Endotoxin, an LPS found in the outer membrane of Gram-negative bacteria, has been considered by many to be the principal toxin involved in the pathogenesis of Gram-negative septic shock (1). However, it is now clear that endotoxin may cause most (ifnot all) of its biological effects via the release of host factors (2-5) . Several studies have suggested that TNF, a cytokine produced by macrophages during septic shock, is one ofthe endogenous mediators that causes cardiovascular injury and death (2-5).Previous studies examining the effects of endotoxin and TNF in animal models have used a number of techniques to study cardiovascular function during the first 1-12 h after challenge. Most of these investigations have shown evidence of myocardial depression (2-11) . However, recent studies of human and animal septic shock have demonstrated that cardiovascular function continues to change over a period of several days. Typically, the left ventricular ejection fraction (LVEF)' falls to a nadir 2-3 d after the onset of hypotension. This progressive decrease in LVEF is associated with IV dilatation and maintenance of normal or increased cardiac output. In survivors, these cardiovascular changes return to normal in 7-10 d (12-20).The present study shows that dogs given a single intravenous bolus of TNF produce all the same complex cardiovascular changes over 7-10 d as those seen in septic shock over the same period . Furthermore, endotoxin without any other bacterial products can also produce the same serial changes in cardiovascular function seen in human septic shock.

No new therapy for sepsis has shown clinical efficacy. Perhaps more accurate clinical and laboratory predictors are needed to identify patients who may benefit from a given treatment strategy. On the other hand, the therapeutic premises may be flawed. Targeting a single microbial toxin such as endotoxin may not represent a viable strategy for treating a complex inflammatory response to diverse gram-negative bacteria. Similarly, the strategy of inhibiting the host inflammatory response may not be beneficial because immune cells and cytokines play both pathogenic and protective roles. Finally, our scientific knowledge of the complex timing of mediator release and balance during sepsis may be insufficient to develop successful therapeutic interventions for this syndrome.

In TRICC and ARMA, randomization to fixed treatment regimens disrupted preexisting relationships between illness severity and therapy level. This created noncomparable subgroups in both study arms that received care different and opposite from titrated care, that is, practice misalignments. These subgroups reduced the interpretability and safety of each trial. Characterizing current practice, incorporating current practice controls, and using alternative trial designs to minimize practice misalignments should improve trial safety and interpretability.