The successful treatment of haemophagocytic syndrome in patients with human immunodeficiency virus-associated multi-centric Castleman's disease

Abstract

Both virus-associated haemophagocytic syndrome (HPS) and human immunodeficiency virus-associated multi-centric Castleman's disease (HIV-MCD) induced by human herpesvirus-8 (HHV-8) are extremely rare. We therefore wished to investigate their occurrence together, and establish the degree of cytokine activation present. From a prospective cohort of individuals with HIV-MCD, we investigated the incidence and outcomes of HPS and measured 15 inflammatory cytokines and the plasma HHV-8 viral loads before and during follow-up. Of 44 patients with HIV-MCD with an incidence of 4·3/10 000 patient years, four individuals (9%) were diagnosed with HPS. All are in remission (range 6–28 months) following splenectomy, etoposide and rituximab-based therapy. Plasma HHV-8 levels were raised markedly at presentation (median 3 840 000 copies/ml). Histological samples from spleen, splenic hilar lymph nodes and bone marrow demonstrated increased phagocytosis by histiocytes and presence of HHV-8-infected plasmablasts outside the follicles. Surprisingly, many known inflammatory plasma cytokines were not elevated, although interleukin (IL)-8 and interferon-γ were increased in all cases and IL-6 levels were raised in three of four patients. HPS in the setting of HIV-MCD is common and treatment can be successful provided the diagnosis is made appropriately. Systemic activation of cytokines was limited, suggesting that immunosuppressive therapy with steroids is not indicated in HHV-8-driven HPS.

Introduction

Haemophagocytic syndrome (HPS) is a reactive disorder of the reticulo-endothelial system characterized by histiocyte proliferation associated with uncontrolled activation of natural killer, CD4-positive T regulatory and CD8-positive cytotoxic T cells, leading to multi-organ failure when untreated [1]. First described in 1939 in a patient with sickle cell anaemia [2], this rare condition can be divided into primary or hereditary HPS and secondary or reactive HPS, the latter occurring in the setting of malignancies, autoimmune disease and/or infections [3–5]. Virus-associated secondary HPS has additional features more frequently, such as high fever, constitutional symptoms, liver dysfunction, coagulation abnormalities, peripheral blood cytopenias, hepatosplenomegaly and lymphadenopathy [3–5].

Clinically, patients with virus-associated secondary HPS present with similar sign and symptoms to multi-centric Castleman's disease (MCD)[6]. In addition, both have been related aetiologically to infection with the γ-herpesviridae, Epstein–Barr virus (EBV) and human herpesvirus-8 (HHV-8) (also known as Kaposi's sarcoma-associated herpesvirus, KSHV) [7–9]. While diagnostic criteria from the Histiocyte Society are well established [10,11], clinicians should be aware of HPS in individuals with fever, cytopenias, organomegaly and persistent γ-herpesvirus infection; early diagnosis and intervention are required.

We describe HPS in the setting of human immunodeficiency virus (HIV)-MCD, the largest case series in this setting to the best of our knowledge [8,12,13]. In addition, patients' cytokine and HHV-8 profiles were measured; surprisingly, a uniformly raised pattern of known inflammatory cytokines was not observed.

Materials and methods

The Chelsea and Westminster HIV cohort is one of the largest in Europe and we have collected clinical data prospectively on all HIV seropositive patients with histologically confirmed MCD at our organization. This extensive data set, that now contains 42 individuals, includes four patients who presented with HPS, as defined in accordance with the diagnostic criteria [10,11].

Fifteen serum cytokines were measured at the time of diagnosis of HPS and during follow-up in three separate panels; proinflammatory inflammatory cytokines, interleukin (IL)-1β, IL-6, IL-8, tumour necrosis factor (TNF)-α, granulocyte–macrophage colony-stimulating factor (GM-CSF); T helper type 1 (Th1)/Th2 cytokines, IL-2, IL-4, IL-5, IL-10, interferon (IFN)-γ and immunostimulatory cytokines, IFN-α, IL-12p40/p70, IL-13, IL-15 and IL-17, using FIDIS™ cytokine-coated beads (Biomedical Diagnostics, Oosterhout, the Netherlands). All experiments were carried out in compliance with the kit manufacturers' instructions. Briefly, for each experiment, antibody conjugated beads were prepared and aliquoted into a 96-well plate, before addition of either standards of known concentration or samples to be assayed. Subsequent steps involved washes interspersed by the addition of biotinylated detector antibody and later streptavidin–R-phycoerythrin solution. Once this preparation was complete, the plate was placed on the XY platform of the Luminex 100 instrument (Austin, TX, USA) for analysis. STarStation software was used for data acquisition and analysis. Standard curves were generated for each analyte, and the mean fluorescence intensity value of each analyte in each well was converted into a concentration using the standard curve. This value was then multiplied by the dilution coefficient 2 to give the concentration of the analyte in the original serum sample.

The HHV-8 viral load was measured using Lightcycler® quantitative polymerase chain reaction (PCR) (Roche, Lewes, UK) on DNA extracted from plasma using primers specific to the KSHV ORF-73 gene, as described previously in the setting of HIV-MCD[14,15]. Tissue samples of spleen, splenic hilar lymph nodes and bone marrow trephine biopsies were evaluated using a wide panel of immunohistochemistry. Appropriate ethical approval was obtained.

Results

Patients and histology

Four HIV seropositive patients presented with prolonged symptoms of pyrexia of unknown origin associated with hepatosplenomegaly and diffuse lymphadenopathy. All four were anaemic (haemoglobin range: 7·3–9·4 g/dl) and two were thrombocytopenic (21 and 48 × 10⁹/l) at presentation. Polyclonal hypergammaglobulinaemia was present in all patients and none had a detectable plasma paraprotein; all were hypoalbuminaemic (mean 17 g/l) and had raised serum ferritin levels (mean 11 413 µg/l), three of four had raised triglyceride levels (mean 3·3 mmol/l) and two of three had raised fibrinogen levels (mean 4·9 g/l). None of the patients had a previous acquired immune deficiency syndrome (AIDS)-defining illness, two were established on protease inhibitor-based highly active antiretroviral therapy (HAART) with undetectable plasma HIV viral loads and the remaining two individuals were not on HAART and had significant HIV viraemia (Table 1).

Table 1.

Presentation characteristics of four human immunodeficiency virus (HIV) seropositive patients with haemophagocytic syndrome in association with multi-centric Castleman's disease.

Case No	Sex	Ethnicity	Age	Duration HIV+ (years)	CD4 cell count (/mm3)	Plasma HIV viral load (copies/mm3)	Established on HAART	Duration of symptoms (months)	ECOG PS	KS	Plasma HHV-8 VL (copies/ml)
1	M	Black African	37	0·8	492	< 50	Yes	3	2	No	380 000
2	M	Caucasian	24	0·2	193	41 860	No	2	2	No	90 000 000
3	M	Caucasian	51	1·1	71	< 500 000	No	1	3	Yes	7 300 000
4	M	Black African	46	8·1	233	< 50	Yes	4	3	No	30 000

HAART, highly active retroviral therapy; HHV-8, human herpesvirus-8; KS, Kaposi's sarcoma; ECOG, Eastern Cooperative Oncology Group; VL, viral load.

These individuals have been treated successfully with a combination of single-agent etoposide chemotherapy, rituximab monoclonal antibody immunotherapy, and in three cases with splenectomy. All are alive in remission of MCD with no clinical symptoms or lymphadenopathy, undetectable plasma HHV-8 viral loads and normal C-reactive protein after a median follow-up of 10 months (range 6–28 months).

The spleens from three cases were enlarged measuring 19–23 cm in their greatest dimension, with masses of 1170–1600 g. In two cases the white pulp showed follicles with features of MCD. All three cases showed prominent and expanded red pulp, which was congested and showed an increased number of cells within the splenic cords. There were large atypical cells with prominent nucleoli scattered throughout the red pulp. These cells had features of plasmablasts and were seen both within cords and sinusoids. In addition there was a prominent histiocytic component within the sinusoids and these histiocytes showed active phagocytosis of red blood cells and other haemopoietic cells. Histiocytes with foamy cytoplasm and less obvious phagocytosis were seen in one case. The phagocytic histiocytes expressed CD68R. An HHV-8-latent nuclear-1 (LNA1) immunostain identified plasmablasts in both white and red pulp (Fig. 1). In addition, these cells expressed CD79a, multiple myeloma 1, B cell lymphoma 2 (BCL2), immunoglobulin (Ig)M and lambda light chains. They were focally positive for CD30, negative for IgD, kappa light chains, CD20, CD138, BCL6, T cell markers and cytomegalovirus. In situ hybridization for EBV (using an EBER probe) was negative. Splenic tissue was analysed for B cell clonality using PCR analysis for Ig heavy chain gene (FR2 and FR3 region primers) and Ig kappa light chain gene rearrangements. There was no evidence of monoclonal B cell expansion.

Fig. 1.

Section of spleen shows prominent red pulp with increased numbers of histiocytic cells both in the cords and the sinuses. In addition, there are larger atypical cells/plasmablasts. The phagocytic histiocytes are seen expressing CD68R and the plasmablasts show nuclear expression of human herpesvirus-8-latent nuclear-1 (HHV-8-LNA1).

Splenic hilar lymph nodes showed small follicles that had features of MCD. The lymph nodes had normal sinusoidal architecture with prominent medullary cords and sinuses. Medullary cords were rich in differentiated plasma cells, and larger lymphoid cells with features of plasmablasts were seen in two cases. The dilated sinusoids showed plump histiocytic cells with active phagocytosis in one case. Immunohistochemistry was similar to the spleen itself. In one case, the sinusoidal endothelial cells both within the lymph node and the lymphatics surrounding the lymph node demonstrated positivity for HHV-8-LNA1, possibly representing a forme-fruste of Kaposi's sarcoma.

The three bone marrow trephine biopsies were highly oedematous. Two cases showed myeloid predominance and two cases demonstrated focal megaloblastic erythropoiesis. Histiocytes were increased and prominent in two cases and less so in the other case. These cells showed phagocytosis of nucleated haemopoietic cells, apoptotic debris and expressed CD68R; there were no granulomas. The plasma cells were increased in numbers and amounted to 10–20% of the total marrow cellularity, and there was no evidence of light chain restriction. Occasional HHV-8-LNA1-positive large plasmacytoid cells (plasmablasts) were seen in two cases (both with prominent phagocytosis) (Fig. 2). It was not possible to document Ig light chain or IgM expression specifically in the HHV-8-positive plasmablasts. Immunostaining for parvovirus, known to be associated with HPS [5], was negative and in situ hybridization for EBV (EBER probe) did not identify any significant positive population (only an occasional bystander cell was positive). The reticulin was increased (grades 1–2). The stainable iron was moderately increased.

Fig. 2.

Section of a cellular bone marrow shows increased numbers of histiocytic cells amidst other haemopoietic lineage cells. The phagocytic histiocytes are seen expressing CD68R and there are occasional human herpesvirus-8-latent nuclear-1 (HHV-8-LNA1) plasmablasts showing nuclear expression of HHV-8-LNA1.

Cytokines and HHV-8

The HHV-8 levels were raised markedly in all four individuals (median 3 840 000 copies/ml, range 30 000–90 000 000 copies/ml) at presentation with one of the patients having one of the highest HHV-8 viraemias we have ever observed. This compared with the median plasma HHV-8 level of 12 800 copies/ml in 34 patients with HIV-associated MCD without HPS. HHV-8 viraemia decreased in all four patients during the course of therapy (Table 2).

Table 2.

Plasma human herpesvirus-8 (HHV-8) levels before and during therapy (median and range, copies/ml).

Baseline plasma HHV-8 levels	1-month plasma HHV-8 levels	2-month plasma HHV-8 levels	4-month plasma HHV-8 levels
3 840 000	420 500	500	3 700
(30 000–90 000 000)	(0–1 900 000)	(0–24 000)	(0–20 000)

Of the 15 cytokines measured, only IL-8 and IFN-γ were elevated in all the individuals at baseline and follow-up. IL-4, IL-6 and IL-10 were elevated in three of four patients and IL-1β, IL-2, IL-15, GM-CSF, TNF-α and IFN-α were normal at baseline (Table 3). Follow-up measurements were similar to those obtained at baseline. Only differences in IL-10, which was increased in one of four patients at follow-up versus three of four at baseline, and IL-17, which was increased in three of four patients at follow-up and one of four individuals at baseline, were observed in more than one patient comparing baseline with follow-up.

Table 3.

Plasma cytokine measurements at the time of diagnosis and during the follow-up of haemophagocytic syndrome (HPS) in all four patients. The first follow-up sample time-point was 4–6 weeks following the baseline sample, which corresponded to the start of etoposide/rituximab chemotherapy.

Cytokine measured	Normal range (pg/ml)	Median at baseline	Range at baseline	Elevated at baseline	Median at first follow-up	Range at first follow-up	Elevated at first follow-up
IL-1β	0–6	0	0–0	0/4	0	0–8	1/4
IL-2	0–3	0	0–0	0/4	0	0–15	0/4
IL-4	0–10	13	7–36	3/4	24	12–42	4/4
IL-5	0–5	3	0–11	1/4	1	0–11	1/4
IL-6	0–5	32	2–83	3/4	11	3–23	2/4
IL-8	0–8	158	36–587	4/4	49	34–108	4/4
IL-10	0–9	568	0–4844	3/4	0	0–503	1/4
IL-12	50–92	29	14–174	1/4	87	61–283	1/4
IL-13	0–10	0	0–16	1/4	0	0–0	0/4
IL-15	0–5	0	0–0	0/4	0	0–0	0/4
IL-17	0–14	0	0–73	1/4	36	0–36	3/4
GM-CSF	0–19	2	0–7	0/4	0	0–0	0/4
TNF-α	0–12	0	0–0	0/4	0	0–0	0/4
IFN-α	0–31	0	0–0	0/4	10	0–20	0/4
IFN-γ	0–4	14	8–20	4/4	8	8–14	4/4

GM-CSF, granulocyte–macrophage colony-stimulating factor; IFN, interferon; IL, interleukin; TNF, tumour necrosis factor.

Discussion

The HPS, a rare condition, was associated with HIV-MCD in four (9%) cases of our cohort including 44 patients. At present, there is little diagnostic or treatment consensus for secondary HPS which is dealt with generally on a case-by-case basis, depending on the associated infection(s) identified. Here, treatment with etoposide-based chemotherapy and rituximab was successful in all cases thus far, with all patients in remission after a median of 12 months' follow-up.

There is an increasing literature concerning immune dysregulation in HPS, in which a defect in granule (perforin/granzyme)-mediated cytotoxicity, which is important in both the killing of infected cells and the termination of the immune response, seems to be an underlying factor that predisposes individuals to HPS [16]. Since 1999, several genetic loci related to the activity of perforin and granzyme granules have been associated with genetic HPS, thus explaining the impaired or absent function of natural killer cells and cytotoxic T cells characteristic of the disease [17], although such findings cannot necessarily be extrapolated to reactive HPS. The deficiency in cytolytic activity results in persistent activation of lymphocytes and histiocytes, and an uncontrolled immune response leads to hypersecretion of proinflammatory cytokines (such as IFN-γ, IL-6, IL-8, IL-10, GM-CSF and TNF-α), an up-regulation of adhesion molecules and major histocompatibility complex I and II molecules on monomacrophages [18] and an expansion of inflammatory monocytes (i.e. an increase in CD14/CD16 expression) [19]. These mediators are secreted by the activated histiocytes and T lymphocytes, which infiltrate different tissues. In cases of HHV-8-driven HPS, we hypothesize that KSHV-encoded cytokines themselves contribute to the clinical course, although assays including those used here do not distinguish between viral and human homologues of IL-6. However, it is considered that an exaggerated inflammatory response is responsible for necrosis and organ failure and results in uncontrolled proliferation and phagocytic activity of histiocytes [1].

Despite this increasing literature concerning immune dysregulation in HPS, the majority of cytokines we measured were not elevated in all the patients. This is particularly surprising considering the ‘extreme’ HHV-8 viraemia we observed, including levels of more than 7 million copies/ml in two cases. While IL-6, a well-known marker of HHV-8 activity [20], was raised in three of four cases, IL-1β, IL-2, IL-15, GM-CSF, TNF-α and IFN-α were normal in all patients. In one previous study in the literature, the cytokine profile in fatal HIV–tuberculosis–EBV-associated HPS showed increases in the plasma Th1 cytokine IFN-γ, the Th1-related chemokine monokine induced by IFN-γ (MIG)/CXCL9, IFN-induced protein 10 (IP-10)/CXCL10, IL-8/CXCL8 and IL-6, the last group of which increased during therapy with foscarent [21]. The authors of that high-profile case report concluded by stating that immunosuppressive agents (i.e. steroids and anti-cytokine antibodies) may be beneficial in such severe HPS cases and demonstrated that cytokine profiling enhances the understanding and management of immune-mediated diseases.

While we have not measured chemokines, our cytokine data derived from four patients with HHV-8-associated HPS suggest that their approach should be treated with caution. Chemotherapy using dexamethasone, cyclosporine and etoposide (adopted by the Histiocyte Society in 1994 and updated more recently [11,22]) is recommended for severe HPS, including familial and EBV-associated cases. Use of bone marrow transplants has also been reported and the HLH-94 study in 113 children showed a 56% 3-year survival [23]. These data are unlikely to be directly relevant to HIV-1 seropositive adults with HHV-8-driven HPS. We consider that etoposide-based chemotherapy is appropriate treatment here and that rituximab should also be added to the list of potentially useful therapies.

The cytokine data we have shown is notable for the lack of heterogeneity which normally afflicts plasma measurements during the course of disease; data at follow-up were remarkably similar to baseline values. The most prominent change in plasma cytokine concentration was seen with IL-10 (decreases), which is produced by HHV-8-infected and HIV-tat-induced cell lines and has been linked to B cell activation and polyclonal hypergammaglobulinaemia in HIV-1 infection [24–26]; other data in HHV-8-driven disease suggest that it is a better marker of disease activity than IL-6. The role of IL-10 in the control of immune hyperactivation observed in one patient with HPS has been reported previously [27]. There are no studies on the role of IL-17 in HHV-8-associated diseases (to the best of our knowledge), but its role in immunoregulation is consistent with the increased levels we observed at follow-up [28].

We have demonstrated recently that the incidence of HIV-MCD in our cohort of over 10 000 patients measures 4·3/10 000 patient-years [95% confidence interval (CI): 2·7–6·4] (Bower et al., in press). We have not calculated the incidence of HPS here as, because of small numbers, 95% CIs would be extremely wide. However, physicians should be aware of HPS, and have heightened suspicion in cases of MCD.