Early diagnostic and prognostic significance of a specific Th1/Th2 cytokine pattern in children with haemophagocytic syndrome


Yongmin Tang, Division of Haematology-Oncology, Children’s Hospital, Zhejiang University School of Medicine, #57 Zhuganxiang Road, Yan-an Street, Hangzhou 310003, China. E-mail: y_m_tang@zju.edu.cn


The haemophagocytic syndrome (HPS) is a rare but frequently fatal disorder of immune regulation caused by hypercytokinemia. Using cytometric bead array technique, the serum T-helper cell type 1 (Th1) and 2 (Th2) cytokines including interferon-γ (IFN-γ), tumour necrosis factor (TNF), interleukin (IL)-10, IL-6, IL-4 and IL-2 were determined in 24 children with de novo HPS and 87 children as control. The median levels of serum IFN-γ, IL-10 and IL-6 in the acute phase of HPS were 901·7, 879·0 and 63·8 pg/ml, respectively, significantly higher than those after remission, and in the healthy volunteers and patients with viral infection. IL-4 was slightly elevated while IL-2 and TNF were within normal range in acute phase. Patients with bacterial sepsis showed an extremely high level of IL-6 and moderate level of IL-10, whereas IFN-γ was only slightly elevated. Five patients were diagnosed with HPS according to the Th1/Th2 cytokine pattern 3–13 d earlier than they fulfilled the relevant diagnostic criteria. IL-10 level >2000 pg/ml was an unfavorable prognostic factor for HPS treatment response (P = 0·033) and outcome (P = 0·009). We conclude that the significant increase of IFN-γ and IL-10 and a slightly increased level of IL-6 is an early, specific and prognostic cytokine pattern for childhood HPS.

The haemophagocytic syndrome (HPS), also named haemophagocytic lymphohistiocytosis (HLH), is a rare but frequently fatal disorder of immune regulation, characterized by a high-grade fever, hepatosplenomegaly, pancytopenia, high ferritin level, and increased proliferation and activation of benign macrophages with haemophagocytosis throughout the reticuloendothelial system. Numerous well-differentiated macrophages phagocytosing haematopoietic cells can usually be found in the bone marrow, liver, spleen or lymph nodes (Henter et al, 1991a, 1997, 2007). The clinical presentation of HPS is caused by a hyperinflammatory syndrome resulting from hypercytokinemia of various pro-inflammatory mediators, such as interferon-γ (IFN-γ), tumour necrosis factor-α (TNF-α), interleukin (IL)-6, IL-8, IL-12, IL-18, macrophage–colony-stimulating factor (M-CSF), and a number of haematopoietic growth factors released by stimulated lymphocytes and macrophages (Henter et al, 1991b; Akashi et al, 1994; Osugi et al, 1997; Takada et al, 2001; Mazodier et al, 2005).

The diagnosis of HPS is frequently delayed or made at autopsy because the clinical symptoms are not specific in the early stages. Diagnostic work-up should include the search for one or more infectious agents, especially viruses, and the measurement of disease markers, such as fibrinogen, ferritin, triglycerides, soluble CD25, inflammatory cytokines, natural killer (NK) cell function and CD8+ T cells (Henter et al, 2007). Many studies have shown that the early diagnosis of HPS is very important to prevent fatality (Henter et al, 1997, 2004). However, the current diagnostic criteria proposed by International Society of Histiocyte in 2004 (HLH-2004) are not always conclusive (Henter et al, 2007), especially in the early stage of HPS because the relevant clinical parameters are usually not present at that stage of the disease (Henter et al, 1997; Palazzi et al, 2003). For instance, phagocytosed macrophages are not observed even in bone marrow aspiration or biopsy. A diagnostic method practical for the early diagnosis of HPS is needed. A cohort of 24 children with HPS attending our hospital has been studied over a 2-year period by determining the T-helper cell type 1/2 (Th1/Th2) cytokine profile in the sera of the patients and clinically practical findings for the early diagnosis of HPS have been identified.

Patients and methods

Patients and controls

Serum Th1 and Th2 cytokines, including IFN-γ, TNF, IL-10, IL-6, IL-4 and IL-2, were measured in 24 children under 10 years of age with newly diagnosed HPS, while those from 33 healthy volunteers, 22 patients with viral infection and 32 patients with sepsis caused by Gram-negative (G) bacterial infection were used as controls. All the patients were from our hospital between 22 December 2005 and 30 November 2007. Patients who had been treated by steroids before referral to our hospital were excluded. The healthy control samples were from volunteers (17 boys and 16 girls with a median age of 5·5 years, range: 6 months to 15·4 years). The viral infection control samples (14 boys and eight girls with a median age of 3·2 years, range: 8 months to 10·9 years) were from patients diagnosed with infectious mononucleosis (IM) based on positive Epstein–Barr virus (EBV) antibody detection. The bacterial infection control group (sepsis control) contained 32 patients with sepsis caused by G bacterial infection and confirmed by blood culture. This group (23 boys and nine girls with a median age of 5·0 years, range: 1·2 years to 15·1 years) included 24 patients with acute lymphoblastic leukaemia (ALL), two patients with acute myeloid leukaemia (AML), four with non-Hodgkin lymphoma (NHL), one with neuroblastoma, and one without any malignant disease. This study was approved by the Ethic Review Board of the Children’s Hospital, Zhejiang University School of Medicine and informed consent was obtained from their parents or guardians.

Diagnostic criteria

The diagnosis of HPS was based on the revised criteria (HLH-2004) of the International Histiocyte Society for the diagnosis of HPS (Henter et al, 2007). The diagnostic criteria included: (i) fever; (ii) splenomegaly; (iii) cytopenia ≥2 cell lineages; (iv) hypertriglyceridemia and/or hypofibrinogenemia; (v) serum ferritin ≥500 μg/l; (vi) haemophagocytosis in bone marrow, cerebrospinal fluid or lymph nodes; (vii) decreased or absent NK-cell activity; (viii) soluble CD25 (i.e. soluble IL-2 receptor) ≥2400 U/ml (NK-cell activity and sCD25 determination was not performed in this cohort of patients).

Cytokine assay

Peripheral blood samples were collected at the time of acute phase, convalescent and follow-up periods. One millilitre of blood was transferred to a serum-separating tube and centrifuged at 1000 g, 20°C for 20 min after clotting. The sera were carefully harvested, and the aliquots were stored at 2–8°C until analysis (within 12 h). The concentrations of IFN-γ, TNF, IL-10, IL-6, IL-4 and IL-2 in the sera were quantitatively determined by the cytometric bead array (CBA) kit – BD™ CBA Human Th1/Th2 Cytokine Kit II (BD Biosciences, San Jose, CA, USA) as described in the literature (Chen et al, 1999). Briefly, the CBA technique was based on six bead populations with distinct fluorescence intensities that had been coated with capture antibodies specific for IFN-γ, TNF, IL-10, IL-6, IL-4 and IL-2 proteins. The fluorescent dye had a maximal emission wavelength of approximately 650 nm (FL-3), which was detectable by flow cytometry. The cytokine capture beads were mixed with the phycoerythrin-conjugated detection antibodies and then incubated with recombinant standards or test samples to form sandwich complexes. Following the acquisition of sample data on a FACScalibur™ flow cytometer (Becton Dickinson, San Jose, CA, USA), the sample results were generated in graphical and tabular format using the BD CBA Software (BD Biosciences, San Jose, CA, USA). Six standard curves (range from 0 to 5000 pg/ml) were obtained from one set of calibrators and six results were obtained on one test sample. The maximum and minimum limits of detection for all the six cytokines were 1·0 pg/ml and 5000·0 pg/ml, respectively.

Treatment methods

The treatment was based on the HLH-2004 treatment protocol described in the literature (Henter et al, 2007). Briefly, the initial therapy consisted of dexamethasone (DXM, 10 mg/m2 for 2 weeks followed by 5 mg/m2 for 2 weeks, 2·5 mg/m2 for 2 weeks, 1·25 mg/m2 for 1 week, and 1 week of tapering), etoposide (150 mg/m2 twice weekly for the first 2 weeks and then weekly), and cyclosporine A (6 mg/kg/d orally, aiming at blood level of 200 μg/l). After 8 weeks of initial treatment, patients underwent continuous therapy, starting from week 9, which consisted of DXM pulses with 10 mg/m2 for 3 d and etoposide infusions with 150 mg/m2 every alternative week until a total 12 months were reached. No patients underwent haematopoietic stem cell transplantation in this cohort.

Statistical analysis

Serum concentrations of individual cytokines were compared between HPS patients (the acute phase, after remission), healthy volunteers and infectious patients, using the Bonferroni-adjusted Mann–Whitney U-test followed by the Kruskal–Wallis test. Survival analysis was performed by the Kaplan–Meier method, and survival curves were compared using the Log-rank test. The associations between the serum cytokine levels and the treatment response were performed by Chi square test and the serum cytokine levels with each other among the HPS patients were tested by Spearman rank correlation. All statistical analysis was performed using the Statistical Package for the Social Sciences (spss) software, version 12·0 (SPSS Inc., Chicago, IL, USA). In our study, P < 0·05 was considered as significant except in the Bonferroni-adjusted Mann–Whitney U-test (P < 0·008).


Patients’ characteristics

Between 22 December 2005 and 30 November 2007, 24 newly diagnosed HPS patients were admitted to our hospital and underwent Th1/Th2 cytokine detection. There were 12 boys and 12 girls with a median age of 2·0 years (range: 7 months to 9·7 years). All the patients presented high-grade fever with no significant infection in the lung, skin, blood culture for bacteria or fungi and poor response to multiple potent antibiotics or anti-fungi or anti-viral agents. The primary features of these patients were prolonged high-grade fever (>38·5°C in 24/24), hepatosplenomegaly (22/24), pancytopenia (21/24), hyperferritinaemia (24/24), coagulopathy (22/24), haemophagocytosis in bone marrow (21/24), liver dysfunction (19/24) and hypertriglyceridaemia (20/24). NK cells were decreased in most patients tested (17/19) with a median level of 2·37% (0·24–10·54%), and was significant when compared with healthy controls (2·37% vs. 10·07%, P < 0·001). Neurological symptoms, such as cranial nerve palsies or seizures, were not very common (1/24). All the patients were followed up by phone calls or outpatient clinic until death.

Serum Th1 and Th2 cytokine levels in HPS

The serum Th1 and Th2 cytokines (IFN-γ, TNF, IL-10, IL-6, IL-4 and IL-2) were determined in all 24 patients at acute phase, including initial onset (Table I) and early phase of the disease recurrence, and remission. A total of 96 samples were colleted. The cytokines from healthy control and infectious control groups were also determined using the same method. The IFN-γ, TNF, IL-10, IL-6, IL-4 and IL-2 levels for healthy children were 4·6 (3·3–7·8), 2·3 (<1·0–3·1), 2·4 (1·3–9·9), 4·1 (1·2–8·5), 2·7 (1·1–4·0) and 5·8 (2·7–7·8) pg/ml respectively. However, in IM control, the IFN-γ, IL-10, IL-6, IL-4 levels were 18·9 (7·7–33·4), 9·3 (2·3–61·5), 8·7 (3·1–83·5) and 3·4 (1·2–8·4) pg/ml, which were all slightly higher than healthy control (P < 0·008 for all these groups), but TNF and IL-2 levels were comparable between the two (P > 0·05). In the sepsis controls, IFN-γ, TNF, IL-10, IL-6 and IL-4 were all increased to different extents (P < 0·001 in all these five groups), with median levels at 8·1 (3·2–99·7), 6·9 (<1·0 to >5000), 250·3 (6·8 to >5000), 3334·4 (154·9 to >5000) and 4·1 (<1·0–20·3) pg/ml. [Correction added after online publication 1 September 2008: in the preceding sentence the range, (0 to >5000) was corrected to (<1·0 to >5000).] IL-6 was markedly elevated in most patients with infection (>1000 pg/ml in 26 patients), and 11 samples were exceeded the maximum detection limits (>5000 pg/ml).

Table I.   Th1/Th2 cytokine peak levels in patients with HPS (at diagnosis).
CaseGenderAge (years)Duration (d)IFN-γ (pg/ml)TNF (pg/ml)IL-10 (pg/ml)IL-6 (pg/ml)IL-4 (pg/ml)IL-2 (pg/ml)Outcome

In the acute phase of HPS (including relapsed cases), IFN-γ level was significantly elevated with a median concentration of 901·7 pg/ml (range: 43·9 to >5000 pg/ml), which was significantly higher than those after remission, healthy, IM and sepsis controls (Fig 1A). Of the 24 patients, 17 had peak IFN-γ levels >500 pg/ml at diagnosis, and 12 were >1000 pg/ml. Only one relapsed sample had a relatively low IFN-γ level (43·9 pg/ml), and the levels of other samples at acute phase were all above 130·0 pg/ml. However, IFN-γ level after remission showed no difference when compared with healthy control.

Figure 1.

 Comparison of serum cytokine concentrations among HPS patients (acute phase, remission), healthy, IM and sepsis controls. (A) IFN-γ; (B) TNF; (C) IL-10; (D) IL-6; (E) IL-4; (F) IL-2. bsl00066bsl00066, P < 0·001; bsl00066, P < 0·008; NS, not significant.

Interleukin-10 level was also markedly elevated in the acute phase of HPS. The median peak value was 879·0 pg/ml (range: 26·5 to >5000 pg/ml), which was significantly higher than those after remission, the healthy, IM and sepsis controls as well (Fig 1C). The IL-10 at remission remained slightly higher (9·8 pg/ml) than that of the healthy control (P < 0·001).

The serum IL-6 concentration was moderately higher in the acute phase of HPS (median, 63·8 pg/ml, range, 9·4–698·8 pg/ml) than those after remission, healthy and IM controls (Fig 1D). However, the concentrations were much lower than those from patients with bacterial infection (63·8 pg/ml vs. 3334·4 pg/ml, P < 0·001). However, IL-6 remained at a slightly higher level for HPS patients in remission than healthy children (5·6 pg/ml vs. 4·1 pg/ml, P < 0·008). IL-4 was found slightly higher in the acute phase of HPS, IM and sepsis controls with a median of 3·4, 4·1 and 4·1 pg/ml, respectively (Fig 1E), but decreased when HPS achieved remission (4·1 pg/ml vs. 2·6 pg/ml, P < 0·0001).

The TNF (median: 2·3 and 1·7 pg/ml) and IL-2 (median: 5·0 and 5·3 pg/ml) concentrations were not elevated in HPS in this cohort both in acute phase and at remission as compared with healthy controls (Fig 1B and F). However, TNF level was elevated in some septic children (median, 6·9 pg/ml; range, <1·0 to >5000 pg/ml) when compared to HPS patients in both acute and remission phases, and healthy and IM controls (P < 0·008, P < 0·001, P < 0·001 and P < 0·001 respectively).

Early diagnostic significance of Th1/Th2 cytokine detection

The median duration to a definitive diagnosis of HPS was 12 d from the start of fever in this cohort. The diagnosis for HPS in most of our patients was not very difficult due to the presence of significant abnormalities to meet the criteria. Only five patients with fever of unknown origin did not meet the the minimal diagnostic criteria of HLH-2004 at the time of admission. However, all the five patients were found to have a hypercytokinemia pattern specific for HPS provided by cytokine determination. They eventually met the minimum requirement of five parameters from the HLH-2004 criteria at day 10 (case 8), day 3 (case 11), day 13 (case 12), day 6 (case 13) and day 5 (case 17), respectively, after admission.

The prognostic significance of Th1/Th2 cytokines

For the 18 patients who were survived, the median survival time was 9·5 months at the time of data censoring (range: 30 d to 23 months). [Correction added after online publication 1 September 2008: in the preceding sentence the number of patients, 20 was corrected to 18.] The 18-month overall survival (OS) rate of this cohort was 74·5 ± 9·0%. In order to evaluate the treatment response, patients were divided into two groups according to whether the fever could be reduced within 48 h after receiving DXM. Univariate analysis showed that remaining febrile in first 48 h after DXM administration was associated with an inferior outcome (P = 0·002). The association between serum cytokine levels and the outcome or treatment response was analyzed. Univariate analysis showed that a favorable outcome was achieved in patients with IL-10 level <2000 pg/ml, as compared to those with an initial IL-10 ≥2000 pg/ml, in 18-month OS rates (93·3 ± 6·4% vs. 41·7 ± 17·3%, P = 0·009), and IL-10 level <2000 pg/ml also predicted a favorable treatment response (P = 0·033). No significant relationships between IFN-γ, IL-2, IL-4, IL-6, TNF and the outcome or the treatment response of HPS patients were found.

Associations between the serum cytokine levels with each other in HPS patients

The associations between the serum cytokine levels with each other in HPS patients were analyzed. IFN-γ was significantly associated with IL-10 (r = 0·874, P < 0·001). Levels of both of these cytokines were high in acute phase and returned to relatively low levels upon remission. IFN-γ was associated with IL-6 (r = 0·777, P < 0·001) and IL-4 (r = 0·666, P < 0·001) as associations between IL-10 versus IL-6 (r = 0·734, P < 0·001) and IL-10 versus IL-4 (r = 0·538, P < 0·001) were also found.


Haemophagocytic syndrome is a rare, severe disease that is mostly seen in children. Diagnosis according to the criteria proposed by International Histiocyte Society is usually delayed in most cases. According to the criteria, at least five of the eight parameters should present in a patient before the diagnosis of HPS can be made. A number of patients may present one or more parameters of the minimum diagnostic criteria late in the course of the disease and this usually leads to an indefinite diagnosis with subsequent lack of appropriate treatment (Henter et al, 1997, 2004; Palazzi et al, 2003), and is a common cause of death. It has even been suggested that treatment should be started on the basis of a strong clinical suspicion of HPS before overwhelming disease activity occurs, which usually leads to an irreversible damage in some patients that do not fulfill all the diagnostic criteria (Henter et al, 1997). Therefore, an early diagnosis of HPS is extremely important.

In this cohort, the median duration before HPS diagnosis was 12 d. As our hospital is affiliated to Zhejiang University and is the major treatment unit in the province, many patients were referred here only after the disease could not be controlled by local or community hospital doctors. So, for patients diagnosed at our institution, the disease duration was already long or the situation was severe. As most patients were late in the course of disease with many symptoms and signs appeared, diagnosis was not very difficult. Only five patients presented with fewer symptoms/signs than required by the diagnostic criteria. However, all these five patients were diagnosed as HPS according to the unique Th1/Th2 cytokine pattern and treated accordingly.

Although hypercytokinemia in HPS has been noted, a clinically useful cytokine profile specific for the diagnosis of HPS has not been established. The method of cytokine determination in many hospitals uses the classic enzyme-linked immunosorbent assay (ELISA), which is usually not suitable for the quick diagnosis of a critical disease like HPS. Flow cytometry is a quick and an accurate method for the analysis of both the cells and the soluble components in the plasma or sera via the CBA technique (Chen et al, 1999). Compared with ELISA, it has a wider detection range and comparable analytic sensitivity, requires less sample volume and is less labour intensive. It has recently been shown that the results between ELISA and CBA techniques are comparable (correlation coefficient > 90%) (Cook et al, 2001; Tarnok et al, 2003), while CBA usually provides a quick result that is suitable for clinical diagnosis.

A lot of cytokines were thought to play important roles in HPS, and through the observation of the Th1/Th2 cytokine profile in our patients with HPS, we determined a useful approach for the early diagnosis of this disease. Our study showed that the patients with active HPS had elevated Th1/Th2 cytokines including IL-10, IFN-γ and IL-6, particularly IL-10 and IFN-γ, while TNF, IL-4 and IL-2 were at normal range or slightly elevated. When the patients achieved remission, these cytokines returned to normal or almost normal levels. This is partially consistent with previous reports (Osugi et al, 1997; Murohashi et al, 2006). Some reports also found that TNF-α was moderately elevated in most HPS cases, which is consistent with the factor that macrophages are over-activated in HPS (Henter et al, 1991b; Akashi et al, 1994), but the level was not as high as IFN-γ and IL-10. Most cases had reported TNF levels <20 IU/ml (320 pg/ml) (Henter et al, 1991b). In our cohort, most patients’ TNF levels were in normal range. The reason was unknown.

It has been shown that clinical manifestations, such as age, associated infection, cerebrospinal fluid pleocytosis and family history, are not correlated to the outcome (Arico et al, 1996), while IFN-γ, TNF-α and sCD25 are excellent prognostic factors (Ishii et al, 1991; Imashuku et al, 1995). IFN-γ and IL-18 are also sensitive indicators of disease activity (Ohga et al, 1993; Takada et al, 1999; Mazodier et al, 2005). Our findings showed that IL-10 was prognostic but a prognostic significance between IFN-γ and outcome was not observed. This may be attributed to the time of blood drawing. As shown both in this study and other reports, IFN-γ elevation and reduction was much quicker than that of IL-10 (Osugi et al, 1997). In this study, we did not observe the dynamic variation of these cytokines, and probably did not fully determine the real peak level of IFN-γ. So, we found that some patients with severe symptoms had a relatively low level of IFN-γ. The other possibility for the low IFN-γ level in a few patients might be caused by improper application of steroid agents in local hospitals or clinics, especially for those for whom the pretreatment detail could not be provided.

Interferon-γ is considered to play a pivotal role in the development of HPS (Jordan et al, 2004), whereas there have been fewer reports on IL-10. However, this study found that IL-10 was more prognostic for disease activity and outcome in HPS. As IL-10 can be produced by Th1, Th2 cells and macrophages, it should be considered as an ‘alarm hormone’ that reflects activation status of these cells and disease activity, while IFN-γ is directly involved in mediating the development of HPS and lethal complications.

Clinically, hypercytokinemia is also seen in various infections, especially in sepsis caused by G bacteria (Engervall et al, 1995; de Bont et al, 1999; Lehrnbecher et al, 2004; Stryjewski et al, 2005; Hodge et al, 2006). However, our study showed that HPS and sepsis caused by G bacteria had totally different cytokine patterns. In HPS children, IFN-γ was significantly elevated with moderately (rarely reaching 1000 pg/ml) elevated IL-6. But in sepsis caused by G bacteria, IL-6 was extremely high; in most cases (25/31) it was more than 1000 pg/ml, which has been confirmed by many clinical studies (Hack et al, 1989; Stryjewski et al, 2005). However, IFN-γ is just slightly increased in G bacterial sepsis, rarely more than 100 pg/ml. IL-10 can be highly elevated in both diseases. But in patients with Gram-positive (G+) bacterial infection, IL-6 was moderately elevated, and IL-10 and IFN-γ were normal or just slightly elevated (Engervall et al, 1995; de Bont et al, 1999; Lehrnbecher et al, 2004). And in patients with viral infection, for instance, EBV infection in this study, IFN-γ was predominantly but slightly elevated. In systemic inflammatory response syndrome many reports have shown that IL-6 and IL-10 were both elevated, and IL-6 was higher than IL-10 in acute phase, with TNF in a low level (Miyaoka et al, 2005; Ulloa & Tracey, 2005). Therefore, the pattern of IFN-γ, IL-6, IL-10 levels are important for the differential diagnosis between HPS and above diseases. HPS has a specific Th1/Th2 cytokine pattern, which is of great importance in the early and specific diagnosis of HPS.

Based on the findings in the present study and others, we can depict a schematic picture of HPS development (Fig 2): the pathogens elicit innate immune reactions, with the production of IL-12, or stimulate NK cells or CD8+ T cells to produce IFN-γ, which in turn act on macrophages to secrete IL-12. IL-12 binds to the receptors on antigen-stimulating CD4+ T cells and activates the differentiation of Th0 cells into Th1 cells. Th1 cells secrete a large amount of IFN-γ and IL-10. IFN-γ is the major macrophage-activating cytokine, which can enhance the microbicidal function of macrophages and stimulate macrophages to induce rapid production of IL-6, IL-10, IL-12, TNF-α. IL-10 is an inhibitor of activated macrophages, which inhibits Th1 cells and the production of Th1 cytokines and negatively regulates the production of IL-10 and IL-12 from macrophages, resulting in decreased plasma IFN-γ, IL-10 and IL-12 levels. On the other hand, under the effect of IL-4 and some other cytokines, Th0 cells can be induced to Th2 cells as well, which secrete cytokines such as IL-4, IL-6 and IL-10, stimulating the body to develop specific antibodies, promoting humoral immune responses and attenuating the immune injury mediated by Th1 cells. However, because of perforin deficiency in most FHL patients (Stepp et al, 1999; Kogawa et al, 2002), perforin-mediated cytotoxic activity of CD8+ T cells or NK cells is defective. So the effect of perforin, which downregulates an immune response by the elimination of antigen-presenting cells or by perforin-dependent activation-induced cell death, is impaired and CD8+ T cells may continue to receive activation and proliferation signals. In these circumstances, CD8+ T cells continue to proliferate and secrete large quantities of IFN-γ, stimulating the macrophage uncontrolled activation in turn (Kagi et al, 1994; Sad et al, 1996; Stepp et al, 1999, 2000; Badovinac et al, 2000), secreting high levels of IL-12 and other cytokines to activate CD4+ T cells and CD8+ T cells, which may form a cytokine loop developing into hypercytokinemia.

Figure 2.

 A hypothetic cytokine network in HPS. The network is mediated by Th1 cells, Th2 cells, macrophages, cytotoxic T lymphocyte and NK cell. Solid arrows indicate stimulation or secretion and dotted arrows indicate inhibition or killing. Red crosses indicate the blockage of the pathway. [Correction added after online publication 1 September 2008: in the above figure the term, TFN-γ was corrected to TNF-α.]


The authors thank Dr Chikong Li for his critical review of the manuscript and Mr Hongqiang Shen and Ms Baiqin Qian for their excellent technical assistance.


Yongmin Tang designed the study and wrote/revised the manuscript. Xiaojun Xu analyzed the data, performed the statistics and wrote the manuscript. Hua Song, Shuwen Shi, Shilong Yang, Jianwei, Binghua Pan, Fenying Zhao, Chan Liao were the responsible physicians. Chunfang Luo performed the flow cytometric determination of cytokines using the cytometric bead array (CBA).