All patients enrolled in this study were being treated at the Yokohama City University Hospital, the Yokohama City University Medical Center Hospital, or the National Hospital Organization Yokohama Medical Center (Table 1). Seven patients with secondary HPS met the diagnostic guideline for hemophagocytic lymphohistiocytosis 2225, except as regards hypertriglyceridemia and hypofibrinogenemia, neither of which is generally applicable to secondary HPS in adults. In these seven patients, the underlying diseases were systemic lupus erythematosus (SLE) in two; hematological malignancy, including non-Hodgkin's lymphoma, multiple myeloma, and acute myeloid leukemia, in three; and ASD and viral infection in the others. The patients having more than two lineages of cytopenia, liver dysfunction, fever above 39°C, and hyperferritinemia were categorized as having active disease. Remission of the diseases was defined as disappearance of these findings after therapy. Ten patients with ASD met the criteria of Cush 26 and Yamaguchi 27 and their colleagues. An ASD patient who also met the diagnostic guidelines for hemophagocytic lymphohistiocytosis was classified in the HPS group in this study. Patients with active ASD were those presenting with polyarthritis, typical skin rashes, and fever above 39°C, in addition to hyperferritinemia. When the symptoms and signs had subsided, the patients were considered to be in remission.

We also studied 73 patients with other rheumatic diseases, including 30 with rheumatoid arthritis (RA), 18 with SLE, 9 with dermatomyositis/polymyositis (DM/PM), and 16 with Behçet's disease (BD). The diagnosis of individual diseases was based on the following criteria: for RA, the 1987 American College of Rheumatology (formerly, the American Rheumatism Association) criteria 28; for SLE, the 1997 updating of the American College of Rheumatology revised criteria for the classification of systemic lupus erythematosus 29; for polymyositis and dermatomyositis, the diagnostic criteria described by Bohan and Peter 3031; and for Behçet's disease, the International Study Group criteria for diagnosis of Behçet's disease 32. The disease activity was evaluated at the time of blood sampling. All of the RA patients were considered to have active disease, because their disease activity scores (DAS) based on 28 joints and C-reactive protein (CRP) (DAS28-CRP) were more than 3.2 33. The mean CRP level at the time of blood sampling was 2.4 ± 2.7 mg/dl. On the basis of the Systemic Lupus Disease Activity Index 34, 12 of the 18 SLE patients had a score above 9 and were regarded as having active disease, while the other 6 were in remission. Two other SLE patients who met the diagnostic guidelines for hemophagocytic lymphohistiocytosis were included in the HPS group 2225. All of the DM/PM patients had active diseases, inasmuch as their creatine kinase concentrations were more than twice the normal upper limit and they had muscle weakness and/or active interstitial pneumonia. Six of the 16 BD patients presented active symptoms of uveitis, erythema nodosum, genital ulcers, deep vein thrombus, central nervous system involvement, arterial occlusion, or gastrointestinal lesions in addition to positive CRP, indicating active disease; the other 10 were regarded as having inactive disease.

Twenty patients with liver diseases were enrolled in this study (Table 1). Of the five with acute hepatitis, three had hepatitis B, one had drug-induced hepatitis, and one had Epstein–Barr viral hepatitis. The seven patients with chronic hepatitis included one with hepatitis B and six with hepatitis C. Serum alanine aminotransferase levels were measured as an indicator of liver injury. The means ± standard deviations (IU/l) found for these 20 patients were as follows: acute hepatitis, 770.0 ± 568.6; chronic hepatitis, 61.9 ± 28.9; liver cirrhosis, 59.0 ± 24.0; hepatocellular carcinoma, 27.5 ± 14.8; primary biliary cirrhosis, 98.5 ± 14.1; autoimmune hepatitis, 259; and alcoholic hepatitis, 56.

Ten patients who had received frequent blood transfusions were also included. The underlying hematological diseases were myelodysplastic syndrome in six patients and aplastic anemia in four. Healthy volunteers served as normal controls. All the studies were performed after obtaining written informed consent, which was approved by the local Institutional Review Board.

Serum ferritin and HO-1 levels were measured by an EIA detection system (Tosoh, Tokyo, Japan), and a human HO-1 ELISA kit (Stressgen, Victoria, Canada), respectively. Concentrations of serum tumor necrosis factor (TNF)-α were determined by specific ELISA systems using pairs of capture and biotin-conjugated detecting antibodies, which were purchased from R&D (Minneapolis, MN, USA). Serum IL-18 level was determined using a human IL-18 ELISA kit in accordance with the manufacturer's protocol (MBL, Nagoya, Japan).

Peripheral blood mononuclear cells (PBMCs) were isolated by centrifugation over Ficoll-Hypaque (ICN, Aurora, OH, USA). 10⁶cells/ml were cultured with 100 μM hemin (Sigma-Aldrich, Saint Louis, MO, USA) in Hepes modified RPMI 1640 (Sigma-Aldrich) containing 10% fetal calf serum (Equitech-Bio, Kerrville, TX, USA), 2 mM L-glutamine (Sigma-Aldrich), 100 U/ml penicillin, plus 100 μg/ml streptomycin (Sigma-Aldrich) in a 5% CO₂ in an air incubator at 37°C for 24 hours.

Total RNA was isolated from cells by using TRIzol reagent (Invitrogen, Carlsbad, CA, USA). Reverse transcription was performed using a SuperScript™ reverse transcriptase (Invitrogen). Panels of primers of human HO-1 and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNA were purchased from PE Applied Biosystems (Foster City, CA, USA). Real-time PCR was performed using a TaqMan Universal Master Mix (PE Applied Biosystems), and the data were analyzed by the ABI prism 7700 sequence detection system (PE Applied Biosystems). Briefly, 1/50 amounts of of cDNA derived from 1 μg of total RNA, 200 nmol/l of probe, and 800 nmol/l of primers were incubated in 25 μl at 50°C for 2min and 95°C for 10min, followed by 40 cycles of 95°C for 15s and 60°C for 1 min. The amounts of cDNA obtained from transcriptions of mRNA were semiquantified in comparison with those of serially diluted standard cDNA, which was prepared using a conventional PCR technique. The expression level of HO-1 mRNA in a sample was expressed as arbitrary units, which were determined by the formula 1AU = (HO-1 mRNA/GAPDH mRNA) × 100.

Cells expressing HO-1 were determined with anti-HO-1 monoclonal antibody (Stressgen) using a Dako LSAB2 kit (Dako, Glostrup, Denmark).

The Mann–Whitney U test, the Wilcoxon signed rank test, and multiple regression analyses were used to test for differences. P values less than 0.05 were considered significant. Values are reported as means ± standard deviations.

Serum HO-1 levels in patients with inflammatory rheumatic diseases were monitored by ELISA. In the healthy controls, only very low levels of serum HO-1 were detectable (2.6 ± 1.3 ng/ml) (Fig. 1). Age and sex did not influence HO-1 levels. In contrast, HO-1 levels were significantly elevated in patients with active ASD and HPS (Table 1; Fig. 1). HO-1 protein levels exceeded 10 ng/ml in all but one patient with ASD, who was classified as having active disease in the study because of high fever with elevated levels of CRP and ferritin during maintenance therapy with a low dose of prednisolone (PSL). However, the clinical manifestations were less serious and serum ferritin was lower (1201 ng/ml) than in any other patient with active ASD in this study. Although subjects with active DM/PM also had significantly increased serum HO-1 levels (Table 1; P = 0.001), these were still significantly lower than in subjects with active HPS or ADS (P = 0.0007 and P = 0.003, respectively). Serum HO-1 levels were not increased in other rheumatic diseases including RA, SLE, and BD, regardless of disease activity (except for two patients with SLE complicated by HPS) (Table 1). These findings suggest that increased serum HO-1 levels are characteristic of active ASD and HPS.

Serum HO-1 levels were monitored before and after remission-inducing therapy that included corticosteroids with or without cyclosporin A in three patients with HPS and five with ASD. Serum HO-1 levels were significantly reduced after successful therapy (Fig. 2a) (P = 0.0078).

Serum HO-1 and ferritin were serially monitored in one patient with ASD and one with HPS during the course of disease (Fig. 2b,c). A 34-year-old man admitted with fever, polyarthralgia, sore throat, and salmon-pink rashes was diagnosed with ASD (Fig. 2b). When this patient was admitted, his serum concentrations of both HO-1 and ferritin were extremely elevated (182 ng/ml and 6,855 ng/ml, respectively). Treatment with methylprednisolone (mPSL) pulse therapy (1,000 mg/day for 3 days) followed by oral PSL (60 mg/day) and cyclosporin A (200 mg/day) led to clinical remission. Associated with this response to therapy, serum HO-1 levels gradually decreased to the normal range over 2 months, as did levels of ferritin and CRP. PSL was tapered to 30 mg/day without relapse.

In a 45-year-old woman with SLE admitted with high fever and cytopenia (Fig. 2c), bone marrow aspiration revealed hemophagocytosis, and her serum ferritin level was 4,588 ng/ml, resulting in a diagnosis of HPS complicated with SLE (the Systemic Lupus Disease Activity Index score was 9). On admission, increased serum HO-1 (74.8 ng/ml) was noted. mPSL pulse therapy (1,000 mg/day for 3 days) followed by oral PSL (60 mg/day) and intravenous gamma globulin (17.5 g/day for 5 days) temporarily reduced her fever and CRP levels. Despite these treatments, serum ferritin and HO-1 peaked at 25,070 ng/ml and 214 ng/ml, respectively. A second course of mPSL pulse therapy also failed, but the patient's condition gradually improved after initiation of cyclosporin A (200 mg/day). Serum levels of CRP, ferritin, and HO-1 reached normal levels by two months after admission. PSL was tapered to 30 mg/day without exacerbation. These findings suggest that the serum HO-1 level is closely correlated with disease activity during the clinical course in patients with HPS and ASD.

We next examined the relation between the serum HO-1 level and other laboratory parameters in the patients with HPS and ASD. Because serum ferritin was widely accepted as a monitoring marker for the diseases, the data included in the analysis were those found when the ferritin level was highest in individual patients during the whole study. The results indicate that serum HO-1 correlates closely with serum ferritin (P = 0.0048, Fig. 3a) but not CRP or lactate dehydrogenase (LDH) levels (Fig. 3b,c), a finding consistent with an association between HO-1 and hyperferritinemia in patients with HPS and ASD. We also measured serum levels of IL-18 and TNF-α, both of which have been shown to be elevated in patients with HPS and ASD 3536. However, we did not find any correlation between serum level of HO-1 and those of cytokines (data not shown).

Besides being found in patients with HPS and ASD, hyperferritinemia is also found in patients with liver diseases and in recipients of frequent blood transfusions. Because ferritin synthesis is stimulated by Fe²⁺, which is generated by HO-1-mediated heme degradation, hyperferritinemia might be caused by high HO-1 activity, irrespective of the underlying diseases. To examine this possibility, the relation between serum HO-1 and ferritin was evaluated in all patient groups. A total of 37 patients had serum ferritin levels >500 ng/ml, which is the cutoff level in the revised diagnostic criteria for HLH 2225. Serum HO-1 levels exceeded 10 ng/ml in 7 of 7 HPS patients and in 9 of 10 ASD patients but in only 2 of 20 patients with other diseases (one with dermatomyositis and the other with Epstein–Barr hepatitis) (Fig. 4). Of all the subjects studied, only one person, with dermatomyositis, had serum HO-1 >10 ng/ml but serum ferritin <500 ng/ml. Thus, simultaneous elevation of serum ferritin and HO-1 was much more common in patients with ASD and HPS than any other disease studied.

Yachie and colleagues reported that PBMCs from children with acute inflammatory illness express elevated HO-1 mRNA levels 37. In our study, HO-1 mRNA expression in PBMCs was semiquantified using real-time PCR. We found that PBMCs from 3 of 5 patients with active HPS and 3 of 10 with active ASD had HO-1 mRNA expression exceeding the mean + 2 standard deviations of healthy controls, whereas no such elevations were found in PBMCs from patients with other rheumatic diseases, irrespective of disease activity (Fig. 5a). The six patients with increased mRNA expression universally manifested elevated serum HO-1 protein levels. Moreover, HO-1 mRNA expression fell when remission was induced in two patients with HPS and one with ASD (Fig. 5b). Changes in HO-1 mRNA expression mirrored changes in serum HO-1 protein levels in a 45-year-old woman (Fig. 5c). This patient, who had had ASD for 4 years and maintained remission with PSL (20 mg/day), was admitted to our hospital because of high fever and cytopenia. Bone marrow aspiration revealed hemophagocytosis, indicating that the patient's ASD was complicated with HPS. Besides increased serum ferritin (8,690 ng/ml) and HO-1 (40.4 ng/ml), HO-1 mRNA expression in PBMCs was much higher than that of healthy controls. Immunocytochemistry showed that HO-1-expressing cells were found in hemin-treated, but not untreated, PBMCs from normal donors (Fig. 6a,b), whereas HO-1 proteins were stained in freshly isolated PBMCs, mainly monocytes, from the patient (Fig. 6c). After clinical remission was achieved by mPSL pulse therapy and subsequent oral PSL, HO-1 mRNA in PBMCs was reduced in parallel with serum HO-1 and ferritin levels (Fig. 5c). These data indicate that circulating PBMCs may contribute to increased serum HO-1 protein levels in some subjects. However, since HO-1 mRNA expression was normal in PBMCs from 9 of 15 patients with active HPS and ASD, despite elevated serum HO-1, it is clear that PBMCs are not a critical source of circulating HO-1.