NK cells and B cells showed no statistically significant change over the first month in the control group. In the sepsis group, the cytokine levels gradually declined (Table 1), IgG declined and IgM and IgA increased between the first and the third study periods (Table 1), MK-2206 in vitro but the lymphocyte subpopulations, CD3+, CD4+, CD8+, NK cells and B cells, remained unchanged. In the suspected infection group,

the cytokine levels and the NK cells and B cells declined, the CD3+, CD4+ and CD8+ subpopulations remained unchanged, while IgG declined and IgM increased. Table 3 shows the sensitivity, the specificity, the positive predictive value (PPV) and the negative predictive value (NPV) of the potential markers examined in predicting sepsis at the first study period.

CRP > 10 mg/l was shown to have relatively good specificity (0.78), but its sensitivity was low (0.64). IL-6 > 60 pg/ml was an excellent marker with high sensitivity and specificity, although the specificity dropped sharply at a lower cut-off value 30 pg/ml. TNF-α > 30 pg/ml was also a precise marker of sepsis, but at the cut-off Small molecule high throughput screening value of 15 pg/ml, its specificity dropped markedly. IL-Ib > 1 pg/ml was a specific but not sensitive index of infection. The area under the ROC (AUC) represents the probability that a randomly selected neonate with sepsis will have a higher test result than a randomly selected control subject. The area under the ROC for CRP, IL-6, TNF-α and IL-1b was 0.77, 0.96, 0.94 and 0.90, respectively. This implies that IL-6 was Isotretinoin the best and CRP the poorest among the four indices examined for predicting sepsis in the full-term infants in this study. The sepsis group was further divided into two subgroups based on the causative micro-organism (gram negative versus gram positive). No significant differences were found between these two subgroups in the serum levels of the immune parameters or in the diagnostic value of interleukins and CRP for predicting

sepsis. This study demonstrated an increase in CRP and in all three studied cytokines, IL-1b, IL-6 and TNF-α, in the group of neonates with sepsis, to values higher than those in neonates with suspected infection or healthy control subjects with no signs of infection. With regard their diagnostic accuracy to detect neonatal sepsis, CRP >10 mg/l was a specific but not sensitive index. This may be because of the fact that the sepsis workup was performed early, with the first suspicion of infection, at which time CRP has not reached its highest levels. Studies have shown that repeated CRP values may be a better diagnostic tool for neonatal infection [14–16], but introduction of antibiotics cannot always be postponed awaiting the results of serial CRP evaluation. A cut-off value of IL-6 > 60 pg/ml proved to be the more sensitive and specific index for early diagnosis of sepsis with both PPV and NPV of above 0.95.