemtuzumab have also yielded promising results [70,71].

Therapy for 'secondary' HLH
Th ere are not suffi cient data to indicate if and which patients with 'secondary' HLH need the full treatment protocol. Although initial treatment with steroids alone or in combination with cyclosporin A may be justifi ed in some patients, the timely use of more aggressive therapy is mandatory for a good outcome. In patients with MAS, immunosuppression with corticosteroids with or without cyclosporin A in most cases leads to a dramatic improve ment of disease within days [72,73]. Etoposide may be added if there is no response or highly active disease [38].
Control of the underlying disease is of key importance in the overall treatment concept. For rheumatological disease this in particular involves the use of IL-1 receptor antagonists and IL-6 antibodies in Still's disease and anti-TNF in some other rheumatological diseases [74,75]. Whether these agents may also trigger MAS is a matter of debate. Of note, they have been shown to be of limited use in the treatment of primary HLH. Control of infectious disease is of similar importance both in primary as well as 'secondary' HLH. EBV-triggered HLH has been treated successfully with rituximab (anti-CD20 antibody) in addition to conventional therapy [27,[76][77][78]. Immuno globulins, which may act against pathological antigens or cytokines, have been used as an adjunct in infection-triggered HLH [1,27].

Hematopoietic stem cell transplantation
HSCT is recommended in genetic cases, in patients with recurrent HLH and disease progression despite adequate therapy [27]. For genetic cases it is the only curative option. To avoid delays in starting HSCT, HLA typing and donor search should be initiated as soon as the diagnosis of primary HLH is established [27]. Due to the highly infl ammatory nature of HLH, myeloablative conditioning has been the standard regimen for many years. However, veno-occlusive disease is a well-known complication when using busulfan-based protocols, occurring in up to 25% of HLH patients [79,80]. Recently, reduced-intensity conditioning regimes have been used successfully and with a much lower incidence of venoocclusive disease [81,82]. Mixed chimerism can be a relevant limitation, but murine studies and some observations in patients indicate that a stable long-term level of about 10 to 15% donor chimerism may be suffi cient to maintain remission [80,81].

Outcome
Despite advances in therapy, up to 40 to 60% of children initially do not respond to treatment and die of HLH or die of infections or complications during therapy [27,83]. Active HLH at the time of HSCT and central nervous system involvement are associated with a worse outcome [84,85]. Th e HLH-94 protocol induced remission or allowed the patient to undergo HSCT in 71% of cases [85]. Patients with a positive family history and who received HSCT after induction therapy according to the HLH-94 protocol had a 5-year survival of 50% [85]. Patients with all forms of HLH who were treated according to the HLH-94 protocol had a 5-year survival of 54%. Th e patient group that underwent HSCT showed a 5-year survival of 66% [85]. Th e ATG-based therapeutic regimes induced remission of HLH in 73% of patients with FHL in one center, and patients had an overall survival after HSCT of 55% [69]. After induction therapy with varying therapeutic regimes, a survival rate of 58.5% 6 years after HSCT has been reported in a single center study [80]. Survival rates of 86% have been reported in EBV-associated 'secondary' HLH after HSCT in Japan Note that the patient only shows 0.17% degranulation, the control 31.17%. Degranulation >15% is considered normal, 5 to 15% abnormal and possibly pathological, requiring additional testing, and <5% defi cient. [86]. Malignancy-related HLH had the worst prognosis of all forms of HLH. A Japanese survey on HLH in all age groups showed a 5-year survival of <15% in malignancyassociated HLH [6]. Long-term complications of HLH encompass therapyrelated morbidity, especially after HSCT, and neurocognitive defi cits. Th e latter can manifest months to years after HLH, but fortunately most patients can return to normal life [87].

Conclusion
HLH is a clinical syndrome of hyperinfl ammation that becomes frequently manifest in patients with certain underlying genetic diseases, but may also occur following particular triggers in patients with no known inherited disorder. Initial symptoms may be non-specifi c and misleading. Th erefore, a high index of suspicion and a thorough clinical, immunological, and genetic work-up is required. Prompt initiation of adequate treatment is the key for survival. However, despite important advances in therapy, the overall survival rates remain unsatisfactory.

Outlook
With increased use of whole genome sequencing approaches, it is likely that the genetic bases of the remaining 10% of familial cases of hereditary HLH will soon be elucidated. Moreover, a potential genetic contribution to 'secondary' HLH is being investigated. Several animal models have been established to further elucidate the pathogenetic basis of the diff erent