Interleukin 6 expression by Hodgkin/Reed–Sternberg cells is associated with the presence of ‘B’ symptoms and failure to achieve complete remission in patients with advanced Hodgkin's disease

Authors


Dr P. G. Murray, Department of Pathology, Division of Cancer Studies, The Medical School, University of Birmingham, Edgbaston, Birmingham, B15 2TT, UK. E-mail: p.g.murray@bham.ac.uk

Abstract

Summary. Interleukin 6 (IL-6) is a potent immunomodulatory cytokine that has pathogenic and prognostic significance in a number of disorders. Previous studies in Hodgkin's disease (HD) have demonstrated the association between elevated serum levels of IL-6 and unfavourable prognosis, including advanced stage and the presence of ‘B’ symptoms and with reduced survival. Although IL-6 expression has been demonstrated in both the malignant Hodgkin/Reed–Sternberg (HRS) cells and in the various non-malignant cells present in HD biopsies, a relationship between expression of IL-6 by the tumour and outcome measures has not been established. The study group comprised of 97 patients with advanced HD who were recruited to two related clinical trials. IL-6 expression was determined on paraffin-wax sections of biopsy material by means of an immunohistochemical assay. Of the 97 patients, 27 (28%) showed staining for IL-6 in HRS cells. IL-6 expression by HRS cells was significantly correlated with a decreased likelihood of achieving a complete response to chemotherapy (P = 0·02) and with an increased prevalence of ‘B’ symptoms (P = 0·04). IL-6 expression by HRS cells was not associated with Epstein–Barr virus status (P = 0·57). In summary, the results suggest that IL-6 expression by HRS cells may contribute to the presence of ‘B’ symptoms and to a decreased likelihood to achieve a complete remission in HD patients.

Hodgkin's disease (HD) is an unusual form of lymphoma which is characterized by a low frequency of tumour cells, the so-called Hodgkin and Reed–Sternberg (HRS) cells, embedded in a background of non-neoplastic (reactive) cellswhich are believed to be recruited and activated by HRS cell-derived cytokines (Pinto et al, 1998). Multiagent combination chemotherapy is effective in HD but a proportion of patients, particularly those presenting with advanced disease, will relapse. Various phenotypic features are helpful predictors of unfavourable outcome. These include the presence of systemic (‘B’) symptoms, such as fever, night sweats, and pruritis, which are thought to be due, at least in part, to altered cytokine levels.

Interleukin 6 (IL-6) is an important immunomodulatory cytokine that can influence B- and T-cell growth and differentiation (Akira et al, 1993; Barton, 1996) and is able to act as a growth factor for malignant B cells or Epstein–Barr virus (EBV)-transformed B cells (Scala et al, 1990). IL-6 serum levels have been reported to be frequently elevated at the time of HD diagnosis. Furthermore, these levels normalize during remission (Seymour et al, 1997) and are associated with specific disease characteristics, including several adverse prognostic features such as the presence of ‘B’ symptoms, and with poorer survival (Kurzrock et al, 1993; Gorschluter et al, 1995; Seymour et al, 1997). IL-6 has been shown to be expressed by HRS cells in a proportion of HD cases and also by several HD-derived cell lines (Tabibzadeh et al, 1989; Jucker et al, 1991; Klein et al, 1992; Tesch et al, 1992; Brown et al, 1997). However, IL-6 is also produced by a number of lymphoid and non-lymphoid cell types in HD tumours, including macrophages, fibroblasts and endothelial cells (Brown et al, 1997).

Despite these studies, it has yet to be established whether IL-6 expression by malignant HRS cells is associated with an unfavourable prognosis. If such a relationship were found, it might reinforce the view that IL-6 has a pathogenic role in HD and could suggest strategies for therapeutic intervention. Accordingly, we have investigated the relationship between IL-6 expression in HD tumours with various patient and disease characteristics and the response to chemotherapy, survival and failure-free survival.

Patients and methods

Patients.  The study patients were from two related clinical trials. The first was a phase II trial, conducted by the United Kingdom Central Lymphoma Group (CLG), investigating the use of alternating ChlVPP/PABlOE (chlorambucil, vinblastine, procarbazine, prednisone/prednisone, adriamycin, bleomycin, vincristine, etoposide) chemotherapy in patients with advanced HD who were unsuitable for treatment with radiotherapy alone (the HO2001 trial). The second was the phase III follow-on trial, which compared the ChlVPP/PABlOE regimen with PABlOE alone (the HO3001 trial). The results of these trials have been reported elsewhere (Cullen et al, 1994; Hancock et al, 2001) In total, 97 histologically confirmed cases from these two trials were available for the present study.

Section preparation.  Haematoxylin and eosin-stained slides were prepared and classified according to the Rye classification system. Prior to the immunohistochemical and in situ hybridization assays, 4 μm paraffin-sections were prepared on charged microscope slides (SurgiPath, Peterborough, UK) and heated for 1 h at 65°C. Sections were dewaxed, rehydrated and endogenous peroxidase activity blocked with 0·3% H2O2 in methanol for 15 min, followed by a wash in tap water.

Immunohistochemistry for IL-6.  Sections were initially subjected to a low temperature retrieval method. Briefly, slides were incubated in 1 mmol/l EDTA pH 8·0/0·1% Tween 20 on a hot-plate stirrer at 65°C for 16 h (overnight). Agitation was achieved using a 30 × 5 mm magnet bar, with the stirrer set to 600 r.p.m.

Sections were washed in tap water and mounted onto a Sequenzer (Shandon, Runcorn, UK). After a Tris-buffered saline pH 7·6 (TBS) wash, primary antibody to IL-6 (cat# sc1266, Autogen Bioclear, Wiltshire, UK) was applied at a dilution of 1:200 for 1 h at room temperature. Slides were washed in TBS pH 7·6 containing 0·001% Tween 20. A universal streptavidin/biotin HRP kit (Binding Site, Birmingham, UK) was then applied to detect bound primary antibody. Briefly, this involved incubation in secondary antibody (diluted 1:100 in TBS) for 20 min, followed by a wash in TBS/0·001% Tween 20. Sections were then incubated in tertiary reagent (also diluted 1:100) for 20 min. Following a final wash in TBS/0·001% Tween 20, Vector NovaRed chromogen (Vector, Peterborough, UK) was applied for 5 min. Sections were then washed in tap water, counterstained in Mayers Haematoxylin, dehydrated, cleared and mounted. Tonsil was used as a positive control for the IL-6 assays. Negative controls consisted of consecutive test sections in which primary antibody was replaced with non-immune serum.

Further validation of the specificity of the antibody reagent for IL-6 was achieved by staining HeLa cells thathad been transfected either with a latent membrane protein-1 (LMP1) expression vector (pSG5-LMP1) or control vector. In cells transfected with pSG5-LMP1, LMP1 expression has previously been shown to induce a 130-fold increase in IL-6 production compared with vector-only transfected cells (as assessed by an IL-6-specific enzyme-linked immunosorbent assay) (Eliopoulos et al, 1999). IL-6 expression was also confirmed in positive HD cases by the application of a mouse monoclonal antibody specific for IL-6 (Research Diagnostics, Flanders, NJ, USA; cat#: RDI-IL6abmX), although this antibody gave generally weaker staining than the polyclonal reagent.

IL-6 expression within each HD tumour was assessed microscopically and cases classified according to whether HRS cells expressed IL-6 or not. The percentage of HRS cells expressing IL-6 was also recorded in all cases. In addition, the percentage of non-malignant cell types showing IL-6 expression was also noted in each tumour. All measurements were performed on a minimum of 25 high-power fields.

Detection of latent EBV infection. In situ hybridization for the detection of EBV-encoded RNA (EBER) was performed according to standard methodology (Barletta et al, 1993). Positive controls for EBER in situ hybridization included paraffin-wax sections of lymphoblastoid cell lines (LCLs) grown as solid tumours in severe-combined immunodeficient (SCID) mice and a known EBER-positive HD case. U6 and sense control probes were also included in all runs and their use has been previously described elsewhere (Barletta et al, 1993). Immunohistochemistry for LMP1 was also performed to confirm the presence of EBV infection in HRS cells. The standard alkaline phosphatase–anti-alkaline phosphatase method was used and sections were microwave-pretreated for 20 min in a standard citrate buffer. Positive controls for LMP1 consisted of paraffin-wax sections of LCLs grown as solid tumours in SCID mice. Negative controls consisted of consecutive test sections in which primary antibody was replaced with non-immune serum of the same IgG subclass.

Statistical methods.  The IL-6-present (IL-6 expression present in HRS cells) and -absent (IL-6 expression absent from HRS cells) groups were compared in terms of baseline patient characteristics. Chi-square tests were used for sex, histology, clinical stage, B-symptoms and EBV status. Chi-square tests or Fisher's exact tests (for small frequencies) were used for sites of disease. The Wilcoxon test was used for age.

Before assessing the effect of IL-6 expression on outcome, the IL-6 groups were checked for comparability of treatment received. Chi-square tests and Fisher's exact tests were used to compare groups in terms of the trial to which they were recruited and the treatment they received prior to and during the trial. Wilcoxon tests were used to compare the groups in terms of the numbers of courses of chemotherapy.

The IL-6 groups were compared in terms of response to chemotherapy using a chi-square test, with complete response (CR) defined as a complete disappearance of all disease and partial response (PR) as a disappearance of at least 50% of known disease. Groups were compared in terms of survival and failure-free survival using a log-rank test. Two and 5 year survival and failure-free survival rates were calculated and compared using Kaplan–Meier estimates. Survival was calculated as the time from beginning chemotherapy to date of death from any cause or date of the last known follow-up. For those patients who responded to treatment (i.e. CR or PR), failure-free survival was calculated as the time from beginning chemotherapy to either date of relapse, or in those patients with no previous documented relapse, either the date of death from any cause or date of the last known follow-up. For those patients who did not respond to chemotherapy, failure-free survival was recorded as zero.

Logistic regression analysis was used to investigate the combined effect of sex, age, presence of B-symptoms, clinical stage, histology and IL-6 status on response to chemotherapy. Similarly, Cox regression analysis was used to assess the combined effects of these variables on survival and failure-free survival. All regression analyses used a stepwise selection method to identify the most significant explanatory variables with an entry criteria of P = 0·10 and a staying criteria of P = 0·10.

For the 27 patients whose HRS cells expressed IL-6, the percentage of cells expressing IL-6 was categorized as < 10%, 10–19%, 20–29%, etc. For all patients, the percentage of non-malignant cell types showing IL-6 expression was categorized as < 1%, 1–4%, 5–9%, 10–19%, 20% or more. The association of each of these ordinal variables with chemotherapy response was assessed using a Wilcoxon test, and with survival and failure-free survival using a Cox regression analysis.

All quoted P-values are two-sided and statistical significance relates to results in which the P-value is < 5%.

Results

IL-6 expression by HRS cells was identified in 27/97 (28%) patients, where the staining appeared both granular and cytoplasmic (Fig 1). The numbers of HRS cells expressing IL-6 in these cases was variable. IL-6 expression was also documented in endothelial cells, macrophages and some lymphocytes in all cases irrespective of IL-6 expression in HRS cells. The specificity of the polyclonal reagent for IL-6 was confirmed by positive staining in HeLa cells transfected with an LMP1 expression vector (Fig 2). These cells have been previously shown to express high levels of IL-6 (Eliopoulos et al, 1999). Control cells (cells transfected with vector only) did not stain with the antibody reagent (Fig 2).

Figure 1.

IL-6 expression in HRS cells of Hodgkin's disease. Immunohistochemical staining for IL-6 was characteristically granular and cytoplasmic in HRS cells (original magnification ×400).

Figure 2.

Validation of the specificity of IL-6 immunohistochemistry. Upper panel (A) shows strong staining for IL-6 in HeLa cells transfected with the EBV latent membrane protein-1 gene. LMP1 has previously been shown to induce high-level expression of IL-6 in these cells. Lower panel (B) shows absence of staining in vector-only transfected cells. Original magnification ×600.

The characteristics of the patients on entry to the trials are given in Table I. There were no statistically significant differences between the IL-6-present and -absent groups in terms of sex, age, clinical stage, subtype histology and sites of disease, but B-symptoms were more prevalent for the IL-6-present patients (74%vs 51%; P = 0·04). There was no association between IL-6 and EBV status. There was no statistically significant differences between the groups in terms of the trials they participated in and the treatment they received (Table II). Most study patients were from the HO2001 trial (79%) and received ChlVPP/PABlOE chemotherapy (87%). The mean number of courses of ChlVPP and PABlOE received by the study patients were 3·2 and 3·6 respectively.

Table I.  Baseline patient characteristics by IL-6 status.
 IL-6 present
n = 27
IL-6 absent
n = 70
P-value
  • *

    Includes nodular sclerosis + lymphocyte depleted.

  • Tests nodular sclerosis against all other types.

Sex:
 Male17 (63%)48 (69%)0·60
 Female10 (37%)22 (31%) 
Age (years):
 Median29330·23
 Inter-quartile range23–4525–50 
 Range15–6515–68 
B-symptoms:
 Present20 (74%)36 (51%)0·04
 Absent7 (26%)34 (49%) 
Clinical stage:
 1/210 (37%)30 (43%)0·57
 3/417 (63%)39 (57%) 
 Not known01 
Histological type:
 Lymphocyte predominant1 (4%)7 (10%)0·93
 Nodular sclerosis*18 (67%)46 (66%) 
 Mixed cellularity5 (19%)13 (19%) 
 Lymphocyte depleted3 (11%)4 (6%) 
Enlarged lymph nodes:
 Cervical22 (81%)50 (71%)0·31
 Axillary10 (37%)25 (36%)0·90
 Inguinal3 (11%)10 (14%)> 0·99
 Abdominal4 (15%)7 (10%)0·49
 Waldeyer's ring0 (0%)2 (3%)> 0·99
 Hepatomegaly1 (4%)5 (7%)> 0·99
 Splenomegaly7 (26%)11 (16%)0·25
EBV status:
 Present8 (35%)17 (28%)0·57
 Absent15 (65%)43 (72%) 
 Not known410 
Table II.  Treatment details by IL-6 status.
 IL-6 present
n = 27
IL-6 absent
n = 70
P-value
  • *

    Tests none vs some.

  • Tests ChlVPP/PABlOE and PABlOE/ChlVPP vs PABlOE alone.

Trial:
 HO200120 (74%)57 (81%)0·42
 HO3001 7 (26%)13 (19%) 
Prior treatment:
 None26 (96%)60 (86%)0·28*
 Radiotherapy only 0 4 
 Chemotherapy only 1 3 
 Radiotherapy and chemotherapy 0 3 
Trial treatment:
 ChlVPP/PABlOE25 (93%)59 (84%)0·28
 PABlOE/CHlVPP 1 (4%) 2 (3%) 
 PABlOE only 1 (4%) 9 (13%) 
Number of courses of ChlVPP:
 n27700·10
 Mean 3·4 3·0 
 Standard deviation 0·97 1·38 
Number of courses of PABlOE:
 n27690·65
 Mean 3·6 3·6 
 Standard deviation 0·97 1·30 

The median follow-up time for the 74 patients on study who were still alive at last known follow-up was 5 years. The response to chemotherapy and survival analysis for the two IL-6 groups is shown in Table III. The complete response rate to chemotherapy was significantly greater in the IL-6-absent group (80%vs 56%; P = 0·02). There was no difference in survival (P = 0·75) and although the observed data showed a trend for the IL-6-absent group to have a better failure-free survival, this was not statistically significant (P = 0·27). There were no significant differences between the groups in terms of 2 and 5 year survival, and failure-free survival rates.

Table III.  Analysis of outcome measures by IL-6 status.
 IL-6 present
n = 27
IL-6 absent
n = 70
Difference (95% CI)P-value
  • *

    Compares complete vs partial response/fail.

Response to chemotherapy:
 Complete15 (56%)55(80%) 0·02*
 Partial10 (37%)11 (16%)  
 Fail2 (7%)3 (4%)  
 Not known01  
Survival:
 2 year survival rate80·7%85·5%5% (−13% to 22%)0·59
 5 year survival rate76·5%79·0%3% (−20% to 25%)0·82
Failure-free survival:
 2 year survival rate69·4%81·2%12% (−8% to 32%)0·24
 5 year survival rate60·9%72·9%12% (−13% to 37%)0·34

The results from the regression analyses are shown in Table IV. The logistic regression analysis selected IL-6 and age as significant explanatory variables for the probability of a complete response to chemotherapy, with IL-6 presence and increasing age reducing the chance of a complete response. IL-6 was the only significant explanatory variable when the analysis excluded the 11 patients who had received prior treatment.

Table IV.  Models from multiple regression analysis.
 Estimate
(Standard Error)
P-valueRisk ratio
(95% CI)
Complete response to treatment:
 IL-6 present−1·38 (0·53)0·0080·25 (0·09–0·70)
 Age−0·03 (0·02)0·040·97 (0·94–1·00)
Survival:
 Nodular sclerosis−0·84 (0·42)0·040·43 (0·19–0·98)
Failure-free survival:
 Nodular sclerosis−0·69 (0·36)0·060·50 (0·25–1·02)

Nodular sclerosis histology was the only explanatory variable for survival and failure-free survival. Patients with this histological type had a reduced risk of death or failure.

For the 27 IL-6-present patients, there was no statistically significant association between the percentage of HRS cells expressing IL-6 and response to chemotherapy (P = 0·41), survival (P = 0·41) or failure-free survival (P = 0·87). There was also no statistically significant association between the percentage of non-malignant cells expressing IL-6 and response to chemotherapy (P = 0·70), survival (P = 0·55) and failure-free survival (P = 0·54).

Discussion

Elevated serum levels of IL-6 are associated with poor prognosis in a number of malignancies, including multiple myeloma, ovarian cancer and renal cancer, as well as in non-Hodgkin's lymphomas and Hodgkin's disease, suggesting a pathogenic role for this cytokine in these tumours. In HD, IL-6 levels have been shown to correlate with adverse disease characteristics, such as the presence of ‘B’ symptoms, and with poorer survival (Kurzrock et al, 1993; Gorschluter et al, 1995; Seymour et al, 1997; Vener et al, 2000). Although there have been numerous reports documenting IL-6 expression in HD biopsies (Tabibzadeh et al, 1989; Jucker et al, 1991; Tesch et al, 1992; Foss et al, 1993; Herbst et al, 1997), it is still a matter of speculation whether the HRS cells are the source of the elevated serum levels. Further, the impact of IL-6 expression by HRS cells on prognosis, including response to treatment and outcome, in HD patients has not been determined. Although one previous study found no association between IL-6 expression in tumours and the presence of ‘B’ symptoms, this study included only 23 patients (Foss et al, 1993). In the present study, we demonstrated that IL-6 expression by the malignant HRS cells of HD tumours does correlate with the presence of ‘B’ symptoms, suggesting that HRS cell-derived IL-6 may be a major contributor to the appearance of ‘B’ symptoms and could be responsible for the elevated serum levels of this cytokine observed in some patients. In a previous study, elevated serum IL-6 levels were shown to correlate with increased immunoreactivity for IL-6 in HD tumours, and a variety of cell types, including the HRS cells, were shown to be responsible for this increased expression (Brown et al, 1997). However in the present study, we have found no relationship between IL-6 expression by the non-malignant cells and any outcome variable.

The presence of ‘B’ symptoms in HD patients is generally associated with a poor prognosis and with a more treatment-resistant phenotype. Thus, the finding that IL-6 expression was correlated with a decreased likelihood of a complete response following chemotherapy in our patient group is perhaps not surprising. However, IL-6 is a major survival factor in other B-cell malignancies and it is possible it may play a more direct role in HRS cell survival. In multiple myeloma, in vitro and in vivo studies show that IL-6 both promotes tumour survival and growth and prevents spontaneous or dexamethasone (DXM)-induced apoptosis (Hardin et al, 1994). In the human myeloma cell line, U266, for example, IL-6 induces STAT3 activation, leading to protection from apoptosis by upregulation of Bcl-xL, and blocking IL-6 receptors in these cells induces apoptosis (Catlett-Falcone et al, 1999). HRS cells and HRS-derived cell lines also express IL-6 receptors (Klein et al, 1992; Tesch et al, 1992; Herbst et al, 1997), suggesting a possible role for autocrine or paracrine stimulation of HRS cells, which may in turn lead to protection from apoptosis. Further work is clearly required to determine whether IL-6 has a direct influence on chemo-resistance in HD.

Consistent with the observation that in a variety of cellular backgrounds IL-6 gene transcription is increased following activation of the EBV-encoded latent membrane protein-1 (LMP1) (Eliopoulos et al, 1997, 1999), a previous study has reported significantly higher proportions of cases with IL-6-expressing HRS cells in LMP1-positive as compared with LMP1-negative HD tumours (Herbst et al, 1997). However, we found no such association. A possible explanation may be that the study by Herbst et al (1997) detected IL-6 transcripts whereas we have detected the IL-6 protein. An alternative possibility is that in EBV-negative HD cases, IL-6 might be transcriptionally regulated by other members of the tumour necrosis factor receptor family, including CD30 and CD40 which are regularly expressed by HRS cells (Younes & Carbone, 1999). These molecules expressed by tumour cell populations may, in the presence of the appropriate ligand, result in cellular signalling leading to IL-6 gene transcription. CD40, for example, is a potent inducer of STAT3 (Hanissian & Geha, 1997) and NF-κB activation (Eliopoulos et al, 1997), which in turn are able to induce IL-6 expression. In fact, constitutive NF-κB activation has been reported to be a consistent feature of HRS cells (Bargou et al, 1997; Wood et al, 1998), even in the absence of EBV-LMP1.

Therapy with anti-IL-6 murine monoclonal antibodies (mAbs) has been shown to be beneficial in some patients with advanced multiple myeloma (Klein et al, 1991; Bataille et al, 1995; van Zaanen et al, 1996). Likewise, administration of an anti-IL-6 mAb to patients seropositive for HIV-1 and suffering from an immunoblastic or a polymorphic large-cell lymphoma resulted in complete abrogation of ‘B’ clinical symptoms (Emilie et al, 1994). More recently, the potentially beneficial effects of anti-IL-6 therapy were reported in transplant recipients with B-lymphoproliferative disorder (Haddad et al, 2001). These and other studies support the notion that anti-IL-6 therapy might be beneficial in a subset of HD patients.

Acknowledgment

We wish to thank Dr Michael Cullen for access to the tumour material and clinical data used in this study.

Ancillary