Volume 9 Supplement 1
Marker or mediator? Changes in endotoxin activity as a predictor of adverse outcomes in critical illness
© BioMed Central Ltd 2005
Published: 7 March 2005
Endotoxin is a potent trigger of the inflammatory cascade in critical illness. We have previously shown that endotoxin activity levels (EAA) on admission to the ICU predict adverse clinical outcomes.
To determine the significance of changes in EAA values over time in patients with critical illness.
Data were extracted from the Multi-centre Endotoxin Detection in Critical Illness trial database for patients who had a minimum of three EAA values (consecutive or nonconsecutive) during the first four ICU days beginning on ICU admission along with associated clinical outcome data. Change in EAA over time was defined qualitatively as an inflection in the curve generated by plotting EAA values over time using the assumption of a normal biologic polynomial distribution. Patients were grouped based on having either zero, one, or two inflections in EAA over the 4-day study period. Infection status was defined by the results of bacteriologic cultures and subsequent Clinical Evaluation Committee adjudication. The primary outcome measure was organ dysfunction over time as defined by the area under the curve generated by plotting MODS score against time (MODS burden) calculated over a 4-day period (trapezoidal rule). Secondary outcomes included ICU and hospital mortality.
In a total of 345 patients, 28 patients had zero inflections, 215 had one inflection, and 102 had two inflections. MODS burden was significantly elevated in patients with two inflections (MODS burden = 486 ± 232 MODS hours) as compared with those with zero (MODS burden = 252 ± 214 MODS hours) or one inflection (MODS burden = 386 ± 206 MODS hours) (P < 0.00001, P = 0.001, respectively; three-way analysis of variance P < 0.00001). EAA variability was predictive of organ dysfunction (MODS burden) over time (P < 0.00264). MODS burden correlated with both ICU mortality (P < 0.00001) and hospital mortality (P < 0.0001). EAA variability was found to be independent of infection status (P = 0.7915). The number of EAA inflections were more predictive of outcome than the absolute EAA value calculated as the total endotoxin activity load over the 4-day period (AUC, trapezoidal rule) (univariate regressions, P < 0.00001, P = 0.5440, respectively). In a separate study of 15 normal healthy volunteers, EAA variability was shown to contain zero inflections in all subjects studied.
These data suggest that increased variability in endotoxin activity is associated with worsening outcomes in the critically ill. Our assumption of a normal biologic polynomial distribution of endotoxin activity requires validation with more frequent EEA sampling. Further studies are warranted to assess whether the dynamic alteration of endotoxin activity may play an important role in the ongoing inflammatory response and organ dysfunction in the critically ill.