Early prediction of outcome in score-identified, postcardiac surgical patients at high risk for sepsis, using soluble tumor necrosis factor receptor-p55 concentrations

Objective: To investigate the prognostic value of increased serum concentrations of soluble tumor necrosis factor (TNF) receptors in patients at high risk for sepsis. Design: Prospective study. Setting: Cardiac surgical intensive care unit in a University Hospital. Patients: Those 27 of 870 consecutive postcardiac surgical patients who met a previously validated high-risk criterion for imminent sepsis (Acute Physiology and Chronic Health Evaluation II [APACHE II] score of greater than or equal to 24 on the first postoperative day [day 1]). In this population, systemic inflammatory response syndrome was present in 96% of the patients and the inhospital mortality rate was 30%. In addition, ten postcardiac surgical patients with an uncomplicated course (mortality rate 0%) were studied for comparison. Interventions: Blood sampling for measurements of serum concentrations of TNF and soluble TNF receptors 55 kilodalton (TNF receptor-p55) and 75 kilodalton (TNF receptor-p75) on days 1, 2, 3, and 5. Measurements and Main Results: Compared with the ten patients with an uncomplicated course (group A), the high risk patients had significantly higher baseline (day 1) serum concentrations of soluble TNF receptor-p55 (9.2 vs. 4.2 ng/ml) and soluble TNF receptor-p75 (9.2 vs. 5.5 ng/ml). These high-risk patients could be further differentiated into two subgroups: one (B) with a prompt decrease in APACHE II score and a good prognosis (mortality rate 0%) and another (C) with a persisting high risk of sepsis and mortality rate (40%, p <.05). Although baseline APACHE II score was similar in both high-risk subgroups, soluble TNF receptor p55 concentrations were significantly higher in subgroup C compared with subgroup B already at baseline (10.7 vs, 4.7 ng/mL). The receiver operating characteristic curve for subgroup classification by soluble TNF receptor-p55 was in a discriminating position with an area (0.773 +/- 0.096), confirming soluble TNF receptor-p55 as a predictor of mortality, TNF and soluble TNF receptor-p75 concentrations were less predictive at baseline. Conclusions: This study suggests that increased soluble TNF receptor-p55 concentrations in the serum of postcardiac surgical patients allow earlier prognostication of subsequent hospital course than APACHE II scores alone, This study further suggests that the combination of physiologic scores and cytokine receptor measurements could improve the predictive power of early postoperative risk stratification.