American Journal of Hematology 39:163-171 (1992)
Radioimmunoassay for the Measurement of Serum 11-6 and Its Correlation With Tumour Cell Mass Parameters in Multiple Myeloma Eric Soiary, Michel Guiguet, Veronique Zeller, Rene-Olivier Casasnovas, Denis Caillot, Pascal Chavanet, Henri Guy, and Gerard Mack Service d’Hematologie Clinique (E.S.,R.O.C.,D.C.,H.G.), Laboratoire de Biochimie (M.G.,V.Z.,G.M.), and Service des Maladies infectieuses (P.C.), C.H.U. Le Bocage, Dijon, France
Interleukin-6(IL-6) was demonstrated to be a strong autocrine or paracrineplasmocytoma cell growth factor in humans. Using a bioassay, high serum IL-6 (S-IL-6) levels were correlated with disease severity in plasma cell dyscrasias. Since other cytokines could interfere with the bioassays, we developed a specific radioimmunoassay to study S-IL6 levels in 102 patients with monoclonal gammopathy (MG). S-IL-6 level was studied by a double antibody radioimmunoassay using a rabbit polyclonal anti-lL-6 antibody and a human recombinant IL-6 as the standard. The lowest value of the standard significantly different from zero was found to be 78 pglml. Within-run and between-runprecisions were characterized by a mean coefficient of variation of 3.72 and 5.5%, respectively. The mean analytical recovery was found to be 113% and the immunochemical identity of 11-6 standard and S-IL-6 was shown by dilution tests. IL-6 was detected in all tested sera. Sera from 66 healthy volunteers and 43 patients with acute leukemia or malignant lymphoma were tested as controls. In healthy subjects, S-IL-6 values were 294 86 pglml. MG were classified as multiple myeloma(MM), macroglobulinemia, and MG of undetermined significance (MGUS). The distribution of S-IL-6 levels in patients with MG was significantly higher than in healthy subjects but lower than in patients with acute leukemia or Hodgkin’s lymphoma. Results obtained in 55 patients with MM were related to other biological parameters. S-IL-6 levels correlated with bone-marrow plasmacytosis (P < .0005), serum-lactate dehydrogenase (S-LDH; P < .005), serum p2 microglobulin (S 432111; P < .Ol), and serum calcium (S-Ca; P < .025) and inversely correlated with haemoglobin (P < .025). Our results indicate that 1) radioimmunoassay is suitable for the measurement of human IL-6 in serum; 2) high S-IL-6 levels are observed in a small number of patients with MG; and 3) S-IL-6 level correlates with tumour cell mass in patients with overt MM.
*
Key words: interleukin-6, monoclonal gammopathy, plasma cell dyscrasias
INTRODUCTION Interleukin-6 (IL-6) is a polypeptide cytokine which is produced by various types of lymphoid and non-lymphoid cells including monocytes, T and B cells, fibroblasts, keratinocytes, endothelial cells, mesangium cells, and various tumor cells. It has been demonstrated to have several biological activities in immune system, hematopoiesis, acute phase reactions, and neural system [l]. IL-6 was originally identified as a T cell-derived lymphokine that is not involved in the growth of normal activated B cells but induces final maturation of B cells into antibody producing cells. It is a potent growth factor for murine plasmocytomas and B cell hybridomas [2] and 0 1992 Wiley-Liss, Inc.
a strong autocrine or paracrine in vitro growth factor in human myelomas [3,4]. The in vitro responsiveness of myeloma cells to IL-6 was demonstrated to be directly correlated with their proliferative status in vivo [ 5 ] . IL-6 was shown to be overproduced in vitro in the bone marrow from patients with active multiple myeloma [4]. Consistent with this observation, serum levels of IL-6 determined in a bioassay using the B9 IL-6 dependent
Received for publication April 30, 1991; accepted August 15, 1991 Address reprint requests to Eric Solary, MD, HCmatologie Clinique, C.H.U. Le Bocage, 21034 Dijon Cedex, France.
164
Solary et al.
murine hybridoma were recently described as a good reflection of disease activity in plasma cell dyscrasias [6]. Thus, 1L-6 seems to be involved in vivo during the progressive phase of multiple myeloma. Nevertheless, since another cytokine named IL- 11 was recently described as a hybridoma growth factor [7J that could interfere in bioassays, we developed a specific radioimmunoassay (RIA) in order to detect IL-6 in serum from each healthy control subject tested. Using this sensitive RIA, we planned this study in order to investigate serum IL-6 level (S-IL-6 level) in patients with monoclonal gammopathies (MG) and evaluated the relationships between S-IL-6 level and other laboratory variables at diagnosis or during the course of multiple myeloma (MM). As increased S-IL-6 level was described in some acute leukemias [8] and malignant lymphomas [9], patients with acute leukemia or Hodgkin’s or non-Hodgkm’s lymphoma at diagnosis were tested as controls. PATIENTS AND METHODS Patients
Serum samples were obtained from three groups of subjects: group I consisted of 66 healthy volunteers; group 2 consisted of 102 patients with MG; and group 3 consisted of 52 patients with acute leukemia (n = 18), Hodgkin’s lymphoma (n = 8), non-Hodgkin’s lymphoma (n = 24), or Castleman’s disease (n = 2) at diagnosis. In all cases, blood samples were collected in 10 ml sterile silicone coated tubes with inert gel barrier and clot activator disc (Venoject VT- I09SAS, Terumo Medical Corporation, Leuven, Belgium). Tubes were centrifugated at 2,500 rpm for 10 minutes and serum was stored at -20°C until assays. Patients with overt inflammatory or infectious process were not included in the study. The I02 patients with MG underwent bone-marrow examination, serum protein electrophoresis, full skeletal radiographic surveys and were divided as follows: macroglobulinemia (n = 19), overt MM (n = 5 3 , and monoclonal gammopathies of undetermined significance (MGUS; n = 28). Fifty-nine samples were obtained at diagnosis before any treatment. Nine of the 19 patients with macroglobulinemia and 34 of the 55 patients with MM were previously treated at the time of the study. Macroglobulinemia was defined by the production of a monoclonal immunoglobulin M (serum level > 5 g/L) and unequivocal infiltration of well-differentiated lymphocytes and atypical or lymphocytic plasma cells in bone marrow biopsies. Among the 9 treated patients, 6 were in a plateau phase and 3 were in evolution. Overt MM was defined by previously described criteria [lo]. An inventory of the clinical and laboratory features of all patients at the time of the study was collected, including age, sex, immunoglobulin (Ig) heavy and light chain type, haemoglobin, platelet count, serum calcium,
serum albumin, serum creatinine, serum lactate dehydrogenase (S-LDH), C reactive protein (CRP), percentage of plasma cells in the bone marrow, and extent of bone lesions. According to the Durie and Salmon’s staging system [ 101, 25 patients had stage I or stage I1 MM and 30 had stage 111MM. Among treated multiple myelomas, 14 were in the plateau phase (partial or complete remission) and 19 had progressive disease or were relapsing. Among the patients, 69% had IgG, 20% had IgA, and 11% had pure Bence Jones MM. Kappa subtype was present in 56.4% and lambda in 43.6%. Sera of the longitudinal studies were always collected immediately before chemotherapy. Normal bone marrow examinations and skeletal surveys were found in 28 patients who fulfilled the criteria of monoclonal gammopathies of undetermined significance (MGUS) [ 1 11. MATERIALS AND METHODS Chemicals
IL-6 used in this study was the recombinant human IL-6 (rhIL-6) standard from the Intertest-6 kit (Genzyme, Boston). As indicated by the manufacturer, the mass of IL-6 contained in this standard was calibrated by comparison with an IL-6 reference standard supplied by the National Institute for Biological Standards and Control (Hertfordshire, UK). The specific activity of the material determined in an IL-6 dependent murine hybridoma cell proliferation assay is reported to be 5. lo6 Ulmg. Polyclonal rabbit anti-human IL-6 antiserum was purchased from Genzyme Inc. As indicated by the manufacturer, this antiserum does not cross-react with IL-I (a or p), TNF ( a or p), GM-CSF, and IFN (a or p). The specificity of this anti-IL-6 antibody for the relevant Jigand in the presence of IL-I (a or p) and TNF a was confirmed [ 121. ’251-rhIL-6(1,000-1,500 Ci/mmol) was obtained from Amersham laboratories (Les Ulis, France). Immunobead goat anti-rabbit immunoglobulin antibody and immunobead rabbit anti-mouse immunoglobulin antibody originated from Bio-Rad laboratories (Richmond, CA). Other chemicals were of reagent grade. RIA Procedures
The assay format chosen was a double antibody radioimmunoassay. The regular buffer (PB) used at all stages of the assay consisted of 0.05 M sodium phosphate, pH 7.5, containing 1 mM EDTA and 0.1% sodium azide. In order to minimize nonspecific protein adsorption, especially I2%rhIL-6, polystyrene test tubes were precoated with bovine serum albumine (B.S.A.; 20 g/L in PB) at 37°C for 2 hours and then washed with PB. The assay was performed in the presence of PB supplemented with B.S.A. 2 g/L. One hundred microlitres either of standards ranging from 78 to 5,000 pglml or of test samples
Serum IL-6 Radioimmunoassay
were incubated at room temperature for 6 hours with 100 pl of the anti-IL-6 serum dilution. This dilution lay between 1:25 and 150 according to anti-IL-6 titration curves performed with each different lot. The second incubation, after addition of 100 pl of '251-rhIL-6 (50 pg), was for 48 hours at 4°C. A third incubation, after adding 1.5 mg of immunobead goat anti-rabbit immunoglobulin antibody, took 2 hours at 37°C and then 2 ml of cold PB was added to each tube. Bound and free IL-6 were separated by centrifugation (3,500g for 30 min at 4"C), supernatants were decanted and the radioactivity of the pellets counted. Test samples and standards were measured in duplicate. Results were expressed as percent bound when compared with samples without IL-6 (B,) and plotted against a logarithmic scale of the IL-6 concentration. In order to point out the specificity of the RIA, we studied the effect of an IL-6 depleted serum on the displacement of Iz5I-rhIL-6 from the antibodies. One millilitre of a high-content sample serum (1,600 pg/ml) was incubated for 3 days at 4°C with an excess (80 pg) of anti-human IL-6 monoclonal antibody (BE-8; Dr Widjenes, C.T.S. Besanson, France). After a 2-hour incubation at 37°C with 1 mg of immunobead second antibody, the immune-precipitate was removed by centrifugation (3,50Og, 15 min at 4°C) and the IL-6 depleted supernatant was collected.
Serum p2 Microglobulin Serum p2 microglobulin (S-parn) levels were measured using the RIA kit purchased from Immunotech laboratories (F- 13288, Marseille, France). Corrections for renal function were not used in the calculation of the S-P2m levels.
165
1
*O 0
10
100
1000
10000
IL6 concentration (pg/rnl)
Fig. 1. Within-assay precision of the standard curve for IL-6 radioimmunoassay. For each standard concentration, the binding of '251-rhlL-6to the antibodies (6)is expressed as the percentage of the binding observed in the absence of IL-6 standard (60).Each point represents the mean of five individual values. The bars indicate the 95% confidence interval for the true mean.
.05 was considered to represent a significant difference or relationship. RESULTS IL-6 Radioimmunoassay
When IL-6 depleted serum was tested, no displacement of '251-rhIL-6 from the antibodies was observed, attesting to the specificity of this assay for IL-6. The within-run precision of the standard curve (Fig. 1) was characterized by a 95% confidence interval placed on the true mean for each standard concentration (n = 5). Serum Thymidine Kinase The coefficient of variation ranged from 0.89% for the Serum thymidine kinase (S-TK) levels were measured low standard values up to 7.73% for the high standard using the Prolifigen TK-REA radioenzymoassay values (mean was 3.72%). The lowest value of the stan(Sangtec, Pharmacia, France). The average S-TK level in dard significantly different from zero (a< 0.05) in this healthy subjects is estimated to be 2.4 units per L serum. typical experiment was found to be 78 pg/ml. Eight conS-TK values > 5 unitdl are referred to as being elesecutive standard curves were used to examine the "bevated. tween-assay'' precision. Within a given experiment, each standard was measured in duplicate and the mean value Other Parameters was considered as an individual IL-6 value for the statisBone-marrow plasmacytosis was determined on tical analysis. The coefficient of variation ranged from smears stained by May-Griinwald-Giemsa. Levels of al- 1.55% for the low standard values up to 11.52% for the bumin, C reactive protein, and immunoglobulins were high standard values (mean was 5.5%). Precision data for measured by nephelometry . All other biological parame- serum samples whose mean IL-6 content lay within the ters were determined by routine analysis. standard range of the assay are summarized in Table I. The recovery of the assay was performed as follows. A Statistical Analysis known quantity of rhIL-6 standard ( I ng) was added to a Comparisons between groups were done by the non serum whose mean IL-6 content was 315 pgiml (n = 5). parametric U test of Mann-Whitney (MW-test). Relation- The mean IL-6 content of the mixture was found to be ships between IL-6 and other biochemical parameters 1,496 pg/ml (n = 5). Thus, the mean recovery in this were done by correlation analysis. A P value of less than experiment was 113%. The immunochemical identity of
166
Solary et al. TABLE 1. Precision Data for Serum Samples Whose Mean IL-6 Content Lies Within the Standard Range of the Assay
Mean IL-6 content (pg/ml)
C.V.a (%)
315 803 1,496 1,687 455
12.8 3.0 2.2 9.6 5.7
Within run (n = 5 )
Between run (n = 7) ”.V.
= coefficient of variation.
TABLE II. Dilution Tests* ~
Dilution
Mean IL-6 content (pglml) (n = 5)
~
Theoretical value (pg/ml)
IL-6 content: theoretical value (%)
~
0 1:2
1:4 1:8
1,687 803 336 195
843.5 422 211
95 80 92
*The immunochemical identity of IL-6 standard and IL-6 measured in the serum was studied by diluting high content serum with RIA buffer.
IL-6 standard and IL-6 measured in serum was studied by diluting a high content serum with the assay buffer. Results of this experiment are summarized in Table 11. IL-6 Serum (S-IL-6) Levels
stage IV Hodgkin’s lymphoma (8,211 pg/ml), and high grade non-Hodgkin’s lymphoma (8,887 pg/ml). The distribution of S-IL-6 levels is not significantly different in the 55 patients with overt MM when compared with patients with non-Hodgkin’s lymphoma at diagnosis. This distribution is significantly lower in the 55 patients with overt MM than in patients with acute non lymphoblastic leukemia (MW test; P < .005), acute lymphoblastic leukemia ( P < .05), or Hodgkin’s lymphoma ( P < .05) (Table 111).
TL-6 was detectable in all tested sera. Results obtained in healthy volunteers and patients with a monoclonal gammopathy are indicated in Figure 2 and Table 111. The normal S-IL-6 level was defined in healthy volunteers as being between 122 and 466 pgiml (mean 2 SD = 294 t 86 pg/ml; median = 287; range = 113-520 pg/ ml). S-IL-6 levels appeared to be significantly higher in IL-6 Serum Level in Relation to Other Serum the 3 groups of patients with MG than in healthy control Parameters in Multiple Myelomas The relations between S-IL-6 level and other serum subjects (Table 111: P < .05; MW-test). Table IV indicates that a S-1L-6 level higher than 466 pg/ml, was parameters which were previously demonstrated to be detected in 3 out of 66 (4.5%) normal subjects, in 2 out of useful in monitoring the course of multiple myeloma 20 (10%)patients with macroglobulinemia, in 4 out of 28 were studied. Considering the 21 patients with MM (14.2%) patients with MGUS, in 11 out of 55 (20%) whose serum was tested at diagnosis, S-IL-6 level was multiple myelomas, in 8 out of 24 (33.3%) patients with shown to correlate significantly with S-LDH (r = 0.736; non Hodgkin’s lymphoma, in 5 out of 8 (62.5%) patients P < .005). The correlations between S-IL-6 level and with Hodgkin’s lymphoma, and in 14 out of 18 (77.8%) either serum calcium (r = 0.392) or bone marrow plasma patients with acute leukemia. Low IL-6 levels (1 95 and cells percentage (r = 0.374) were not statistically signif198 pg/ml) were found in both patients with Castleman icant (P < .1). Mean S-IL-6 level was higher in the 12 disease. S-IL-6 levels higher than 700 pg/ml were de- stage I11 MM (474 t 169 pg/ml) than in the 9 stage tected only in patients with multiple myeloma, acute leu- 1 + I1 MM (346 t 114 pg/ml). Mean S-IL-6 level was kemia, or malignant lymphoma (Table IV). Among the also higher in the 1 1 patients with lytic bone lesions 55 patients with MM, the highest S-IL-6 level was found (458 t 187) than in patients with normal bone X-rays in the only tested plasma-cell leukemia (947 pg/ml) and (377 2 115). In the group of 55 patients with MM at diagnosis or in the highest value observed among the 22 patients with MM tested at diagnosis was 747 pg/ml. The highest evolution (Tables V, VI), S-IL-6 level significantly S-IL-6 levels detected in our study were observed in correlated with bone-marrow plasmacytosis (r = 0.446), patients with acute monoblastic leukemia ( I ,774 pg/ml), S-LDH (r = 0.392), S-P2m (r = 0.365), and serum cal-
Serum IL-6 Radioimmunoassay
.11m
-8271
.8M7
//
/
4
/
167
cium (r = 0.301). No significant correlations between S-IL-6 levels and either serum-creatinine or CRP or serum albumine or S-TK were found. Higher S-IL-6 levels were significantly associated with S432m 3 6 mg/L, S-LDH 3 250 UYL, serum calcium k 105 mg/L, CRP 3 10 mg/L, and bone-marrow plasmacytosis 350% ( P < .05). Considering haematologic parameters, S-IL-6 levels did not correlate with platelet numbers but it significantly and conversely did with haemoglobin levels (-0.304). The distribution of S-IL-6 levels is significantly higher in patients with haemoglobin G 100 g/L than in those with >lo0 g/L ( P < .Ol). S-IL-6 level did not correlate with age, sex, and immunoglobulin heavy or light chain
1000-
E
&ooo-
-
n
DISCUSSION E
2
I 400-
0:
t
i
,
2. *.
a 200
I
-
t
A
I : c
a
x:
m
mgus
mm
anll
all
111
nhl
cd
Fig. 2. Serum IL-6 levels (pglml), determined using a radioimmunoassay (see Materials and Methods), in 66 healthy volunteers (c), 102 patients with monoclonal gammopathy [monoclonal gammopathy of undetermined significance (mgus) = 28; macroglobulinemia (m) = 19; multiple myeloma (mm) = 551, and 52 control samples from patients with acute non-lymphoblasticleukemia (anll; n = 13), acute lymphoblastic leukemia (all; n = 5), Hodgkin’s lymphoma (hl; n = 8), non-Hodgkin’s lymphoma (nhl; n = 24) and Castleman’s disease (cd; n = 2).
IL-6 bioassays were developed using murine hybridoma and myeloma cell lines which were demonstrated not to respond to other well-known cytokines [13-151. This strategy did not take into account unknown or untested cytokines such as IL-11 which has been described recently [7]. Thus, the serum activity detected by using bioassays was the resulting activity of several biological activities including well-known and probably unknown cytokines, steroids, prostaglandins, and circulating autoantibodies or specific inhibitors. Specific neutralizing antibodies were shown to inhibit hybridoma growth factor activity in tested samples [6] but contaminating inflammatory cytokines have been detected in preparations of monoclonal antibodies [16]. Last, the amount of biologically active IL-6 and the total amount of IL-6 in the human serum could be different. We developed a sensitive and reproducible RIA using a polyclonal rabbit antihuman IL-6 antiserum to determine IL-6 level in human serum. This RIA allowed the detection of S-IL-6 at concentrations as low as 78 pg/ml with a within-range and between-range precision in good agreement with that of most published radioimmunoassays. Dilution tests indi-
TABLE 111. Serum Levels of IL-6 Measured Using the RIA in 66 Healthy Subjects, 102 Patients With Monoclonal Gammopathy and 50 Patients With Other Haematological Malignancies*
Control healthy subjects Multiple myeloma Macroglobulinemia MGUS Acute non lymphoblastic leukaemia Acute lymphoblastic leukaemia Hodgkin’s lymphoma Non-Hodgkin’s lymphoma
Number of cases
Serum IL-6 level, mean ik SD (pg/ml)
Median (pg/ml)
Range (pglml)
Statistical significance (Mann-Whitney), P
500
>600
>700
66 55 19 28 13 5 8 24
1 (1.5) 10 (18.3)** 1 (5.2) 4 (14.3)* 10 (76.9)*** 3 (60)*** 5 (62.5)*** 8 (33.3)***
6 (10.9)* 2 (7.2) 7 (53.8)*** 1(20) 3 (37.5)** 3 (12.5)*
3 (5.4) 3 (23.1)** 1(20) 2 (25)* 1 (4.1)
Control healthy subjects Multiple myeloma Macroglobulinemia MGUS Acute non-lymphoblastic leukaemia Acute lymphoblastic leukaemia Hodgkin's lymphoma Non-Hodgkin's lymphoma
-
+ MGUS
= monoclonal gammopathy of undetermined significance. *,**,***Statistical significance (Fisher's exact test): *P < .05; * * P < ,005; ***P < ,0005.
TABLE V. Correlations Between Serum Levels of IL-6 and Parameters of Disease Activity Determined Simultaneously in 55 Patients With Overt Multiple Myeloma Biochemical parameters Bone-marrow plasmocytosis Serum lactate dehydrogenase Serum 02 microglobulin Haemoglobin level Serum calcemia Serum creatinine C reactive protein Serum albumin Serum thymidine kinase Platelet count
*
C
0
, 0
.
I
2
.
I
4
.
1
6
.
I
8
'
1
10
Time (months)
Fig. 3. Longitudinal study of S-IL-6 level in 8 patients. A: Increasing S-IL-6 levels with unfavourable evolution of a patient with fulmigating multiple myeloma ( 0 ) and a patient with end-stage macroglobulinemia (0). B: Decreasing S-IL-6 levels of 3 patients undergoing treatment for multiple myeloma. C: Stable S-IL-6 levels in 3 patients with multiple myeloma in the plateau phase.
cated that IL-6 measured in serum seemed to be identical with the rhIL-6 used as the standard. In addition, the recovery studies showed that the effect of serum components on the RIA was negligible. Using this new RIA, IL-6 was detectable in all tested sera. S-IL-6 levels in healthy subjects averaged 294 t 86 pg/ml. These values are higher than those reported in
r
P
51 50 54 53 55 55 47 55 43 55
0.446 0.392 0.365 -0.304 0.301 0.198 0.176 -0.135 0.075 0.07
.25
studies using bioassays [ 17,181 or ELISA [ 19,201. Using an RIA, Schindler et al. [21] detected IL-6 plasma levels averaging 187 pg/ml in healthy subjects. These discrepancies between RIA and bioassays may be due to the effect of an inhibitory activity in serum or plasma samples. Alternatively or simultaneously, the discrepancies between data from different laboratories could be due to the absence of definite standardization that precludes any comparison [22]. The high S-IL-6 levels detected in patients with MM may be related to the dysregulated expression of IL-6 as a result of its involvement in MM oncogenesis since 1L-6 has been described as an autocrine [3] or paracrine [4] myeloma cell growth factor. High S-IL-6 levels were detected in 11/55 (20%) patients with overt MM. The highest level was found in the only tested plasma cell leukemia (947 pg/ml). These results are in agreement with those of Bataille et al. [6] who found high levels in 35% of overt MM and in all plasma cell leukemia patients. In MM patients, S-IL-6 levels were correlated with tumor cell mass parameters such as bone marrow plasmacytosis, S-LDH, and S+2m [23-261. IL-6 responsiveness of myeloma cells in vitro was demonstrated to be directly correlated with their proliferative status in vivo,
4001 1 200
Number of tested sera
Serum IL-6 Radioimmunoassay
169
TABLE VI. Relations Between Mean Serum IL-6 Levels and Other Parameters of Disease Activty in 55 Patients With Multiple Myeloma’
Parameters Bone-marrow plasmocytosis
Radioimmunoassay for the Measurement of Serum 11-6 and Its Correlation With Tumour Cell Mass Parameters in Multiple Myeloma Eric Soiary, Michel Guiguet, Veronique Zeller, Rene-Olivier Casasnovas, Denis Caillot, Pascal Chavanet, Henri Guy, and Gerard Mack Service d’Hematologie Clinique (E.S.,R.O.C.,D.C.,H.G.), Laboratoire de Biochimie (M.G.,V.Z.,G.M.), and Service des Maladies infectieuses (P.C.), C.H.U. Le Bocage, Dijon, France
Interleukin-6(IL-6) was demonstrated to be a strong autocrine or paracrineplasmocytoma cell growth factor in humans. Using a bioassay, high serum IL-6 (S-IL-6) levels were correlated with disease severity in plasma cell dyscrasias. Since other cytokines could interfere with the bioassays, we developed a specific radioimmunoassay to study S-IL6 levels in 102 patients with monoclonal gammopathy (MG). S-IL-6 level was studied by a double antibody radioimmunoassay using a rabbit polyclonal anti-lL-6 antibody and a human recombinant IL-6 as the standard. The lowest value of the standard significantly different from zero was found to be 78 pglml. Within-run and between-runprecisions were characterized by a mean coefficient of variation of 3.72 and 5.5%, respectively. The mean analytical recovery was found to be 113% and the immunochemical identity of 11-6 standard and S-IL-6 was shown by dilution tests. IL-6 was detected in all tested sera. Sera from 66 healthy volunteers and 43 patients with acute leukemia or malignant lymphoma were tested as controls. In healthy subjects, S-IL-6 values were 294 86 pglml. MG were classified as multiple myeloma(MM), macroglobulinemia, and MG of undetermined significance (MGUS). The distribution of S-IL-6 levels in patients with MG was significantly higher than in healthy subjects but lower than in patients with acute leukemia or Hodgkin’s lymphoma. Results obtained in 55 patients with MM were related to other biological parameters. S-IL-6 levels correlated with bone-marrow plasmacytosis (P < .0005), serum-lactate dehydrogenase (S-LDH; P < .005), serum p2 microglobulin (S 432111; P < .Ol), and serum calcium (S-Ca; P < .025) and inversely correlated with haemoglobin (P < .025). Our results indicate that 1) radioimmunoassay is suitable for the measurement of human IL-6 in serum; 2) high S-IL-6 levels are observed in a small number of patients with MG; and 3) S-IL-6 level correlates with tumour cell mass in patients with overt MM.
*
Key words: interleukin-6, monoclonal gammopathy, plasma cell dyscrasias
INTRODUCTION Interleukin-6 (IL-6) is a polypeptide cytokine which is produced by various types of lymphoid and non-lymphoid cells including monocytes, T and B cells, fibroblasts, keratinocytes, endothelial cells, mesangium cells, and various tumor cells. It has been demonstrated to have several biological activities in immune system, hematopoiesis, acute phase reactions, and neural system [l]. IL-6 was originally identified as a T cell-derived lymphokine that is not involved in the growth of normal activated B cells but induces final maturation of B cells into antibody producing cells. It is a potent growth factor for murine plasmocytomas and B cell hybridomas [2] and 0 1992 Wiley-Liss, Inc.
a strong autocrine or paracrine in vitro growth factor in human myelomas [3,4]. The in vitro responsiveness of myeloma cells to IL-6 was demonstrated to be directly correlated with their proliferative status in vivo [ 5 ] . IL-6 was shown to be overproduced in vitro in the bone marrow from patients with active multiple myeloma [4]. Consistent with this observation, serum levels of IL-6 determined in a bioassay using the B9 IL-6 dependent
Received for publication April 30, 1991; accepted August 15, 1991 Address reprint requests to Eric Solary, MD, HCmatologie Clinique, C.H.U. Le Bocage, 21034 Dijon Cedex, France.
164
Solary et al.
murine hybridoma were recently described as a good reflection of disease activity in plasma cell dyscrasias [6]. Thus, 1L-6 seems to be involved in vivo during the progressive phase of multiple myeloma. Nevertheless, since another cytokine named IL- 11 was recently described as a hybridoma growth factor [7J that could interfere in bioassays, we developed a specific radioimmunoassay (RIA) in order to detect IL-6 in serum from each healthy control subject tested. Using this sensitive RIA, we planned this study in order to investigate serum IL-6 level (S-IL-6 level) in patients with monoclonal gammopathies (MG) and evaluated the relationships between S-IL-6 level and other laboratory variables at diagnosis or during the course of multiple myeloma (MM). As increased S-IL-6 level was described in some acute leukemias [8] and malignant lymphomas [9], patients with acute leukemia or Hodgkin’s or non-Hodgkm’s lymphoma at diagnosis were tested as controls. PATIENTS AND METHODS Patients
Serum samples were obtained from three groups of subjects: group I consisted of 66 healthy volunteers; group 2 consisted of 102 patients with MG; and group 3 consisted of 52 patients with acute leukemia (n = 18), Hodgkin’s lymphoma (n = 8), non-Hodgkin’s lymphoma (n = 24), or Castleman’s disease (n = 2) at diagnosis. In all cases, blood samples were collected in 10 ml sterile silicone coated tubes with inert gel barrier and clot activator disc (Venoject VT- I09SAS, Terumo Medical Corporation, Leuven, Belgium). Tubes were centrifugated at 2,500 rpm for 10 minutes and serum was stored at -20°C until assays. Patients with overt inflammatory or infectious process were not included in the study. The I02 patients with MG underwent bone-marrow examination, serum protein electrophoresis, full skeletal radiographic surveys and were divided as follows: macroglobulinemia (n = 19), overt MM (n = 5 3 , and monoclonal gammopathies of undetermined significance (MGUS; n = 28). Fifty-nine samples were obtained at diagnosis before any treatment. Nine of the 19 patients with macroglobulinemia and 34 of the 55 patients with MM were previously treated at the time of the study. Macroglobulinemia was defined by the production of a monoclonal immunoglobulin M (serum level > 5 g/L) and unequivocal infiltration of well-differentiated lymphocytes and atypical or lymphocytic plasma cells in bone marrow biopsies. Among the 9 treated patients, 6 were in a plateau phase and 3 were in evolution. Overt MM was defined by previously described criteria [lo]. An inventory of the clinical and laboratory features of all patients at the time of the study was collected, including age, sex, immunoglobulin (Ig) heavy and light chain type, haemoglobin, platelet count, serum calcium,
serum albumin, serum creatinine, serum lactate dehydrogenase (S-LDH), C reactive protein (CRP), percentage of plasma cells in the bone marrow, and extent of bone lesions. According to the Durie and Salmon’s staging system [ 101, 25 patients had stage I or stage I1 MM and 30 had stage 111MM. Among treated multiple myelomas, 14 were in the plateau phase (partial or complete remission) and 19 had progressive disease or were relapsing. Among the patients, 69% had IgG, 20% had IgA, and 11% had pure Bence Jones MM. Kappa subtype was present in 56.4% and lambda in 43.6%. Sera of the longitudinal studies were always collected immediately before chemotherapy. Normal bone marrow examinations and skeletal surveys were found in 28 patients who fulfilled the criteria of monoclonal gammopathies of undetermined significance (MGUS) [ 1 11. MATERIALS AND METHODS Chemicals
IL-6 used in this study was the recombinant human IL-6 (rhIL-6) standard from the Intertest-6 kit (Genzyme, Boston). As indicated by the manufacturer, the mass of IL-6 contained in this standard was calibrated by comparison with an IL-6 reference standard supplied by the National Institute for Biological Standards and Control (Hertfordshire, UK). The specific activity of the material determined in an IL-6 dependent murine hybridoma cell proliferation assay is reported to be 5. lo6 Ulmg. Polyclonal rabbit anti-human IL-6 antiserum was purchased from Genzyme Inc. As indicated by the manufacturer, this antiserum does not cross-react with IL-I (a or p), TNF ( a or p), GM-CSF, and IFN (a or p). The specificity of this anti-IL-6 antibody for the relevant Jigand in the presence of IL-I (a or p) and TNF a was confirmed [ 121. ’251-rhIL-6(1,000-1,500 Ci/mmol) was obtained from Amersham laboratories (Les Ulis, France). Immunobead goat anti-rabbit immunoglobulin antibody and immunobead rabbit anti-mouse immunoglobulin antibody originated from Bio-Rad laboratories (Richmond, CA). Other chemicals were of reagent grade. RIA Procedures
The assay format chosen was a double antibody radioimmunoassay. The regular buffer (PB) used at all stages of the assay consisted of 0.05 M sodium phosphate, pH 7.5, containing 1 mM EDTA and 0.1% sodium azide. In order to minimize nonspecific protein adsorption, especially I2%rhIL-6, polystyrene test tubes were precoated with bovine serum albumine (B.S.A.; 20 g/L in PB) at 37°C for 2 hours and then washed with PB. The assay was performed in the presence of PB supplemented with B.S.A. 2 g/L. One hundred microlitres either of standards ranging from 78 to 5,000 pglml or of test samples
Serum IL-6 Radioimmunoassay
were incubated at room temperature for 6 hours with 100 pl of the anti-IL-6 serum dilution. This dilution lay between 1:25 and 150 according to anti-IL-6 titration curves performed with each different lot. The second incubation, after addition of 100 pl of '251-rhIL-6 (50 pg), was for 48 hours at 4°C. A third incubation, after adding 1.5 mg of immunobead goat anti-rabbit immunoglobulin antibody, took 2 hours at 37°C and then 2 ml of cold PB was added to each tube. Bound and free IL-6 were separated by centrifugation (3,500g for 30 min at 4"C), supernatants were decanted and the radioactivity of the pellets counted. Test samples and standards were measured in duplicate. Results were expressed as percent bound when compared with samples without IL-6 (B,) and plotted against a logarithmic scale of the IL-6 concentration. In order to point out the specificity of the RIA, we studied the effect of an IL-6 depleted serum on the displacement of Iz5I-rhIL-6 from the antibodies. One millilitre of a high-content sample serum (1,600 pg/ml) was incubated for 3 days at 4°C with an excess (80 pg) of anti-human IL-6 monoclonal antibody (BE-8; Dr Widjenes, C.T.S. Besanson, France). After a 2-hour incubation at 37°C with 1 mg of immunobead second antibody, the immune-precipitate was removed by centrifugation (3,50Og, 15 min at 4°C) and the IL-6 depleted supernatant was collected.
Serum p2 Microglobulin Serum p2 microglobulin (S-parn) levels were measured using the RIA kit purchased from Immunotech laboratories (F- 13288, Marseille, France). Corrections for renal function were not used in the calculation of the S-P2m levels.
165
1
*O 0
10
100
1000
10000
IL6 concentration (pg/rnl)
Fig. 1. Within-assay precision of the standard curve for IL-6 radioimmunoassay. For each standard concentration, the binding of '251-rhlL-6to the antibodies (6)is expressed as the percentage of the binding observed in the absence of IL-6 standard (60).Each point represents the mean of five individual values. The bars indicate the 95% confidence interval for the true mean.
.05 was considered to represent a significant difference or relationship. RESULTS IL-6 Radioimmunoassay
When IL-6 depleted serum was tested, no displacement of '251-rhIL-6 from the antibodies was observed, attesting to the specificity of this assay for IL-6. The within-run precision of the standard curve (Fig. 1) was characterized by a 95% confidence interval placed on the true mean for each standard concentration (n = 5). Serum Thymidine Kinase The coefficient of variation ranged from 0.89% for the Serum thymidine kinase (S-TK) levels were measured low standard values up to 7.73% for the high standard using the Prolifigen TK-REA radioenzymoassay values (mean was 3.72%). The lowest value of the stan(Sangtec, Pharmacia, France). The average S-TK level in dard significantly different from zero (a< 0.05) in this healthy subjects is estimated to be 2.4 units per L serum. typical experiment was found to be 78 pg/ml. Eight conS-TK values > 5 unitdl are referred to as being elesecutive standard curves were used to examine the "bevated. tween-assay'' precision. Within a given experiment, each standard was measured in duplicate and the mean value Other Parameters was considered as an individual IL-6 value for the statisBone-marrow plasmacytosis was determined on tical analysis. The coefficient of variation ranged from smears stained by May-Griinwald-Giemsa. Levels of al- 1.55% for the low standard values up to 11.52% for the bumin, C reactive protein, and immunoglobulins were high standard values (mean was 5.5%). Precision data for measured by nephelometry . All other biological parame- serum samples whose mean IL-6 content lay within the ters were determined by routine analysis. standard range of the assay are summarized in Table I. The recovery of the assay was performed as follows. A Statistical Analysis known quantity of rhIL-6 standard ( I ng) was added to a Comparisons between groups were done by the non serum whose mean IL-6 content was 315 pgiml (n = 5). parametric U test of Mann-Whitney (MW-test). Relation- The mean IL-6 content of the mixture was found to be ships between IL-6 and other biochemical parameters 1,496 pg/ml (n = 5). Thus, the mean recovery in this were done by correlation analysis. A P value of less than experiment was 113%. The immunochemical identity of
166
Solary et al. TABLE 1. Precision Data for Serum Samples Whose Mean IL-6 Content Lies Within the Standard Range of the Assay
Mean IL-6 content (pg/ml)
C.V.a (%)
315 803 1,496 1,687 455
12.8 3.0 2.2 9.6 5.7
Within run (n = 5 )
Between run (n = 7) ”.V.
= coefficient of variation.
TABLE II. Dilution Tests* ~
Dilution
Mean IL-6 content (pglml) (n = 5)
~
Theoretical value (pg/ml)
IL-6 content: theoretical value (%)
~
0 1:2
1:4 1:8
1,687 803 336 195
843.5 422 211
95 80 92
*The immunochemical identity of IL-6 standard and IL-6 measured in the serum was studied by diluting high content serum with RIA buffer.
IL-6 standard and IL-6 measured in serum was studied by diluting a high content serum with the assay buffer. Results of this experiment are summarized in Table 11. IL-6 Serum (S-IL-6) Levels
stage IV Hodgkin’s lymphoma (8,211 pg/ml), and high grade non-Hodgkin’s lymphoma (8,887 pg/ml). The distribution of S-IL-6 levels is not significantly different in the 55 patients with overt MM when compared with patients with non-Hodgkin’s lymphoma at diagnosis. This distribution is significantly lower in the 55 patients with overt MM than in patients with acute non lymphoblastic leukemia (MW test; P < .005), acute lymphoblastic leukemia ( P < .05), or Hodgkin’s lymphoma ( P < .05) (Table 111).
TL-6 was detectable in all tested sera. Results obtained in healthy volunteers and patients with a monoclonal gammopathy are indicated in Figure 2 and Table 111. The normal S-IL-6 level was defined in healthy volunteers as being between 122 and 466 pgiml (mean 2 SD = 294 t 86 pg/ml; median = 287; range = 113-520 pg/ ml). S-IL-6 levels appeared to be significantly higher in IL-6 Serum Level in Relation to Other Serum the 3 groups of patients with MG than in healthy control Parameters in Multiple Myelomas The relations between S-IL-6 level and other serum subjects (Table 111: P < .05; MW-test). Table IV indicates that a S-1L-6 level higher than 466 pg/ml, was parameters which were previously demonstrated to be detected in 3 out of 66 (4.5%) normal subjects, in 2 out of useful in monitoring the course of multiple myeloma 20 (10%)patients with macroglobulinemia, in 4 out of 28 were studied. Considering the 21 patients with MM (14.2%) patients with MGUS, in 11 out of 55 (20%) whose serum was tested at diagnosis, S-IL-6 level was multiple myelomas, in 8 out of 24 (33.3%) patients with shown to correlate significantly with S-LDH (r = 0.736; non Hodgkin’s lymphoma, in 5 out of 8 (62.5%) patients P < .005). The correlations between S-IL-6 level and with Hodgkin’s lymphoma, and in 14 out of 18 (77.8%) either serum calcium (r = 0.392) or bone marrow plasma patients with acute leukemia. Low IL-6 levels (1 95 and cells percentage (r = 0.374) were not statistically signif198 pg/ml) were found in both patients with Castleman icant (P < .1). Mean S-IL-6 level was higher in the 12 disease. S-IL-6 levels higher than 700 pg/ml were de- stage I11 MM (474 t 169 pg/ml) than in the 9 stage tected only in patients with multiple myeloma, acute leu- 1 + I1 MM (346 t 114 pg/ml). Mean S-IL-6 level was kemia, or malignant lymphoma (Table IV). Among the also higher in the 1 1 patients with lytic bone lesions 55 patients with MM, the highest S-IL-6 level was found (458 t 187) than in patients with normal bone X-rays in the only tested plasma-cell leukemia (947 pg/ml) and (377 2 115). In the group of 55 patients with MM at diagnosis or in the highest value observed among the 22 patients with MM tested at diagnosis was 747 pg/ml. The highest evolution (Tables V, VI), S-IL-6 level significantly S-IL-6 levels detected in our study were observed in correlated with bone-marrow plasmacytosis (r = 0.446), patients with acute monoblastic leukemia ( I ,774 pg/ml), S-LDH (r = 0.392), S-P2m (r = 0.365), and serum cal-
Serum IL-6 Radioimmunoassay
.11m
-8271
.8M7
//
/
4
/
167
cium (r = 0.301). No significant correlations between S-IL-6 levels and either serum-creatinine or CRP or serum albumine or S-TK were found. Higher S-IL-6 levels were significantly associated with S432m 3 6 mg/L, S-LDH 3 250 UYL, serum calcium k 105 mg/L, CRP 3 10 mg/L, and bone-marrow plasmacytosis 350% ( P < .05). Considering haematologic parameters, S-IL-6 levels did not correlate with platelet numbers but it significantly and conversely did with haemoglobin levels (-0.304). The distribution of S-IL-6 levels is significantly higher in patients with haemoglobin G 100 g/L than in those with >lo0 g/L ( P < .Ol). S-IL-6 level did not correlate with age, sex, and immunoglobulin heavy or light chain
1000-
E
&ooo-
-
n
DISCUSSION E
2
I 400-
0:
t
i
,
2. *.
a 200
I
-
t
A
I : c
a
x:
m
mgus
mm
anll
all
111
nhl
cd
Fig. 2. Serum IL-6 levels (pglml), determined using a radioimmunoassay (see Materials and Methods), in 66 healthy volunteers (c), 102 patients with monoclonal gammopathy [monoclonal gammopathy of undetermined significance (mgus) = 28; macroglobulinemia (m) = 19; multiple myeloma (mm) = 551, and 52 control samples from patients with acute non-lymphoblasticleukemia (anll; n = 13), acute lymphoblastic leukemia (all; n = 5), Hodgkin’s lymphoma (hl; n = 8), non-Hodgkin’s lymphoma (nhl; n = 24) and Castleman’s disease (cd; n = 2).
IL-6 bioassays were developed using murine hybridoma and myeloma cell lines which were demonstrated not to respond to other well-known cytokines [13-151. This strategy did not take into account unknown or untested cytokines such as IL-11 which has been described recently [7]. Thus, the serum activity detected by using bioassays was the resulting activity of several biological activities including well-known and probably unknown cytokines, steroids, prostaglandins, and circulating autoantibodies or specific inhibitors. Specific neutralizing antibodies were shown to inhibit hybridoma growth factor activity in tested samples [6] but contaminating inflammatory cytokines have been detected in preparations of monoclonal antibodies [16]. Last, the amount of biologically active IL-6 and the total amount of IL-6 in the human serum could be different. We developed a sensitive and reproducible RIA using a polyclonal rabbit antihuman IL-6 antiserum to determine IL-6 level in human serum. This RIA allowed the detection of S-IL-6 at concentrations as low as 78 pg/ml with a within-range and between-range precision in good agreement with that of most published radioimmunoassays. Dilution tests indi-
TABLE 111. Serum Levels of IL-6 Measured Using the RIA in 66 Healthy Subjects, 102 Patients With Monoclonal Gammopathy and 50 Patients With Other Haematological Malignancies*
Control healthy subjects Multiple myeloma Macroglobulinemia MGUS Acute non lymphoblastic leukaemia Acute lymphoblastic leukaemia Hodgkin’s lymphoma Non-Hodgkin’s lymphoma
Number of cases
Serum IL-6 level, mean ik SD (pg/ml)
Median (pg/ml)
Range (pglml)
Statistical significance (Mann-Whitney), P
500
>600
>700
66 55 19 28 13 5 8 24
1 (1.5) 10 (18.3)** 1 (5.2) 4 (14.3)* 10 (76.9)*** 3 (60)*** 5 (62.5)*** 8 (33.3)***
6 (10.9)* 2 (7.2) 7 (53.8)*** 1(20) 3 (37.5)** 3 (12.5)*
3 (5.4) 3 (23.1)** 1(20) 2 (25)* 1 (4.1)
Control healthy subjects Multiple myeloma Macroglobulinemia MGUS Acute non-lymphoblastic leukaemia Acute lymphoblastic leukaemia Hodgkin's lymphoma Non-Hodgkin's lymphoma
-
+ MGUS
= monoclonal gammopathy of undetermined significance. *,**,***Statistical significance (Fisher's exact test): *P < .05; * * P < ,005; ***P < ,0005.
TABLE V. Correlations Between Serum Levels of IL-6 and Parameters of Disease Activity Determined Simultaneously in 55 Patients With Overt Multiple Myeloma Biochemical parameters Bone-marrow plasmocytosis Serum lactate dehydrogenase Serum 02 microglobulin Haemoglobin level Serum calcemia Serum creatinine C reactive protein Serum albumin Serum thymidine kinase Platelet count
*
C
0
, 0
.
I
2
.
I
4
.
1
6
.
I
8
'
1
10
Time (months)
Fig. 3. Longitudinal study of S-IL-6 level in 8 patients. A: Increasing S-IL-6 levels with unfavourable evolution of a patient with fulmigating multiple myeloma ( 0 ) and a patient with end-stage macroglobulinemia (0). B: Decreasing S-IL-6 levels of 3 patients undergoing treatment for multiple myeloma. C: Stable S-IL-6 levels in 3 patients with multiple myeloma in the plateau phase.
cated that IL-6 measured in serum seemed to be identical with the rhIL-6 used as the standard. In addition, the recovery studies showed that the effect of serum components on the RIA was negligible. Using this new RIA, IL-6 was detectable in all tested sera. S-IL-6 levels in healthy subjects averaged 294 t 86 pg/ml. These values are higher than those reported in
r
P
51 50 54 53 55 55 47 55 43 55
0.446 0.392 0.365 -0.304 0.301 0.198 0.176 -0.135 0.075 0.07
.25
studies using bioassays [ 17,181 or ELISA [ 19,201. Using an RIA, Schindler et al. [21] detected IL-6 plasma levels averaging 187 pg/ml in healthy subjects. These discrepancies between RIA and bioassays may be due to the effect of an inhibitory activity in serum or plasma samples. Alternatively or simultaneously, the discrepancies between data from different laboratories could be due to the absence of definite standardization that precludes any comparison [22]. The high S-IL-6 levels detected in patients with MM may be related to the dysregulated expression of IL-6 as a result of its involvement in MM oncogenesis since 1L-6 has been described as an autocrine [3] or paracrine [4] myeloma cell growth factor. High S-IL-6 levels were detected in 11/55 (20%) patients with overt MM. The highest level was found in the only tested plasma cell leukemia (947 pg/ml). These results are in agreement with those of Bataille et al. [6] who found high levels in 35% of overt MM and in all plasma cell leukemia patients. In MM patients, S-IL-6 levels were correlated with tumor cell mass parameters such as bone marrow plasmacytosis, S-LDH, and S+2m [23-261. IL-6 responsiveness of myeloma cells in vitro was demonstrated to be directly correlated with their proliferative status in vivo,
4001 1 200
Number of tested sera
Serum IL-6 Radioimmunoassay
169
TABLE VI. Relations Between Mean Serum IL-6 Levels and Other Parameters of Disease Activty in 55 Patients With Multiple Myeloma’
Parameters Bone-marrow plasmocytosis
