To assess the efficacy of anakinra treatment in patients with adult-onset Still's disease (AOSD) that is refractory to corticosteroids, methotrexate (MTX), and etanercept.
To assess the efficacy of anakinra treatment in patients with adult-onset Still's disease (AOSD) that is refractory to corticosteroids, methotrexate (MTX), and etanercept.
Four patients with AOSD were treated with prednisone and MTX and 2 patients were also treated with etanercept for worsening symptoms and indicators of systemic inflammation. White blood cells (WBCs), C-reactive protein (CRP) levels and/or erythrocyte sedimentation rate, and ferritin levels were measured and, in 1 patient, serum creatinine levels were determined. Treatment with anakinra at 100 mg/day was initiated.
The index patient's disease was refractory to treatment with prednisone (30 mg/day) and MTX, with spiking fevers, rash, synovitis, a serum ferritin level of 8,400 ng/ml (normal ≤200), and a CRP level of 86 mg/liter (normal <8). Levels of interleukin-1β (IL-1β), IL-1α, IL-6, IL-1 receptor antagonist, and IL-18 were elevated. Just prior to anakinra treatment, the WBC count was 14,600/mm3, the CRP level was 86 mg/liter, and the ferritin level was 573 ng/ml, with daily spiking fevers to 104°F, rash, and swollen joints. Within hours of the first injection, the patient was afebrile and asymptomatic; within days, the WBC count, ferritin level, and CRP level decreased into the normal range. On 2 occasions, anakinra was withheld. Within a few days, the WBC count rose to >20,000/mm3 with prominent neutrophilia, the CRP level rose to >200 mg/liter, and the ferritin level rose to >3,000 ng/ml. Upon restarting anakinra, the patient became afebrile, the WBC count fell to 8,000/mm3, the CRP level fell to <3 mg/liter, and the ferritin level fell to <300 ng/ml. Three additional patients with refractory AOSD who experienced rapid reductions in fever, symptoms, and markers of inflammation when treated with anakinra are reported.
Refractory AOSD appears to be IL-1–mediated since anakinra decreases hematologic, biochemical, and cytokine markers and also produces rapid reductions in systemic and local inflammation. Reported efficacy of tumor necrosis factor–blocking therapies in AOSD may be due to a reduction in IL-1.
Adult-onset Still's disease (AOSD), a rheumatologic condition found worldwide, is characterized by a variety of clinical features, including intermittent fever, arthritis, evanescent rash, sore throat, leukocytosis, and hepatic dysfunction, polyserositis, lymphadenopathy, splenomegaly, and other systemic symptoms. Because of the diverse presentation (1), classifications have been developed to standardize the diagnosis of AOSD. The most widely accepted criteria set, as presented by Yamaguchi and colleagues (2), is a compilation of major and minor criteria with the exclusion of infections, malignancies, and other rheumatic or systemic diseases. More recently, Fautrel and associates have proposed classification criteria utilizing diagnostic markers of serum ferritin and glycosylated ferritin, thought to be more specific for AOSD (3). These criteria are said to provide a sensitivity of 80.6% and a specificity of 98.5%.
The response to the prognosis and therapy of AOSD can be as varied as its clinical presentation. In the series reported by Pouchot et al (4), the course of AOSD was categorized as self-limited in 34% (median time to remission 9 months), intermittent in 24% (recurrent flare at a median of 54 months), and chronic in 36% (at least 1 episode lasting >1 year). Severe, erosive polyarthritis is not uncommon in AOSD, and death in association with systemic features such as hepatic failure, amyloidosis, infection, and disseminated intravascular coagulation as well as progression to macrophage activation syndrome has been reported. The etiology of AOSD remains unknown, but viral infection, including Epstein-Barr virus (5), human herpes virus 6 (6), and rubella (7), has been suspected in its pathogenesis.
Several clinical aspects as well as biochemical markers of inflammation observed in AOSD are similar to those of the systemic inflammatory response syndrome (8). Fever, leukocytosis (neutrophilia), and hepatic acute-phase protein synthesis are prominent in AOSD and, similar to findings in patients with systemic inflammatory response syndrome, are associated with elevated levels of serum cytokines, such as interleukin-1 (IL-1), tumor necrosis factor α (TNFα), IL-6, and, more recently, IL-18 (9). However, in any disease manifestation, cytokine causality is determined by specific blockade and not by correlations of a particular cytokine level with disease severity. It is apparent that TNFα plays a role in several manifestations of AOSD since patients whose disease is refractory to standard treatments, including corticosteroids and methotrexate (MTX), exhibit a rapid response upon intravenous neutralization of TNFα using infliximab (10–15). We now report that, in patients whose AOSD was refractory to conventional therapies, the blockade of IL-1 receptors with the IL-1 receptor antagonist (IL-1Ra) anakinra resulted in a rapid resolution of systemic and local manifestations of the disease within hours and days of the initial subcutaneous injection.
Serum was kept frozen at −20°C, and then thawed, aliquoted into smaller volumes, and refrozen at −70°C until assayed in duplicate. The average of the duplicates is reported. IL-1β was measured using the enhanced chemiluminescence (ECL) assay (BioVeris, Gaithersburg, MD), as previously described (16). For this assay, anti-human IL-1β antibodies were purchased from R&D Systems (Minneapolis, MN). The limit of detection of the ECL assay was 10–20 pg/ml, and levels in sera from healthy subjects were below the detection limit. A known amount of recombinant human IL-1β (PeproTech, Rocky Hill, NJ) was added to 1 aliquot of the index patient's serum in order to test for recovery or for the presence of interfering substances. An IL-1α ECL assay was performed as previously described (17), using anti-human IL-1α (BAF 200 and AF-200-NA; R&D Systems). The detection limit was 10 pg/ml and in sera from healthy subjects, the levels were consistently below the detection limit.
IL-6 was measured using an ECL assay and antibodies purchased from R&D Systems (BAF 206 and MAB 206). Serum IL-6 levels in healthy subjects ranged from 10 to 200 pg/ml. Serum IL-18 was measured using the ECL assay, as previously described (18). Using this assay, the mean ± SEM measurement of serum IL-18 levels in healthy controls was 64 ± 15 pg/ml (18). Endogenous IL-1Ra was measured using an ECL assay with a sensitivity of 50 pg/ml, with biotinylated and ruthenylated antibodies obtained from R&D Systems (BAF 280 and AF-280-NA). Serum IL-1Ra levels in healthy subjects ranged from 150 to 300 pg/ml. Levels of all other serum reactants and enzymes were determined by the respective hospital laboratory services.
A 39-year-old Korean woman developed intermittent fever, rash, arthralgias, weight loss, and sore throat in May 2001. She was referred to a rheumatologist (AAF) in September 2001 for assessment of a possible connective tissue disease as a cause for fever of unknown origin, after findings of an evaluation for infectious agents were negative. Results of the initial examination were normal. Laboratory results included the following: hemoglobin level 11.3 gm/dl, white blood cell (WBC) count 8,300/mm3, C-reactive protein (CRP) level 47 mg/liter (normal <8.0), antinuclear antibody (ANA) negative, rheumatoid factor (RF) negative, anti–streptolysin O (ASO) negative, antineutrophil cytoplasmic antibody (ANCA) negative, and normal findings on chest radiography and transthoracic echocardiography. At that time, the serum ferritin level was modestly elevated at 442 ng/ml (normal 12–200). A diagnosis of AOSD was considered, but the patient continued treatment with antiinflammatory medications. Sera were obtained and later assayed for cytokine levels. Two months later, while visiting in Korea, the woman had an exacerbation with fever, rash, polyarthritis, and an elevated serum ferritin level of 8,400 ng/ml. AOSD was diagnosed and prednisone 30 mg/day was started. Prednisone improved the fever, arthritis, and rash and, 6 months later, the ferritin level had decreased to 686 ng/ml.
Upon returning to Canada in May 2002, however, the patient redeveloped intermittent fevers, a maculopapular rash on her arms, and synovitis of her hands and wrists while taking prednisone 25 mg/day. At that time, the hemoglobin level was 10.2 gm/dl with a WBC count of 14,400/mm3 (84% neutrophils), a serum ferritin level of 543 ng/ml, and a CRP level of 116 mg/liter. Oral MTX (15 mg/week) was added at that time for improved disease control and steroid sparing. By September 2002, the patient was receiving MTX 25 mg/week and prednisone 20 mg/day but with only partial improvement. A decision was made to discontinue MTX and start anakinra. Prior to the first injection of anakinra, the hemoglobin level was 10.7 gm/dl with a WBC count of 14,600/mm3 (86% neutrophils), the CRP level was 86 mg/liter, and the serum ferritin level was 573 ng/ml. The clinical course with biochemical markers and cytokine levels is shown in Figure 1.
The first injection of anakinra (100 mg) was administered on December 16, 2002. Overnight, the patient had profuse sweating, and by the following day she experienced no further fevers, rash, or joint pain. Her husband, who kept detailed records of his wife's illness and laboratory results, documented all temperature readings (Figure 2). Five weeks following the start of anakinra, the patient remained asymptomatic, with a hemoglobin level of 11.6 gm/dl, a WBC count of 5,700/mm3 (68% neutrophils), a CRP level of 9.6 mg/liter, and a serum ferritin level of 170 ng/ml.
The patient remained healthy until March 2003, when she presented with shortness of breath, cough, fever, and right middle and lower lobe interstitial pulmonary infiltrates. She was hospitalized for evaluation. Transbronchial biopsy results were normal and bacterial cultures were negative. A presumptive diagnosis of viral pneumonia was made. As shown in Figure 3, anakinra was withheld pending pulmonary investigations and the patient was treated with prednisone 15 mg/day. Within 4 days of stopping anakinra, the patient had a clinical relapse of AOSD with arthritis and rash. Her WBC count increased to 21,400/mm3 (83% neutrophils), the CRP level to 185 mg/liter, and the serum ferritin level to 1,618 ng/ml (Figure 3). Alanine aminotransferase (ALT), alkaline phosphatase (AP), and lactate dehydrogenase (LDH) levels were mildly elevated.
After a 14-day lapse, daily anakinra was restarted. Within 4 hours of administration, the patient became afebrile, and arthritis and rash resolved within 1 day. Three days later, the WBC count decreased to 8,000/mm3 and the CRP level to 56 mg/liter. Within 3 weeks, her WBC count was 8,800, her CRP level was 13 mg/liter, and her serum ferritin level was 2,156 ng/ml. Chest radiographs and computed tomography (CT) showed improvement over several weeks, but the patient continued to experience dyspnea. Her pulmonary artery pressure was mildly elevated, with a reduction of diffusing capacity for carbon monoxide. Dyspnea persisted and the pulmonary artery pressure increased over the next 4 months, and the patient was diagnosed as having idiopathic pulmonary hypertension. Once again, anakinra was discontinued pending investigations. Within days, spiking fevers, evanescent rash, arthritis, leukocytosis, and increased levels of acute-phase reactants developed and elevated IL-6 levels returned, each of which decreased within days upon reinstitution of anakinra (Figure 3). Liver function abnormalities continued. Prednisone was tapered to 7.5 mg/day. The patient was started on bosentan for pulmonary hypertension in August 2003.
Subsequently, in October 2003, the patient developed shingles. In December 2003, she developed fever and further increases in her AP and LDH levels. Liver biopsy results suggested steatosis and possible autoimmune hepatitis. Prednisone was increased to 15 mg/day. By late April 2004, the patient was experiencing high fevers, increased dyspnea, and continued abnormalities on liver function tests with an increased ferritin level of 7,127 ng/ml, a CRP level of 161 mg/liter, hypofibrinogenemia, and thrombocytopenia of 104,000 × 109/liter. The patient was critically ill, with ascites and a pericardial effusion. Because of these clinical and laboratory features, as well as elevated serum IL-18 levels, macrophage activation syndrome was suspected. Bone marrow tests revealed hypercellularity and rare hemophagocytosis. Intravenous and then oral cyclosporine with dexamethasone at 10 mg/day was added to the anakinra treatment, after which the ferritin and CRP levels fell dramatically. The patient improved within days except for the dyspnea and pulmonary hypertension. Steroid therapy was reduced to prednisone 20 mg/day. The patient is stable on a regimen of cyclosporine (150 mg twice daily), anakinra 100 mg/day, prednisone 5 mg/day, and sildenafil for pulmonary hypertension.
The progression from AOSD to macrophage activation syndrome is well documented (19), particularly in children with Still's disease. The association of IL-18 with macrophage activation syndrome has recently been reported (20) and is highly consistent with the elevated levels of IL-18 in this patient.
In November 2001, a 17-year-old girl of Pakistani origin developed a diffuse, pruritic, eczematous rash with arthralgias and myalgias a few weeks after a hepatitis B vaccination. Although these resolved in 2 months, 4 months later she developed a diffuse rash, arthralgias, sore throat, weight loss, and fever. At that time, the hemoglobin level was 10.1 gm/dl with a WBC count of 9,200/mm3 (80% neutrophils), and a Westergren erythrocyte sedimentation rate (ESR) of 95 mm/hour (normal 0–20). ANA was negative and protein electrophoresis revealed a nonspecific inflammatory pattern. Upon examination, minor cervical lymphadenopathy and synovitis of the wrists and knees were present. When the patient was examined by a rheumatologist (AY) in September 2002, the symptoms had improved after treatment with standard cyclooxygenase inhibitors. However, during the next month, fever spikes to 104°F, joint pains, left upper quadrant tenderness, and a palpable spleen tip developed.
The patient was admitted for investigation of fever of unknown origin. Microbial cultures, including a bone marrow aspirate, were negative. Chest CT revealed bilateral pleural effusions, a pericardial effusion, and a minor interstitial infiltrate at the base. Abdominal CT showed some ascitic fluid and mild hepatosplenomegaly. The hemoglobin level was 7.7 gm/dl, the WBC count was 28,700/mm3 (70% neutrophils), and the ESR was 95 mm/hour. The serum ferritin level was >1,650 ng/ml. RF, ANA, extractable nuclear antigen (ENA), and anti-DNA antibodies were negative. The patient was treated with intravenous methylprednisolone followed by prednisone 60 mg/day and was discharged with a diagnosis of AOSD.
The lichenified rash and arthritis persisted, and oral MTX (12.5 mg/week) was added to the regimen. After a few weeks, the rash resolved, the arthritis improved, and prednisone was tapered to 10 mg/day. With the tapering of prednisone, the arthritis worsened and the dosage of MTX was increased to 17.5 mg/week. Etanercept (25 mg twice weekly) was added to the regimen of weekly MTX and prednisone 10 mg/day. Despite treatment with this combination for 10 weeks, the arthritis persisted. MTX and etanercept were discontinued and subcutaneous administration of anakinra 100 mg/day was started. Before anakinra was instituted, the WBC count was 16,900/mm3 (84% neutrophils) and the ESR was 73 mm/hour. Within 2 weeks, there was a dramatic improvement in joint pain and swelling. Four weeks after institution of anakinra, the hemoglobin level was 11.2 gm/dl with a WBC count of 7,300/mm3 and a normal differential cell count. The ESR was 19 mm/hour. Prednisone was discontinued in October 2003, 3 months after initiation of anakinra.
In November 2003, after 4 months of anakinra treatment, the patient developed a flulike illness with fever, nausea, vomiting, and diarrhea with bilateral pulmonary infiltrates. She was hospitalized, but no source of infection was identified. Anakinra was withheld for 1 week because of infection concerns, during which time a severe flare of arthritis took place. Anakinra was restarted, with a rapid improvement in clinical and biochemical parameters, and the patient was discharged from the hospital after 2 weeks. She continues to be asymptomatic after 14 months of treatment with anakinra at 100 mg/day.
In February 2003, a 40-year-old white man developed acute synovitis of the left elbow. Over the next 2 weeks, this evolved into polyarthritis of the hands and feet. When seen in the emergency department, he reported a sore throat and indicated that, 3 months previously, he had experienced an erythematous rash on the trunk and limbs that lasted for 10 days. The WBC count was 9,900/mm3 (75% neutrophils), the Westergren ESR was 30 mm/hour, and results of a throat culture were negative.
He presented to a rheumatologist (SAL) in mid-March 2003 with joint pain and swelling, myalgias, and fatigue. The examination revealed generalized small and large joint synovitis, pitting edema of the left calf and foot, mild cervical adenopathy, and proximal muscle wasting. There was no rash. The WBC count was 14,400/mm3 (95% neutrophils), the CRP level was 232 mg/liter, and the serum ferritin level was >1,659 ng/ml. ANA, ENA, anti-DNA antibodies, and RF were negative. Creatine kinase was normal, and ALT and AP levels were elevated at 223 and 320 units/liter, respectively (normal <50 and 30–130, respectively). Findings of ultrasound of the patient's left leg were normal. The synovial effusion fluid had a WBC count of 31,200/mm3. Findings of chest radiography and echocardiography were normal. Viral serology, ASO titer, and cultures were negative. The patient was treated with standard cyclooxygenase inhibitors.
Within 3 weeks, an erythematous rash appeared on the patient's limbs and trunk, with daily spiking fevers to 104°F. A clinical diagnosis of AOSD was made and prednisone was started at 20 mg 3 times per day. After 4 weeks, the rash and fever resolved but the polyarthritis and myalgias continued. The WBC count was 16,700/mm3 (84% neutrophils) and the CRP level was 38 mg/liter. Due to the ongoing arthritis, etanercept (25 mg subcutaneously twice weekly) was added to the prednisone regimen, which had been tapered to 10 mg twice daily. Despite treatment with this combination for 6 weeks, the polyarthritis was still active and the CRP level remained elevated at 104 mg/liter. Results of liver function tests had normalized and MTX was added in increasing dosages, to 25 mg subcutaneously per week by 3 months. However, the combination of prednisone, MTX, and etanercept did not improve the patient's polyarthritis. Furthermore, the WBC count remained elevated at 13,900/mm3 (90% neutrophils) and the CRP level at 14 mg/liter. When the prednisone dosage was reduced from 20 mg/day to 15 mg/day, the arthritis worsened and the CRP level rose to 65 mg/liter.
In December, after a further 2 months of therapy, etanercept was discontinued but subcutaneous MTX at 25 mg/week and prednisone 20 mg/day were continued. Anakinra was started at 100 mg/day. Prior to the initiation of anakinra, the WBC count was 13,400/mm3 (78% neutrophils), with a CRP level of 29 mg/liter and a serum ferritin level of 234 ng/ml. Within 24 hours of the first subcutaneous injection of anakinra, the patient reported a significant improvement in joint stiffness, pain, and swelling. His grip formation and grip strength improved, and joint pain and stiffness resolved within 1 week. Two weeks after the initiation of daily anakinra, the WBC count was 9,500/mm3 (70% neutrophils), the CRP level was 0.9 mg/liter, and the serum ferritin level was 149 ng/ml. Six months later, the patient is healthy on a regimen of anakinra and MTX.
In September 2002, a 31-year-old Canadian man presented with a 5-day history of fever, pain in his wrist and ankle, sore throat, and a nonpruritic rash. He described a spiking fever to 104°F with a 10–20-pound weight loss in the preceding 10 days. Upon examination, it was determined that he had synovitis of his wrists. He was admitted to the hospital, where he continued to have spiking fevers to 104°F, during which time a pale maculopapular rash was present over the trunk and upper extremities. At that time, the hemoglobin level was 11.8 gm/dl with a WBC count of 6,500/mm3 (61% neutrophils with 16% bands) and a Wintrobe ESR of 49 mm/hour (normal 0–10). Liver function tests revealed an AP level of 308 units/liter and an ALT level of 174 units/liter. The serum ferritin level was 1,474 ng/ml. Findings on investigations for sepsis were negative except for Streptococcus pyogenes group C found on throat culture. Liver biopsy showed nonspecific inflammation. The patient was administered a course of oral penicillin and discharged without a definite diagnosis.
For the next 2 months, the patient continued to feel unwell, with fever, rash, and arthritis, but these symptoms gradually subsided. After he had been asymptomatic for 2 months, the rash, sore throat, and fever returned. At that time, he was examined by a rheumatologist (JEH) and AOSD was diagnosed. The patient was admitted to the hospital, at which time laboratory studies revealed the following: hemoglobin 12.9 gm/dl, WBC count 8,900/mm3 (80% neutrophils), CRP 79 mg/liter, ESR 45 mm/hour, ALT 275 units/liter, AP 374 units/liter, and serum ferritin 2,424 ng/ml. The patient was treated with prednisone 20 mg/day. Although fever and systemic symptoms subsided, the rash and synovitis persisted. Two weeks prior to anakinra administration, while the patient was being treated with prednisone 10 mg/day, the hemoglobin level was 14.2 gm/dl, the WBC count was 7,400/mm3, the CRP level was 146 mg/liter, the ESR was 39 mm/hour, the ALT level was 114 units/liter, and the serum ferritin level was 1,221 ng/ml.
In May 2003, the patient was started on anakinra 100 mg/day subcutaneously with prednisone 7.5 mg/day. Forty-eight hours after receiving the first injection of anakinra, the patient's WBC count decreased to 3,400/mm3 (51% neutrophils) and the CRP level fell to 32 mg/liter, but the serum ferritin level remained elevated at >1,650 ng/ml. However, after 2 weeks of anakinra therapy, the fever, rash, and arthritis resolved and the hemoglobin level rose to 14.3 gm/dl, the WBC count rose to 3,900/mm3 (49% neutrophils), the ESR increased to 12 mm/hour, the ALT level was 83 units/liter, while the serum ferritin level decreased to 728 ng/ml. After 4 weeks of anakinra treatment, prednisone was discontinued. After 8 weeks of anakinra treatment, the hemoglobin level was 14.6 mg/dl, the WBC count was 4,400/mm3 (46% neutrophils), the ESR was 2 mm/hour, the ALT level was 25 units/liter, and the serum ferritin level was 95 ng/ml. After 1 year of daily anakinra treatment, the patient remains well.
Since the initial descriptions of AOSD by Bywaters (21), AOSD patients have been initially treated with high dosages of acetylsalicylic acid or nonsteroidal antiinflammatory drugs. It is reported that ∼20% of patients respond to this therapy, whereas 80% require more aggressive treatment with corticosteroids with or without disease-modifying antirheumatic drugs (DMARDs) in order to control both arthritis and systemic features (22). In a retrospective review of 26 patients with refractory AOSD in France (3), 23 responded to MTX (up to 17.5 mg/week), including 18 with complete and 5 with partial responses. Three failed to respond. Lack of response to the combination of MTX and prednisone in some patients is well recognized. Characteristically, patients with refractory AOSD continue to experience spiking fevers despite treatment with high-dose prednisone, but the fevers improve with salicylates (1).
Etanercept has been administered to patients with refractory AOSD (13, 23, 24). Although some patients have exhibited a response, others have discontinued treatment and some have experienced disease flares while receiving etanercept, as observed in cases 2 and 3 reported herein. In general, responses were not observed until therapy had been administered for several weeks. By comparison, infliximab treatment generally results in more rapid improvement of clinical and biochemical markers in those who respond (10–14). Although some patients respond to infliximab relatively rapidly, in others disease activity has improved only after several weeks; in some patients infusions have been required every 4 weeks, and some have discontinued treatment (15). There are also reports of complications in these patients (13), including a report of fulminant hepatitis 2 weeks after the second infliximab infusion (25).
In the present study, we observed rapid and impressive reductions (within hours to a few days after the first anakinra injection) in disease activity, hematologic parameters, and biochemical markers in 4 patients. In the index case, stopping anakinra on 2 different occasions resulted in a rapid return of fever, leukocytosis with a prominent neutrophilia, and rising levels of acute-phase reactants, which fell dramatically within a few days upon restarting anakinra. In patient 2, stopping and restarting anakinra resulted in a similar flare and resolution cycle of the disease. Rapid responses following anakinra treatment of refractory AOSD have also been observed by others (26, 27).
As shown in the index case (Figure 2), but also in each of the other cases, a marked defervesence took place within a few hours after the first anakinra injection. A single subcutaneous injection of 100 mg of anakinra in patients with rheumatoid arthritis (RA) results in mean peak IL-1Ra serum levels of 1.2–1.35 μg/ml after 5 and 6 hours followed by a rapid decrease to <0.035 μg/ml after 24 hours. In contrast, following an intravenous infusion of infliximab (3–5 mg/kg), peak serum levels of the antibody reach 100 μg/ml, are maintained at similar levels for several days, and slowly decrease over 2–4 weeks. In fact, 6 weeks after a single infliximab infusion of 3–5 mg/kg, inhibition of TNFα biologic activity is still present (28).
Given the rapid and near-complete resolution with anakinra therapy and given the relatively low and short-lived concentrations of anakinra compared with infliximab, any benefit of TNFα neutralization in AOSD is likely due to a reduction in IL-1. Levels of IL-1 in AOSD before and after infliximab treatment have not been reported. However, experimental studies reveal that TNFα induces IL-1 (29, 30). Patients with a purely TNF-mediated disease called TNF receptor–associated periodic syndrome (TRAPS) experience fever, systemic inflammation, and acute-phase responses due to a mutation in TNF receptors. Although these patients exhibit decreased disease severity when treated with exogenous soluble TNF receptors, the improvement is short-lived and some patients do not respond to TNF blockers. However, they respond to anakinra (31). Therefore, in TRAPS, IL-1 is downstream from TNF, and IL-1 mediates the manifestations of the disease. Results of the present study support a similar concept in AOSD.
The rapid defervesence following a single injection of anakinra is illustrative of IL-1 mediating the fever of AOSD. Although humans injected with TNFα for the treatment of cancer develop rigors and fevers, these occur at doses starting at 100 ng/kg and reaching 1,500 ng/kg, without a clear dose response (32–35). By comparison, injection of IL-1α or IL-1β into humans induces rapid-onset fevers at doses as low as 3–10 ng/kg (36–42). Following doses of 3 ng/kg or 10 ng/kg, the maximal blood level of IL-1 in a 70-kg subject would be 40 pg/ml (2 pM) and 140 pg/ml (8 pM), respectively. Steady-state circulating levels of IL-1 differ in health and disease in the low pg/ml level due to variations in commercially available kits (43, 44). Moreover, circulating levels represent a “spillover” from the levels at the end organ. Therefore, attribution of a causative role for IL-1 in disease should not be based on circulating levels but rather on clinical responses to agents that reduce IL-1 activities.
Supporting the role of IL-1 in AOSD is the rapid reduction in leukocytosis (with prominent neutrophilia) following anakinra treatment in each of the 4 cases described herein. In the index patient, each time anakinra was withheld, the WBC count rose to >20,000/mm3 within a few days, but upon restarting anakinra, the WBC count rapidly decreased to a normal range (Figure 3). Unlike TNFα, IL-1 is a true hematopoietic cytokine (45), whereas TNFα suppresses the bone marrow precursors. Humans administered intravenous injections of either IL-1α or IL-1β show a rapid increase in levels of circulating neutrophils (for review, see ref. 46). IL-1 has been used therapeutically in patients undergoing bone marrow transplantation (37, 39–42). In humans injected with only 3 ng/kg of IL-1β, there is a brisk neutrophilia within 1 hour (36). In contrast, in humans infused with TNFα at doses ranging from 100 ng/kg to 15,000 ng/kg, a profound neutropenia is observed (33–35, 47, 48).
Like IL-1, IL-6 is a hematopoietic growth factor, and injections of IL-6 increase peripheral neutrophil counts in bone marrow transplant patients (49, 50). The fall in IL-6 levels following infliximab administration is well established in patients with RA (51) but also occurs with anakinra. In humans with septic shock, there is a dose-dependent fall in serum IL-6 levels within 24 hours of anakinra treatment, whereas in placebo-treated patients, IL-6 levels continue to rise (52).
The fall in CRP and serum ferritin levels with anakinra can be attributed to a fall in IL-6 activity because IL-6 is an inducer of hepatic acute-phase protein synthesis. For example, there is no acute-phase response in mice deficient in IL-6 (53, 54). However, IL-6 is also not present in mice deficient in IL-1β, and these mice similarly exhibit no acute-phase response to inflammatory stimuli (55, 56). IL-1–induced fever in mice requires the presence of IL-6 (57, 58) and cyclooxygenase 2 (59).
In humans with familial cold autoinflammatory syndrome (60), levels of circulating IL-1β are not detectable, whereas circulating IL-6 rises within a few hours after cold exposure (61). In these patients, cold-induced fever, leukocytosis, and acute-phase proteins are prevented by pretreatment with 100 mg of anakinra (61). In familial cold autoinflammatory syndrome, Muckle-Wells syndrome, and neonatal-onset multisystem inflammatory disease, a single amino acid mutation in the NALP3 gene results in increased secretion of IL-1β from blood monocytes due to dysfunctional control of caspase 1, the intracellular enzyme that controls the processing and release of IL-1β (62). In each of these systemic inflammatory syndromes, anakinra, at the same dosing schedule used to treat the patients with refractory AOSD, rapidly reverses the fever, serositis, arthralgias, and fatigue as well as the leukocytosis (neutrophilia) and CRP levels (61, 63–66). Indeed, elevation of serum IL-6 in our index patient decreased upon anakinra therapy (Figure 3). Therefore, it is not unexpected that anti–IL-6 receptor antibodies were effective in treating a patient with refractory AOSD (67). Figure 4 depicts the likely cytokine cascade in AOSD, based on preclinical and clinical data.
Of particular interest were the high levels of IL-18 in the circulation of the index patient. Although these decreased somewhat upon treatment cycles with anakinra, they were markedly elevated compared with findings in other inflammatory diseases. Others have reported elevated serum IL-18 levels in AOSD (9, 68). Interestingly, there was no correlation between IL-18 bioactivity and levels measured by enzyme-linked immunosorbent assay, the bioactivity being 10-fold lower (68). In one study, serum IL-18 levels in AOSD remained elevated even when the disease was in remission. In our index patient, high levels to IL-18 heralded the progression to macrophage activation syndrome. IL-18 is thought to be pathologic in macrophage activation syndrome (20). In 16 patients with AOSD, 8 polymorphisms were identified in the IL-18 gene and 3 haplotypes were defined. The frequency of homozygosity for 1 of the haplotypes, which contained all 8 IL-18 polymorphisms, was significantly higher in the AOSD patients than in the healthy controls (P = 0.0006, odds ratio [OR] 7.81, 95% confidence interval 2.48–24.65), whereas in RA patients, the OR was 4.0 (P = 0.015) compared with healthy controls (69).
Similar to RA patients, patients with refractory AOSD require long-term anticytokine treatment. However, the safety of infliximab, particularly regarding risk of opportunistic infections, is a major concern. Infection with Mycobacterium tuberculosis is severalfold greater in RA patients treated with a DMARD plus infliximab compared with those not receiving infliximab (70–72). By comparison, in a large, placebo-controlled trial of DMARD plus anakinra treatment in RA, there were no opportunistic infections (73). Moreover, based on findings in our cohort of patients, anakinra treatment allows for reduction or cessation of corticosteroid therapy, providing an additional margin of safety for long-term treatment options in AOSD.
We thank David Dryland, MD, and Alan Solinger, MD, for helpful discussions. We also thank Leonid L. Reznikov and Jacqueline K. Larsen for their help with cytokine assays, as well as Ms Pat Boulton and Dr. Cheryl Hanson for their assistance in these studies.