Results
Table 1 shows the baseline demographic, clinical, and laboratory characteristics of the case-cohort data set, classified by outcome status. Of the 4,197 participants, 332 (7%) had missing or unusable blood samples. Clinical details and the mean or median values of the inflammatory biomarkers relating to the remaining 3,865 individuals are shown in Table 1. The mean age of the study cohort was 66.9 years (SD 6.6), and 61% of the cohort were male. During 5 years of follow-up, 709 patients experienced a major macrovascular event, 439 patients experienced a microvascular complication, and 706 patients died (Table 1). (Note that patients could appear in more than one of the three columns.) Supplementary Table 1 shows these results for subjects without a history of microvascular or macrovascular disease at baseline (n = 2,270).
Supplementary Table 2 shows the associations among CRP, fibrinogen, and IL-6 levels, and baseline levels of macrovascular CVD and risk factors (as well as associations among CRP, fibrinogen, and IL-6 levels), after adjustment for age, sex, and randomized treatment allocations. Male patients had lower levels of fibrinogen and CRP than female patients, but higher levels of IL-6. Current smokers had higher levels of all three biomarkers. IL-6, but not CRP or fibrinogen, was associated with baseline macrovascular CVD. All three biomarkers were modestly associated with BMI (Spearman correlations [r]: fibrinogen r = 0.133; CRP r = 0.242; IL-6 r = 0.207) and moderately with each other (fibrinogen and CRP r = 0.414; fibrinogen and IL-6 r = 0.301; CRP and IL-6 r = 0.500). Associations with other continuous variables were weaker.
After adjustment for age, sex, and randomized treatment (model 1), fibrinogen and IL-6 levels were associated with all three study outcomes, whereas CRP level was associated with macrovascular events and death, but not with microvascular complications (Table 2). Further adjustment for key clinical risk factors (model 2) removed the effects of fibrinogen and IL-6 levels on microvascular complications and fibrinogen on macrovascular events; the remaining associations were attenuated, but still significant (Table 2). The risk of a macrovascular event was increased by 11% for every extra SD of log CRP, and by 37% for every extra SD of log IL-6; corresponding results for death were similar, at 15 and 35%. After additional adjustments for the other two inflammatory markers (model 3), only the associations of IL-6 with macrovascular events and death remained significant (Table 2). Results were similar in participants with and without baseline CVD (Fig. 1), except that the positive association between fibrinogen and death was only significant in the subgroup without overt disease at baseline. Results were also similar in participants with (n = 1,047) and without (n = 2,818) renal dysfunction at baseline (Supplementary Fig. 1). The numbers of individual macrovascular and microvascular outcomes in the whole study population are shown in Supplementary Table 3. Supplementary Table 4 shows that IL-6 level was significantly (P < 0.0001) associated with risks of myocardial infarction, stroke, cardiovascular death, and noncardiovascular death but not with risks of new or worsening nephropathy, retinopathy, or neuropathy.
Figure 1.
Comparison of associations of circulating inflammatory biomarkers with study outcomes in participants with and without previous CVD. The P values refer to tests of interaction by history of previous CVD.
Table 3 shows the discrimination and reclassification statistics for all three inflammatory markers and outcomes. Although the IDI and (with one exception) the continuous version of the NRI were significant in each case, the change in C-statistic and categorical NRI were only significant in the case of IL-6, which added significantly to both the discrimination and reclassification of both macrovascular events and death. Discrimination increased by 0.012 for macrovascular events and by 0.016 for death. The categorical NRI values were 0.03 and 0.06 for macrovascular events and death, respectively.
Diabetes. 2014;63(3):1115-1123. © 2014 American Diabetes Association, Inc.
