Fluids in septic shock: too much of a good thing?

In a recent issue of Critical Care, Brandt and colleagues report the effects of a 'liberal' fluid loading protocol compared to a more 'restrictive' protocol on hemodynamics and mortality in pigs in which septic shock had been induced. It appears that the former protocol was associated with higher mortality in spite of improved hemodynamics compared to the latter. The results of the paper are discussed here in view of the scope and mechanisms of these findings. With regard to fluid resuscitation, they indicate that too much of an otherwise good thing is harmful, even if overhydration and edema formation seem to have been prevented. They also do not exclude a specific toxic effect of the larger volumes of hydroxyethyl starch in the 'liberal' strategy. The precise nature of a toxic effect remains obscure, however, but may involve the kidneys.


Introduction
In a recent issue of Critical Care, Brandt and coworkers [1] performed fl uid resuscitation on pigs with endo toxemia or fecal peritonitis using either moderate volumes of crysta l loids (10 ml/kg/h) or larger volumes of crystalloids supple mented by 130/0.4 hydroxyethyl starch (combined 20 ml/kg/h). Th ese protocols were designed to mimic the relative 'restrictive' and 'liberal' fl uid resuscitation policies that have been previously compared in human major surgery and acute lung injury/acute respira tory distress syndrome [2][3][4]. In both sepsis models, mortality increased with 'liberal' fl uid loading protocols in spite of better haemo dynamic stabi liza tion. Th is interesting study raises a number of questions.

Volume and type of fl uid in the resuscitation from sepsis
Although supplemental hydroxyethyl starch use in both study arms was partially 'goal-directed' -on the basis of cardiac output responses assessed by esophageal Dopplerthe doses used for fl uid loading were relatively fi xed rather than completely based on cardiac fl uid responses (fl uid responsiveness). Th e latter is preferable at the bedside, even if we do not formally know whether such therapy causes less morbidity and mortality in septic shock than using fi xed volumes or guiding infusion according to relatively crude hemodynamics, such as the central venous pressure, as currently recommended [5]. It is likely, however, that tailored 'liberal' therapy decreases the risk for iatrogenic and detrimental fl uid overload compared to fi xed 'liberal' therapy [6][7][8]. Th e debate about fi xed 'restrictive' versus 'liberal' versus 'goaldirected' therapy in the case of major surgery is also unresolved [2,3,9]. Diff ering results among studies, which may relate to diff ering case mixes, defi nitions, hemodynamic monitoring techniques/endpoints and treatment strategies, preclude unequivocal conclusions [9].
Th e authors used diff erent types of fl uid in the 'restrictive' and 'liberal' arms, with hydroxyethyl starch used particularly in the latter. A toxic eff ect of hydroxyethyl starch can not thus be ruled out, so it is possible that the higher mortality in the 'liberal' arm was caused, in part, by toxicity rather than large volumes. Indeed, mortality in the control non-septic pigs receiving the 'liberal' protocol was 13% (1 out of 8). Toxicity may include renal damage, as was particularly noted from the histology of the 'liberal' endotoxin-challenged animals. In any case, the histology of several tissues suggested that overhydration and (pulmonary) edema had not increased in the 'liberal' compared to the 'restrictive' fl uid loading groups, even in the presence of so-called colloid plaques observed in lungs, for instance, although the nature of these remains relatively unclear. Finally, starch preparations may have multiple anti-infl ammatory eff ects, but we do not know whether this is good or bad during sepsis [10]. Collectively, the experiments reported raise the interest ing idea that too much of a good thing is detrimental, whether related to relative overtreatment or to toxicity of the hydroxyethyl starch colloid.
A comparison of these experimental results with the literature is diffi cult because of, for example, highly varying study goals and endpoints. Morisaki and colleagues [11] found that starches (more so than Ringers lactate)

Abstract
In a recent issue of Critical Care, Brandt and colleagues report the eff ects of a 'liberal' fl uid loading protocol compared to a more 'restrictive' protocol on hemodynamics and mortality in pigs in which septic shock had been induced. It appears that the former protocol was associated with higher mortality in spite of improved hemodynamics compared to the latter. The results of the paper are discussed here in view of the scope and mechanisms of these fi ndings. With regard to fl uid resuscitation, they indicate that too much of an otherwise good thing is harmful, even if overhydration and edema formation seem to have been prevented. They also do not exclude a specifi c toxic eff ect of the larger volumes of hydroxyethyl starch in the 'liberal' strategy. The precise nature of a toxic eff ect remains obscure, however, but may involve the kidneys. ameliorated progression of microvascular and parenchymal injury during the development of peritonitis in sheep. Su and colleagues [12] noted that starch, albumin, gelatin and Ringers lactate fl uid resuscitation aff orded similar survival benefi ts during protracted fecal peritonitis in sheep, in spite of greater hemodynamic eff ects with the fi rst two. Th is illustrates that the current data provided by Brandt and colleagues [1] may need to be confi rmed. Th e observations that hemodynamic and mortality endpoints may not go in the same direction also deserve further explanation.

Clinical implications
What are the clinical implications of these experimental results? Th e potential but unconfi rmed (renal) toxicity of hydroxyethyl starch is indeed a subject of ongoing research in human septic shock and the current experimental observations may further fuel these eff orts [13][14][15]. For instance, the potential renal toxicity of starch preparations may depend on volume, type, substitution of starch and the underlying condition of patients in whom fl uids are infused, so that general conclusions are hard to draw at this stage [13][14][15]. Th at colloids have greater hemodynamic eff ects, for a given fl uid infusion volume, than crystalloids, even in sepsis with increased permeability and potential leakage of the compounds, is corroborated by recent clinical observations [16].

Conclusion
Th e outcome benefi ts and drawbacks of fl uid resuscitation in sepsis and shock may not solely relate to hemodynamic eff ects, so that more is not always better, even if overt overhydration and (pulmonary) edema do not occur. Th e experimental fi ndings remind us that outcome may also be a matter of the type of fl uid used for initial resuscitation during septic shock. Obviously, this relates, among other factors, to the increasing evidence that starch solutions have important side eff ects, particularly when exceeding recommended maximum daily doses. Further comparative research is needed.