Characteristics and Long-Term Outcome of 15 Episodes of Systemic Lupus Erythematosus-Associated Hemophagocytic Syndrome

Abstract

Reactive hemophagocytic syndrome (HS) occurs mainly in the setting of serious infections and lymphomas. HS can occur in the course of 2 active systemic diseases, without simultaneous infection: adult Still disease and systemic lupus erythematosus (SLE). Observations of specific lupus-associated HS are rare, and the long-term outcome of these patients with active SLE is unknown.

We retrospectively studied 15 episodes of SLE-associated HS in 12 patients (10 women, 2 men) and noted the long-term outcome. HS occurred at a mean age of 25 years. All patients were febrile with ≥2 cytopenias, and bone marrow aspiration indicated hemophagocytosis. HS revealed SLE in 9 patients and recurred in 3. The main features of SLE-associated HS were a low frequency of hepatosplenomegaly, a high frequency of heart involvement (5 pericarditis, 4 myocarditis requiring transfer to intensive care unit), and a low C-reactive protein level (mean, 15 mg/L). Cutaneous-mucous symptoms of SLE, arthritis, and nephritis were present respectively in 8 (53%), 6 (40%), and 4 (27%) episodes, but symptoms of SLE were absent in 4 episodes at admission. All patients had anti-nuclear antibodies when the HS occurred. Anti-double-stranded DNA antibodies were present in 12 episodes. Treatment was steroids in 14 cases but cyclophosphamide was the only treatment able to control HS in 2 cases. All the cases of SLE-associated HS were controlled by the immunosuppressive regimen. Intravenous immunoglobulins seemed poorly effective. No infectious agent was found.

Clinical presentations of the 23 patients with SLE-associated HS described in the literature were reviewed and were similar to those of the current series. The mean follow-up was 88 months (range, 7-240 mo). One patient died at 15 months (sepsis). Among the 5 patients with a follow-up >8 years, 4 always had active disease. During the follow-up of SLE, immunosuppressive drugs were added in 8 patients (cyclophosphamide in 7, azathioprine in 3, mycophenolate mofetil in 2) with significant adverse drug reactions. In the long-term, SLE-associated HS seems to define a severe SLE form with frequent flares, possible HS recurrences, and the need for prolonged immunosuppression.

Abbreviations: CRP = C-reactive protein, HS = hemophagocytic syndrome, ICU = intensive care unit, IV Ig = intravenous immunoglobulins, SLE = systemic lupus crythematosus, SLEDAI = Systemic Lupus Erythematosus Disease Activity Index.

INTRODUCTION

Reactive hemophagocytic syndrome (HS) is characterized by the infiltration of morphologically benign hemophagocytic histiocytes in the bone marrow (Figure 1) and various organs, including the lymph nodes, liver, and spleen. Patients usually present with an acute febrile illness, sometimes fulminant and lethal. Common manifestations include high fever, pancytopenia, hepatosplenomegaly, elevated liver enzymes, and high blood triglyceride and ferritin levels. Coagulopathy and central nervous system dysfunction often ensue, and less frequently the lungs and cardiac tissues are involved. Overproduction of pro-inflammatory cytokines, uncontrolled activation of T cells, and macrophages associated with decreased natural killer cell and cytotoxic cell functions seem to be the hallmarks of the immunologic abnormalities in HS^4,10,13.

FIGURE 1: Bone marrow smear showing an activated macrophage (vacuolated cytoplasm) with phagocytosis of an erythroblast (May Grünwald-Giemsa stain, magnification × 640).

This puzzling disorder, identified in 1979²⁵, is currently classified among the reactional histiocytoses. It reflects an inappropriate immune response of the host and is triggered by various events. The constitutive form is familial hemophagocytic lymphohistiocytosis which is secondary to mutations in genes involved in cytotoxicity¹⁸. Secondary HS is described in the setting of infections especially in immunocompromised patients, malignancies (mostly lymphoma), autoimmune disorders, and severe allergic drug reactions^10,13,14. The underlying disease may be ascertained as causative when HS is cured with the specific treatment of the disease, such as chemotherapy in lymphoma.

HS occurring in the course of adult systemic disease has rarely been reported and can be separated into HS associated with an active infection, often a complication of immunosuppressive treatments, and HS specifically associated with a flare of the systemic disease. A 2003 study⁷ has shown that the latter situation may be encountered only with systemic lupus erythematosus (SLE) and adult Still disease. Data on SLE-associated HS, the "acute lupus hemophagocytic syndrome," are limited, and no large series has been described. The first cases were reported by Wong et al³⁶ in 1991 and by our group in 1993⁶. In 2 studies of 50 patients^7,28 focusing on HS associated with systemic disease, lupus was involved in 16 cases, although in 1 of the studies, by Dhote et al⁷, infection was the trigger factor in 7 cases. Moreover, no data are available about the long-term outcome of the patients and the subsequent course of SLE.

We report here the long-term outcome of the 3 SLE patients reported in 1993⁶ and describe 9 other patients with a total of 15 episodes of SLE-associated HS, providing to our knowledge the largest series of acute lupus HS. HS related to infections in patients with SLE were excluded. The characteristics of these 15 episodes were compared with the 23 observations of SLE-associated HS reported in the literature^{7,8,11,12,20,22,30,36}.

PATIENTS AND METHODS

We collected clinical data on cases of SLE-associated HS observed over a 20-year period in the Departments of Internal Medicine of the 3 University Hospitals in the south-east of Paris (Bicêtre Hospital, Mondor Hospital, and Pitié-Salpêtrière Hospital). Over this 20-year period, about 1700 patients with SLE were treated in these departments. One author (HH) provided an additional case.

A case of lupus-associated HS had to meet 3 criteria. First, HS was diagnosed in febrile patients with otherwise unexplained cytopenias and identifiable hemophagocytic histiocytes in bone marrow, liver, or lymph node biopsies. Second, the patient presented with SLE defined according to international criteria (American College of Rheumatology 1997 criteria)⁹, but the criteria could be complete only after the first episode of HS. Third, the HS was effectively treated with specific lupus treatment, and extensive screening for underlying infections was negative.

Clinical and laboratory data at the time of the HS were recorded as well as its course and the treatment received. For hematologic and biochemical values, the nadir or peak values were recorded. The course of the lupus was also recorded with clinical and immunologic data. Anti-double-stranded DNA antibodies were quantified, according to the center and the time of sampling, either with indirect immunofluorescence on Crithidia luciliae or using the Farr test. The minimum and maximum Systemic Lupus Erythematosus Disease Activity Index (SLEDAI) scores during the clinical course were calculated. We separated the treatment for lupus and the treatment given to control the HS. For steroid therapy, we recorded whether patients received intravenous pulses of steroids, whether they received more than 2 months of steroids at a dose above 60 mg/d, and the length of time spent on more than 0.5 mg/kg per d of prednisone. The amount of cyclophosphamide was also recorded, as were all the immunosuppressive therapies received.

We completed the study with a systematic review of the published cases of HS occurring in the course of SLE. The MEDLINE database (National Library of Medicine, Bethesda, MD) was searched using a strategy that included the following medical subject headings: "hemophagocytosis," "hemophagocytic histiocytosis," or "macrophage activating or activated syndrome." The subheadings were "lupus" or "systemic disease." When "lupus" was in the title, the command "related articles" was used. This screening was supplemented by a manual search of references in the articles. Well-described cases of specific lupus-associated HS (not triggered by an infection) were included.

RESULTS

Seventeen episodes of lupus-associated HS were recorded in 14 patients over a 20-year period, but complete data were obtained from 12 patients. Patients previously reported⁶ are Patients 1, 2, and 3.

Clinical Features of HS

Tables 1 and 2 show the clinical characteristics of the 15 episodes of lupus-associated HS that occurred in 12 patients. Three patients (Patients 1, 3, 5) had recurrent lupus-associated HS (each episode is represented in the table as a and b). The mean age at the time of HS diagnosis was 25 ± 9 years (range, 11-40 yr). All patients had fever. Acute onset of fever was observed in all patients except Patient 3, who had low-grade fever for 1 month. Recent weight loss was present in 80% of patients. Hepatosplenomegaly was rare: hepatomegaly was present in only 2 episodes (13%), splenomegaly in 4 (27%). Superficial lymphadenopathy was present in 11 episodes (73%).

TABLE 1: Clinical Manifestations in 12 Patients With SLE-associated HS

TABLE 2: Clinical and Biological Manifestations at the Onset of SLE-Associated HS

Clinical manifestations suggestive of SLE such as polyarthralgia/polyarthritis, cutaneous vasculitis, or malar rash were present respectively in 6 (40%), 4 (27%), and 2 (13%) episodes of HS. Kidney involvement with glomerular nephritis was present in 27% with isolated proteinuria in Patient 2, and stage III-IV nephritis (biopsy proven) in 3 episodes (20%). Heart involvement was symptomatic in 8 (53%) HS episodes with pericarditis in 5 (33%) and myocarditis in 4 (27%). Five episodes required patient admission to the intensive care unit (ICU), with myocarditis in 4 and anasarca in 1. Lung involvement was present in 6 (40%) episodes but was clearly related to lung edema in the setting of myocarditis in 4 cases. In 4 episodes, no symptom suggestive of SLE was present when the patient was admitted.

Laboratory Features of HS

Tables 2 and 3 report laboratory data at the time of HS. Neutropenia (≤1.5 × 10⁹/L) and anemia with hemoglobin less than 11 g/dL were present, respectively, in 13 (87%) and in all cases of HS. Lymphopenia (<1 × 10⁹/L) was present in all but 1 episode. Thrombocytopenia was also frequent: <150 × 10⁹/L in 12 episodes (73%). According to these criteria, all patients had depression of at least 2 cell lines. Activation of coagulation was less frequent, with only 2 cases with fibrinogen level less than 2 g/L. Liver cytolysis was present in 11 cases (73%), usually moderate at presentation. Cholestasis with normal bilirubin level was present in 6 cases (40%). Lactate dehydrogenase level was increased in 82%, and blood ferritin level was always elevated in the episodes in which it was measured (mean, 8509 μg/L). Triglyceride levels, quantified in 10 episodes, were increased in 7. Despite the high fever and the "sepsis-like" presentation of some patients, the C-reactive protein (CRP) level was always normal or low: less than 30 mg/L except in 1 patient with pericarditis.

TABLE 3: Laboratory Data at the Time of HS^†

Extensive virologic and bacteriologic investigations were performed for all patients with no demonstrated active infection. Repeated blood cultures, urine and stool cultures, and serologic testing for herpes virus, cytomegalovirus, Epstein-Barr virus, herpes 6 virus, parvovirus B19, and hepatitis B and C virus were usually performed.

At presentation, all patients had positive antinuclear antibodies, 12 (80%) had anti-double-stranded DNA antibodies, and 9 (60%) had a C3 fraction decrease. Hemophagocytosis was seen on bone marrow aspiration in all patients except Patients 3 and 5, in whom the diagnosis was made at the second episode of HS. Hemophagocytic cell infiltration was not quantified in the bone marrow in most cases. Hemophagocytosis was also present in lymph node biopsies in Patients 1 and 2; in bone marrow biopsies in Patients 3, 5, and 11; and in liver biopsies in Patient 11.

Immediate Outcome of HS

Table 4 shows the outcome of the HS. All patients survived the HS but in 5 episodes (33%), the patient was admitted to the ICU. Five patients (33%) required red blood cell concentrates, and 1 patient required platelet concentrates. Empirical antibiotic therapy was administered in 9 of the 12 first HS episodes but not when HS recurred. Antituberculous treatment was given empirically in 3 episodes, and antiviral therapy in Patient 11.

TABLE 4: Treatment and Immediate Outcome of the Lupus-Specific HS

In 6 episodes, patients received high-dose intravenous immunoglobulins (IV Ig), given alone in 4 episodes (1 g/kg per d, 2 d). Transient efficacy (disappearance of fever) was observed in Patient 6. IV Ig were ineffective in Patients 7, 11, and 12. Clinical condition and cytopenia did not improve, and myocarditis in Patient 12 worsened, leading to admission to the ICU 6 days later. All patients except 1 received corticosteroids (prednisone, with pulse methylprednisolone in 9 episodes) with initially good results, but immunosuppressive therapy was required in 2 patients because of insufficient control of the HS (Patients 10 and 11). Cyclophosphamide was the most effective drug in controlling HS and SLE in these 2 patients in whom steroids were insufficient. In Patient 11, who was critically ill, a delayed effect of rituximab and cyclosporine was also possible. In 5 other patients (Patients 1, 2, 3, 4, 5), cyclophosphamide was rapidly added as baseline treatment for severe SLE.

At the time of the first HS episode, only Patient 6 was receiving immunosuppressive therapy (steroid) but azathioprine had been stopped 1 month earlier. Relapse of HS was observed in 3 patients (Patient 1, 3, 5). Relapse occurred 4 years after the first episode in Patient 1, despite 35 mg/d prednisone, and 14 months after the first episode in Patient 3, despite 10 mg/d prednisone. Recurrent HS led to rapid introduction of cyclophosphamide in these patients. Minimal relapses with febrile moderate cytopenias occurred later several times in Patient 3 with good response to steroid dose increase. In Patient 5, immunologic testing had not been performed at the first HS episode; after initial spontaneous improvement, the HS relapsed leading to the SLE diagnosis.

SLE at Diagnosis

HS and SLE were diagnosed simultaneously in 8 patients (see Table 1). SLE was diagnosed after the first HS with the onset of other symptoms in Patient 3, and at the recurrence of HS in Patient 5. In Patient 8, the initial diagnosis following lymph node biopsy was Kikuchi disease (a histiocytic necrotizing lymphadenitis, usually a self-limiting disease), but the onset of malar rash and immunologic abnormalities led to diagnosis of SLE. SLE was diagnosed before HS only in Patient 6. HS was the first manifestation of SLE in 9 patients. In the 3 other cases, autoimmune manifestations preceded the HS: isolated thrombocytopenia 7 years before SLE diagnosis in Patient 3, polyarthritis in Patient 9, and polyarthritis, vasculitis, and stage IV nephritis in Patient 6.

All patients had 4 or more American College of Rheumatology criteria for SLE during disease progression. Table 5 presents the clinical and immunologic symptoms of SLE with their time of occurrence. At the time of SLE diagnosis, the main symptoms were malar rash/cutaneous vasculitis in 9 patients (75%); polyarthritis in 8 (67%); nephritis in 6 (50%), with 4 kidney biopsies performed showing stage IV nephritis; cytopenias in all patients except Patient 6; and pericarditis in 6 (50%) patients. All patients were febrile except for Patient 6. Mean SLEDAI at diagnosis was 17 (range, 10-34).

TABLE 5: SLE Description and Long-Term Outcome in 12 Patients

Markedly elevated titers of antinuclear antibodies were found in all patients; anti-double-stranded DNA antibodies were present in 11 (92%) patients (see values reported in Table 3). Serum levels of the C3 and C4 complement fractions were decreased in 9 (75%) patients. The Coombs test was positive in 7 (58%) patients. Other autoantibodies were found at SLE diagnosis in 8 patients (see Table 5), mainly anticardiolipin IgG antibodies in 6 (50%) patients.

Autoimmune diseases with 2 SLE were reported in the family in 3 patients, all of Asian origin. We have previously reported familial SLE¹⁹.

SLE Course

Table 5 shows the course and long-term outcome of SLE. Treatment of SLE consisted of corticosteroids in all patients, with methylprednisolone pulses in 10 patients, hydroxychloroquine in 7, and immunosuppressive drugs required in 8 patients. Cyclophosphamide was used in 7 patients, azathioprine in 3, mycophenolate mofetil in 2, and rituximab in 2.

Mean follow-up was 88 months but follow-up of more than 8 years was available in 5 patients. One patient was lost to follow-up and 1 died of sepsis 15 months after SLE diagnosis. Considering the 5 patients with the longest follow-up, repeated SLE flares occurred in 4 (Patients 1, 2, 3, 6). At the last visit, these 4 patients still had active disease requiring significant doses of steroids and immunosuppressive drugs. Symptomatic antiphospholipid syndrome with anticardiolipin antibodies developed during follow-up in Patients 1 and 2. However, in all patients, except Patient 3, cytopenias resolved during the follow-up and seemed to be mainly related to the HS.

The mean maximal SLEDAI score quantified during the disease course was 22. SLE activity led to the use of high steroid doses. The 7 patients with a follow-up of more than 4 years spent a mean of 15 months on a daily prednisone dose of more than 0.5 mg/kg, and 5 had periods of more than 2 months on more than 60 mg prednisone/d. High cyclophosphamide doses were used in 4 of these 7 patients, azathioprine in 3, mycophenolate mofetil in 2. Adverse drug reactions were significant among these patients: obesity, hypertension, vertebral collapse, head femoral avascular necrosis with significant morbidity (3 cases). Three patients had severe infections (pneumococcal pneumonia, disseminated zona [2 cases]², and cytomegalovirus infection).

Antinuclear antibodies remained positive in all patients, as did anti-double-stranded DNA antibodies except for Patient 8, in whom they became negative. C3 complement fraction level was variable and correlated well with the flares. Anticardiolipin antibodies became detectable in 3 patients during the follow-up.

Lupus-Associated HS in the Literature

Two main series have focused on HS in systemic diseases^7,28: the study by Dhote et al⁷ in adults, and the study by Stephan et al²⁸ in children in whom Still disease was prevalent. Dhote et al reported 7 cases of SLE-associated HS among 26 cases of adult systemic disease, while Stephan et al reported 2 children among 24 cases. Apart from our first report, 2 series have specifically focused on SLE-associated HS: the 6-patient series of Wong et al³⁶, which was to our knowledge the first description of this entity, and 3 cases reported by Papo et al²². Among the 378 patients of the 10 largest series of HS in the literature^{1,8,12,23,25,26,30-32,35}, 17 cases of HS occurred in SLE patients. However, only 5 cases of SLE-associated HS were detailed. Some isolated cases^11,20 have also been reported. In summary, we have found in the literature 23 detailed cases of SLE-associated HS, which are summarized in Table 6.

TABLE 6: Specific SLE-Associated HS, Previous Reports^*

The clinical presentation of the patients described in the literature and in the current series is similar: patients with SLE-associated HS were febrile with infrequent hepatosplenomegaly. Lymphadenitis might be present but was not a constant feature. Symptoms suggestive of SLE such as skin rash, pericarditis, or polyarthritis were often present. Laboratory abnormalities such as cytopenia were unremarkable. Steroids were used in most cases; high-dose IV Ig, when used, seemed to be not sufficient to control the HS, and cyclophosphamide was necessary in 2 cases. Immediate outcome of the HS was favorable in most cases, but 2 patients died^22,35. No data about the long-term outcome of SLE were available in the literature.

Infection-Associated HS in Patients With SLE

During the study period, 4 patients with SLE had HS due to infection. Their clinical and laboratory characteristics are described in Table 7 for comparative purposes. The major point is the patient's history: infection-associated HS occurred in all cases in patients with known lupus, 3 of whom received immunosuppressive drugs. The activity of the SLE was very low in 2 patients. No particular clinical or laboratory finding was clearly distinctive in the infection-associated HS versus lupus-associated HS, except for CRP which can be high in some infections. Herpesviridae were involved in 3 cases.

TABLE 7: Infection-Associated HS in 4 patients with SLE

Infection-associated HS is life threatening because infection can trigger a lupus flare requiring strong immunosuppressive treatment as in Patient A, can lead to end-stage organ damage because immunosuppressive drugs are withdrawn (Patient C), or can occur in septic patients thus worsening the prognosis (Patient D). The effective treatment is both to treat the microorganism involved and to reduce the immunosuppressive treatment, but transiently, as the risk of lupus flare after infection is high. IV Ig might be useful in this setting.

DISCUSSION

HS is a life-threatening condition which is usually associated with lymphoma or infectious diseases. It has been described infrequently in the course of systemic diseases. In this setting, HS may either complicate the underlying disease or result from an infection, which frequently occurs in immunocompromised patients. In the 10 largest reported series^{1,8,12,23,25,26,30-32,35} of HS, systemic diseases represent only 0-5% of the underlying disorders, with SLE as the leading disease. We have found in the literature 23 detailed cases of SLE-associated HS in which an infectious trigger has not been identified (see Table 5) and 15 cases^{5,7,16,22,23,27,32} of HS in SLE with an infection identified as a trigger factor. We report here what is, to our knowledge, the largest study to date on SLE-associated HS, providing the description of 15 episodes and the outcome of SLE in 12 patients.

Wong et al³⁶ reported an incidence of SLE-associated HS of 6 cases during a 3.5-year period among 250 SLE patients. This incidence is low but the rarity of SLE in the large series of HS is striking. SLE-associated HS might be underdiagnosed. Cytopenias are common and may have various origins in SLE. Bone marrow aspiration and/or biopsy may not be systematically performed in a febrile, cytopenic lupus patient. Moreover, bone marrow aspirates may be falsely negative and not indicate hemophagocytosis⁷. It is noteworthy that SLE was diagnosed simultaneously with HS in 8 patients (67%) in our series, as also noted in the literature⁷. Pancytopenia with high ferritin level, uncommon in SLE, is highly suggestive of HS.

On the other hand, HS may lead to a delay in SLE diagnosis. Clinical symptoms suggestive of SLE were not always seen at presentation in our patients. Patient 8 was diagnosed with Kikuchi disease, a condition that has been associated with HS¹⁷, whereas culture-negative tuberculosis was suspected in Patient 5, leading to ineffective treatment and HS recurrence with lupus kidney involvement. However, antinuclear antibodies were found in all patients screened, indicating systemic disease. Immunologic testing should therefore be systematically performed in the setting of HS.

The characteristics of the HS itself may be informative. Contrasting with other diagnoses such as infections or lymphoma, hepatomegaly and splenomegaly were rare in our series, present in 13% and 27% of cases, respectively. In the literature survey of SLE-associated HS, they were reported in respectively 36% and 22% of cases. This contrasts with the 88% and 77%, respectively, reported in the study by Takahashi et al³⁰ of HS associated with lymphomas. In the main studies of hemophagocytosis grouping infection and lymphoma-associated HS^1,23,30, the frequency of hepatosplenomegaly is always above 50%-60%. On the other hand, lymphadenitis was frequent among our patients, but this may also be the case in patients with lymphoma. It should be underlined that lymph-node cytology and biopsy are useful when seeking evidence of hemophagocytosis and in ascertaining or excluding the diagnosis of blood disease. This point should be stressed as the incidence of lymphoma is increased in SLE³.

The high incidence of heart involvement in SLE-associated HS, that is, 4 episodes of myocarditis (27%) and 5 of pericarditis (33%), is striking. Pericarditis is a frequent feature of SLE. Myocarditis may be present both in SLE flare and in HS. Within the former, myocarditis, although well described in postmortem studies, is rarely symptomatic-less than 10% of cases in recent clinical studies³³. But in some patients, acute myocarditis complicated by heart failure may be the presenting clinical manifestation of SLE³³. In HS, myocarditis is very rare in adults²⁴ but seems to be more frequent in children^5,21. Therefore, HS with heart involvement might be indicative of a systemic disease, at least in adults.

All patients had cytopenias, high ferritin level, a high frequency of liver cytolysis, and hemophagocytosis. We have not found a specific biologic feature for SLE-associated HS except for the low serum CRP level contrasting with the high fever. This point should be underlined, as a high CRP level is strongly indicative of infection (see Table 7 and reference 15).

Prognostic factors of adult HS have been identified¹²: low hemoglobin and platelet values on admission, disseminated intravascular coagulation, jaundice, and increased ferritin level were associated with death. All patients of the study had at least 2 cytopenias and elevated ferritin levels, but no patient died from the HS. Disseminated intravascular coagulation was absent, and a very low (<20 × 10⁹/L) platelet count was present in only 1 episode. No patient had jaundice. Is specific HS in SLE less severe than HS in lymphoma, for instance? The literature reports the death of 2 (4%) patients with SLE-associated HS, and 4 of our patients were admitted to the ICU because of the HS. SLE-associated HS is therefore a life-threatening condition, but the good treatment response should be underlined. HS and SLE were both controlled in all cases with steroids and/or immunosuppressive drugs.

It is important to note that in 2 patients in our series, and in 2 in the literature, cyclophosphamide was the only treatment able to control both SLE-associated HS and SLE itself. Therefore, cyclophosphamide should be used as soon as steroids seem to be insufficient. High-dose IV Ig have been suggested to be active in some cases of HS but their efficacy remains controversial^5,8. Though limited, our data do not support their use in SLE-associated HS. Indeed, IV Ig were either ineffective or transiently and/or incompletely effective, and steroids were required in all cases. However, when an infection is suspected in a patient with SLE under immunosuppressive treatment developing HS, IV Ig seems to be useful (see Table 7). Their role in HS in the lupus setting remains to be determined.

Wong and colleagues³⁶ and our group⁶ described SLE-associated HS 15 and 13 years ago, respectively, but there are no literature data about the long-term outcome of SLE. We report here these data with more than 15 years of follow-up for the 3 cases presented in our first report. SLE revealed or complicated by a specific HS is severe and can remain active for years despite prolonged immunosuppressive therapy. Eight patients received immunosuppressive treatment in addition to steroids. Several drugs were used in 5 patients. For 5 patients, more than 8 years have elapsed since diagnosis. Four patients still have active disease with flares requiring cyclophosphamide or mycophenolate mofetil. In these patients, severe infections and bone complications illustrate the adverse reactions to these treatments. High-dose prednisone (≥60 mg/d for ≥2 mo) and repeated methylprednisolone pulses were associated³⁷ with an increased risk of avascular necrosis and osteoporotic fracture as observed in our studies. The initial HS-related cytopenias resolved and no patient developed other serious hematologic complications in the course of SLE. As previously shown², immunologic abnormalities increased over time in some patients, with the appearance of new autoantibodies.

The sequence of events leading to HS is incompletely understood^10,13, and hypotheses are mainly derived from observations of primary hemophagocytic lymphohistiocytosis¹⁸. The inability of NK and CD8 T cells to efficiently terminate an immune response leads to sustained activation of lymphocytes and macrophages. The result is widespread hemophagocytosis and overproduction of cytokines such as interferon γ, TNF α, IL-6. We had previously shown⁶ a large production of TNF α and IL-6 in SLE-associated HS, but the levels were within the same range in patients with active SLE. Elevation of plasma interleukin 18 concentration was correlated with disease activity in SLE³⁴ and was also increased in patients with hemophagocytic lymphohistiocytosis²⁹ with a correlation with interferon γ levels. This cytokine might therefore play a significant role in SLE-associated HS. SLE by itself, during a flare, might be able to switch on the immune system.

In practice, HS should be considered in a febrile cytopenic patient with SLE and should be carefully investigated by bone marrow aspiration. Determination of blood ferritin and triglyceride levels is useful. Infection should be screened for especially in inactive SLE and is related to iatrogenic immunosuppression, which should be transiently reduced. SLE-associated HS occurs when SLE is active and may be the presenting manifestation of SLE. SLE should be systematically suspected in the setting of an unexplained hemophagocytosis syndrome, especially in a young woman. Given that suggestive clinical symptoms of SLE may be absent at admission, serologic tests (ANA and anti-double-stranded DNA antibodies) are mandatory. The presence of heart involvement is indicative of a systemic disease (SLE, but also Still disease). Onset of the disease may be the only factor triggering HS. Steroids are the cornerstone of the treatment but cyclophosphamide may be necessary. In the long-term, SLE-associated HS seems to define a severe subset of SLE with repeated flares, possible HS recurrences, and a frequent requirement of prolonged immunosuppression.