Correlations between the systemic concentrations of 14 immune mediators in patients with recent onset type 1 diabetes yielded a web with central and peripheral elements. A central position was found for IL-1RA which was the only mediator associated with the pro-inflammatory core elements CRP+IL-6, with soluble adhesion molecules (E-selectin), and with several chemokines (MCP-4, MIP-1ß, MDC). The Th1 immunity-associated cytokine IFNγ and chemokine IP-10 did not associate with the network but remained separate.
In view of the many immune parameters analyzed, significant correlation coefficients may have occurred by chance in this descriptive analysis. A biological relevance of most r-values is supported by the observation of significant correlations of individual blood levels between pairs of immune mediators known to share regulatory control. All chemokines (MCP-4, MIP-1ß, IL-8, MDC, MCP-1, TARC) were found to be associated with the core chemokine MCP-4 or with MIP-1ß. The exception is IP-10, which is known to be induced by Th1 cytokine IFNγ 
. Indeed, systemic concentrations of IP-10 and IFNγ were correlated. The Th2 type chemokines MDC and TARC have been described to be regulated in parallel 
, and their concentrations correlated in the current analysis. IL-6 has been reported to regulate the expression of CRP 
, and this fits with the association of systemic concentrations of the two immune mediators observed here. Soluble adhesion molecules are often expressed in parallel 
, and this is reflected by the association between blood levels of sICAM-1 and E-selectin in our analysis.
In cases of moderate upregulation of circulating levels of CRP and IL-6, such as in subclinical systemic inflammation of type 2 diabetes, this is usually accompanied by increased concentrations of soluble adhesion molecules 
. The association found here for both CRP and IL-6 with the two soluble adhesion molecules sICAM-1 and E-selectin probably indicates similar co-regulation of these four immune mediators. However, the associations observed do not seem to be mediated by the obesity status of patients. After statistical adjustment for individual BMI five of the six associations between CRP, IL-6, sICAM-1 and E-selectin remained significant, only the association between IL-6 and E-selectin was lost. For many significant associations between immune mediators the observed correlation coefficients were below 0.5. This was also the case for pairs of immune mediators which are known to be co-regulated, such as IFNγ and IP-10 (r
0.33) or CRP and IL-6 (r
0.45). In persons with type 2 diabetes or impaired glucose tolerance, the correlation coefficient for CRP and IL-6 was reported to be in a similar range (r
. Additional internal and external factors appear to contribute to the regulation of systemic concentrations of immune mediators.
A remarkable finding is that all correlations found are positive, also between putative antagonistic pro-inflammatory and anti-inflammatory mediators (such as CRP versus IL-1RA). This is the characteristic of a highly buffered system where increased production of agonistic mediators is counterbalanced by increased levels of antagonistic mediators. Because of its central position, levels of IL-1RA may be particularly relevant for limiting aggressive/inflammatory immune reactivity. This is supported by the observation that parameters of beta cell function (fasting and meal stimulated serum C-peptide concentrations) associate only with IL-1RA and with none of the other 13 immune mediators analyzed, after adjustment for age, sex and BMI. The positive association of IL-1RA blood levels with beta cell function has been observed also in a cohort of children with type 1 diabetes, during the first year after diagnosis 
. In the latter study, a negative association of stimulated C-peptide concentrations was found during the first year after diagnosis with the pro-inflammatory Th1-associated chemokine MIP-1α (CCL3) 
. Since MIP-1α was not included in the panel of immune mediators measured in the context of the DIATOR study, its position in the immune mediator network remains to be determined.
The clinical relevance of the association of serum IL-1RA concentrations with beta cell function is uncertain. In rheumatoid arthritis the administration of recombinant IL-1RA is clinically effective in dampening local inflammation although clinical efficacy is less than seen for TNF blocking agents 
. Down regulation of systemic inflammation by treatment with IL-1RA was also observed in patients with long standing type 2 diabetes, and there was significant preservation of residual beta cell function, lasting for at least 9 months after the end of treatment, in the majority of patients 
. However, administration of the antagonist protein in patients with recent onset type 1 diabetes did not result in improved beta cell function or other clinical parameters, although the treatment protocol resembled that being effective in rheumatoid arthritis 
It is noteworthy that residual beta cell function in patients with recent onset type 1 diabetes is associated with systemic concentrations of an inhibitor of pro-inflammatory or aggressive immune reactivity. IL-1RA suppresses central pathways of inflammatory and destructive immunity via blockade of the IL-1 receptor 
. Hence, the extent of beta cell destruction may be determined by the ability of the organism to contain aggressive immunity. The ability to respond to beta cell destructive processes by up-regulating counter-reactive protective mechanisms may also be the reason behind resistance to type 2 diabetes in healthy obese people 
The association between beta cell function and circulating immune mediators appears to change with the progression of type 1 diabetes. There is substantial loss of beta cell function and probably of beta cell mass in the years following diagnosis 
, and a regression of the insulitis process 
. Islet inflammation was found to persists when there was still substantial beta cell mass present 
. Indeed, we found in longer term patients with type 1 diabetes a positive association of residual beta cell function with pro-inflammatory immune mediators IL-6 and TNFα, whereas increased concentrations of anti-inflammatory/regulatory immune mediators (Il-1RA, IL-10, transforming growth factor-ß1 and -ß2) were seen in patients with low serum C-peptide concentrations and assumed little residual beta cell mass 
We also searched for associations between the systemic immune mediator network and further metabolic or general patient characteristics. There were only a small number of significant associations except for BMI. Similar as reported from studies in obese persons, BMI correlated with CRP, IL-6, IL-1RA, MIP-1ß, MCP-4 
. HbA1c levels correlated with systemic concentrations of the two soluble adhesion molecules. Since both mediators are markers of endothelial dysfunction, increased concentrations may be expected with poor metabolic control 
. No association was found for insulin dose. Blood lipid parameters also showed no association with immune mediators, with one exception, IL-6. The negative association of IL-6 with HDL-cholesterol fits with the positive association of IL-6 with BMI 
. An association with sex was observed for IL-1RA, CRP and IL-6, with higher immune mediator levels in female patients. In a large case-chort and a population-based study similar results were obtained for IL-1 RA and IL-6, respectively, but not for CRP 
. Of the 14 immune mediators only MDC concentrations were (negatively) correlated with age. The limited age range of 18–39 years for patient recruitment probably did not allow to detect the increase of blood levels of pro-inflammatory immune mediators seen with advanced age 
. Graphic presentation of these findings () shows that peripheral elements of the network exhibit only few correlations with metabolic parameters, and none with general characteristics. Multiple associations are seen only for core elements of the network, IL-1RA, CRP and IL-6.
In conclusion, acute phase proteins, cytokines, chemokines and soluble adhesion molecules show significant correlations of their blood levels in patients with recent-onset type 1 diabetes. Among the core elements of the network IL-1RA has a dominant role, being associated with all groups of immune mediators. The network appears to be regulated, i.e., high levels of pro-inflammatory mediators are accompanied by high levels of antagonistic mediators. The central role of IL-1RA is supported by the finding that it is the only mediator showing association with residual beta cell function. There are only few significant correlations between concentration of immune mediators and other metabolic or general patient characteristics.