Recently Du Villard et a1 (1995) have reported serum IL-6 measurements in monoclonal gammopathies. Elevated concentrations (> 3 3 5 pglml) were found in 45% of patients with non-Hodgkin's lymphoma (NHL). 35% of patients with multiple myeloma (MM), and 0% of patients with chronic lymphocytic leukaemias (CLL). The serum concentrations found in their controls (200 pg/ml) and patients were much higher than those reported in previous studies for controls (5 pg/ml) and MM patients (40 pg/ml) (Ludwig et al, 1991; Greco et al, 1992).Using our collection of serum samples, taken during the last 9 months and stored at -8O"C, serum IL-6 levels were measured with IL6 ELISA kit (TEBU, France) in 25 patients with NHL (mean age 62 f 4 years, sex ratio man/woman 1*5), 1 2 patients with Hodgkin's lymphoma (HL) (mean age 43 f 5, sex ratio 2) and six patients with CLL (mean age 70 f 2, sex ratio 1) and compared with 41 control subjects (mean age 66 f 1, sex ratio 5). As shown in Fig 1, serum IL-6 concentrations were significantly higher ( P < 0.001, Mann-Whitney U test) in patients with NHL (28.3 f 6.0pg/ml) and HL (26.0 * 3.5 pg/ml) than in patients with CLL (8.3 f 1*6pg/ml) and control subjects (6.7 f 1.1 pg/ml). The percentages of patients with IL-6 serum levels >2Opg/ ml (upper limit in our normal sera) were 40% (10/25) for NHL. 75% (9/12) for HL, and 0% (0/6) for CLL.We confirm the results of Du Villard et al(1995) showing higher levels of serum IL6 in 45% of their patients with NHL but not in CLL patients. However, as in several previous studies using HA for the dosage of serum IL6 (Ludwig et al. Greco et al. 1992), the concentrations found in our controls and patients are much lower than those found in their controls and patients. In a previous paper describing their RIA technique (Solary et al, 1992), the authors carefully tested the reliability of their assay, the immunochemical identity between the IL6 measured in serum and the recombinant IL6 standard, and eliminated nonspecific interferences with serum components. Moreover, their results were in agreement with those of Schindler et a1 (1 990) using RIA. They suggested that the discrepancies between their data and those from other laboratories could be due to the effect of inhibitory activities and/or to the absence of standardization. This last hypothesis seems improbable, because of the fact that most of these immunoassays use as standards purified recombinant human IL-6 of controlled structural and biological activity. The presence of soluble gp80 IL6 receptors in serum is now well established and the presence of IL6-IL6 soluble receptors in the circulation is demonstrated (Gaillard et 41, 1993). Due to the different types of antibodies used in these different immunoassays. these complexes could be detected or not. As the ratio between free and bound IL6 is unknown in physiological and pathological conditions, comparison of these techniques remains impossible at present. This problem is of some clinical importance if we consider that the balance between free and boun...
