SHORT REPORTS

to affinity and preferentially detects high-affinity antibodies. In our diabetic population, most IAA+ sera were identified by RBA, suggesting that these lAAs are of high affinity. As such, they might be the witness of a mature process directed against endogenous insulin. In each of the two groups of children, there was an ELISA+ serum containing low-affinity human insulin-specific lAAs, similar to the clonally-restricted antibodies found in the nondiabetic population at the frequency of 1% (10). In conclusion, RBA and ELISA yield discrepant results for IAA detection in. newly diagnosed type I diabetic children. In terms of IAA prevalence and age-related positivity, RBA definitely appears to be more appropriate than ELISA for IAA detection. From the Departments of Diabetology and Biochemistry, Robert Debrg Hospital, Paris; and the Regional Institute of Health, Tours, France; and the Departments of Pediatrics and Endocrinology, Sart-Tilman University, Liege, Belgium. Address correspondence and reprint requests to Claire Levy-Marchal, MD, Service de Diabe'tologie, Hdpital Robert Debr6, 48 Boulevard Se>urier, 75019 Paris, France. Received for publication 16 April 1990 and accepted in revised form 27 July 1990. ACKNOWLEDGMENTS

This work was supported by grants from the Fonds de la Recherche Scientifique Me"dicale (Brussels), Caisse Nationale d'Assurance Maladie-lnstitut National de la Sant£ et de la Recherche Me"dicale (998-197, Paris), and Novo Nordisk (Copenhagen). We are grateful to Drs. C. de Beaufort, J. Doutreix, V. Froment, and J. Voirin for dedicated and skillful assistance. REFERENCES

1. Palmer JP, Asplin CM, demons P, Lyen K, Tatpati O, Raghu PK, Paquette ZT: Insulin antibodies in insulindependent diabetes before insulin treatment. Science

222:1237-39, 1983 Wilkin TJ, Hoskins PJ, Armitage M, Rodier M, Casey C, Diaz AJ, Pyke DA, Leslie RDG: Value of insulin autoantibodies as serum markers for insulin-dependent diabetes mellitus. Lancet 2:480-82, 1985 Wilkin TJ, Palmer JP, Bonifacio E, Diaz JL: The second international workshop on the standardisation of insulin autoantibodies (IAA) measurements. Diabetologia 31: 449-50, 1988 Levy-Marchal C, Papoz L, de Beaufort C, Doutreix J, Froment V, Voirin J, Collignon A, Garros B, Schleret Y, Czernichow P: Incidence of juvenile type I diabetes in France. Diabetologia 33:465-69, 1990 Srikanta S, Ricker AT, McCulloch DK, Soeldner JS, Eisenbarth GS, Palmer JP: Autoimmunity to insulin, beta cell dysfunction, and development of insulin-dependent diabetes mellitus. Diabetes 35:139-42, 1986 Koch M, Sodoyez JC, Sodoyez-Goffaux F, Dozio N, Di Silvio LS, Kurtz AB: Is quantitative assessment of insulin antibodies and autoantibodies feasible? Diabetologia 32:774-78, 1989 Arslanian SA, Becker DJ, Rabin B, Atchison R, Eberhardt M, Cavender D, Dorman J, Drash AL: Correlates of insulin antibodies in newly diagnosed children with insulindependent diabetes before insulin therapy. Diabetes 34:926-30, 1985 Arquilla ER, Edwards S, McDougall BR, Mosqueda L, Stenger DP: Differences in humoral insulin-antibody response among inbred Lou/M rats and epitope presentation differences in ELISA and radioimmune titration. Diabetes 38:868-73, 1989 Sodoyez-Goffaux F, Koch M, Dozio N, Brandenburg D, Sodoyez JC: Advantages and pitfalls of radioimmune and enzyme linked immunosorbent assays of insulin antibodies. Diabetologia 31:694-702, 1988 10. Sodoyez JC, Sodoyez-Goffaux F, Koch M, Sonolag G, Bouillenne C, Franc,ois-Ge>ard C, Bosi E: Clonally restricted insulin autoantibodies in a cohort of 2200 healthy blood donors. Diabetologia. In press

Serum Levels of Tumor Necrosis Factor and IL-lcv and IL-lp in Diabetic Patients Objective: To determine whether chronic hyperglycemia causes increased levels of serum tumor necrosis factor (TNF) and interleukin 1a (IL-1a) and IL-10. Research Design and Methods: Sera were obtained from 59 diabetic patients, 44 chronically ill nondiabetic patients, and 34 age-matched healthy control subjects. Mononuclear cells were isolated from a subgroup of diabetic patients and healthy control subjects. Results: Except for a modest increase in the prevalence of detectable serum TNF levels in diabetic patients, the serum cytokines measured in this study did not appear to be altered in diabetes. In vitro TNF production by mononuclear cells was not altered in diabetic patients. However, in vitro IL-1(5 secretion, in response to lipopolysaccharides, was reduced. Conclusions: Diabetes mellitus is not associated with significant

Arshag D. Mooradian, MD Richard L. Reed, MD Keith E. Meredith, PhD Philip Scuderi, PhD

changes in serum levels of TNF, IL-1a, or IL-1JJ. In vitro secretion of IL-1(J in response to lipopolysaccharides may be reduced in diabetes. Diabetes Care 14:63-65, 1991

V

arious biological sequelae have been attributed to the advanced glycosylation end products (AGE) of various proteins (1,2). Of particular interest are the recent in vitro studies that have shown that AGE attached to proteins can specifically stimulate monocyte production of tumor necrosis factor (TNF) and interleukin 1p (IL-1 (3; 3). This observation

DIABETES CARE, VOL. 14, NO. 1, JANUARY 1991

63

©

SHORT REPORTS

suggests that diabetic patients with chronic uncontrolled hyperglycemia have elevated serum TNF or IL-1 levels with possible long-term deleterious effects on various metabolic parameters. To test this hypothesis, an enzyme-linked immunosorbent assay (ELISA) system specific for human TNF, IL-1 a, and IL-1 (3 was used to screen sera from diabetic patients. In vitro cytokine production by mononuclear cells in response to various stimuli was also evaluated in preliminary experiments.

RESEARCH DESIGN AND METHODS Fifty-nine diabetic men (mean age 68.1 ± 14.3 yr) were recruited from the outpatient clinics of Tucson Veterans Administration Medical Center. Thirty-nine patients were taking insulin, 11 were receiving oral hypoglycemic agents, and 9 were on weight-reducing diets. The mean fasting plasma glucose concentration was 9.8 ± 4.2 mM, and mean glycosylated hemoglobin was 10.7 ± 2.7% (normal range 4.0-6.8%). The mean duration of diabetes in this group was 14.1 ± 6 . 7 yr. The diabetic patients were compared with 44 nondiabetic patients with a mean age of 72.9 ± 7.0 yr. This group was chosen to control for a possible nonspecific effect of the stress of chronic illness other than diabetes on cytokine secretion. The most common diagnoses in this group were hypertension, congestive heart failure, and arthritis. Diuretics and antihypertensive and nonsteroidal anti-inflammatory agents were the most commonly used medications. A second control group consisted of 34 healthy men with a mean age of 69.1 ± 13.0 yr. The ELISA system specific for human TNF and IL-1 a has been described previously (4,5). The IL-1 (3 was measured with a commercially available ELISA for human IL-1 ^ (R and D Systems, Minneapolis, MN). The cell culture method has been described previously (5). The supernatant of the cell cultures was assayed for secreted TNF. The total cell lysate was assayed for IL-1 p. The number of monocytes in diabetic patients (595 ± 55) was not different from that in the control group (546 ± 48). The ACE-bovine serum albumin (BSA) was prepared by incubating BSA with 100 mM glucose for 30 days followed by dialysis against phosphate-buffered saline. Results are reported as means ± SD. One-way fixed effects of variance were used to examine differences in TNF and IL-1 levels among the three groups and proportions of patients with detectable serum levels of the cytokines. Tukey's honestly significant difference post hoc tests of pairwise differences were used if the F test for the main effect was significant.

RESULTS The prevalence of detectable serum TNF levels in diabetic patients, nondiabetic patients, and healthy control

subjects was 30.5, 29.6, and 11.8%, respectively. The mean serum TNF level in the diabetic group (5.86 ± 7.99 pM) was not found to be different from the level in nondiabetic patients (5.36 ± 8 . 1 8 pM) or healthy control subjects (5.76 ± 13.83 pM). Serum IL-1 a was detectable in 27.1% of diabetic patients, 27.3% of nondiabetic patients, and 17.7% of healthy control subjects. The mean serum IL-1 a level in diabetic patients (5.65 ± 9.67 pM) was not significantly different from the level in nondiabetic patients (7.19 ± 12.32 pM) or healthy control subjects (11.73 ± 29.90 pM). Serum IL-1 (3 was detectable in 1 of 18 control subjects and 1 of 16 diabetic patients tested. Within the diabetic group, there was no correlation between serum TNF or IL-1 a levels and either fasting plasma glucose or glycosylated hemoglobin levels. The serum TNF levels correlated positively with serum IL-1 a levels (Pearson's r = 0.49, P < 0.001). There was no statistically significant association between the serum cytokine levels and any diabetic complications such as nephropathy, retinopathy, or neuropathy (data not shown). The results of in vitro production of TNF or IL-l (3 by mononuclear cells from control subjects and diabetic patients are shown in Table 1.

CONCLUSIONS The serum levels of TNF, IL-1a, and IL-13 in nondiabetic patients recruited from the same outpatient clinics were similar to those of diabetic patients. Moreover, there was no correlation between either the plasma glucose levels or glycosylated hemoglobin level and the serum level of TNF or IL-1 a. These observations taken collectively indicate that chronic hyperglycemia in vivo is not associated with increased serum levels of these cytokines. Thus, as a clinical tool, the measurement of serum levels of TNF or IL-1 is not useful. The lack of a net AGE effect on TNF or IL-1 (3 secretion is in variance with a previously published study (3). One difference in these studies is that we used total mononuclear cells to study in vitro cytokine secretion, whereas the previous study used cultured monocytes. The choice of total mononuclear cells for this study was based on the assumption that cytokine secretion in vitro by monocytes in conjunction with other mononuclear cells probably simulates the in vivo situation better than the monocytes isolated from the rest of mononuclear cells. The possibility that both AGE protein and BSA were equally contaminated with endotoxin was excluded by the Limulus amebocyte lysate assay (