Commentary

The colloid-crystalloid debate has lingered for decades, resulting in the overall conclusion that composition of fluids for resuscitation does not influence morbidity or mortality in the general intensive care unit (ICU) population and that the only difference involves cost [1]. Even with the advent of 'safer' hydroxyethyl starches(HESs) [2, 3], a mortality benefit remains elusive. However, human studies [4, 5] suggest that fluid therapies may not be as innocuous as once thought and that they may cause renal injury and perhaps affect mortality in specific subgroups.

The first adequately powered, randomized, blinded study drawing attention to these potential differential effects was the SAFE trial. This study found no differences in mortality in the general ICU population, but did find trends towards increased survival in patients with sepsis and increased mortality in patients with traumatic brain injury [6], suggesting that these differential effects do exist and that they may be determined by the population studied. More recently, the VISEP (Efficacy of Volume Substitution and Insulin Therapy in Severe Sepsis) trial [7] addressed the safety and efficacy of HES versus lactated ringer's solution in patients with severe sepsis and septic shock. These investigators demonstrated that HES increases risk of acute kidney injury (AKI) and renal replacement therapy. Similarly, a meta-analysis that included the VISEP trial showed an increased risk of AKI in the general population and an increased risk of AKI and use of renal replacement therapy in patients with sepsis. However, the VISEP trial used high doses of hyperoncotic HES and may not be relevant to usual practice. Finally, a meta-analysis by Perel and colleagues [8], which included both SAFE and VISEP trials, failed to show differences in mortality in hospitalized patients but recommended that future trials focus on specific subgroups. Taken together, these data suggested that fluid composition may be important, at least in certain subgroups of critically ill patients, especially in patients with sepsis. The Scandinavian 6S trial attempted toanswer this question by randomly assigning patients with severe sepsis to receive HES in a ringer's acetate solution compared with carrier solution alone. The 6S trial found a higher risk of 90-day mortality (relative risk = 1.17, P = 0.03) and greater use of renal replacement therapy with HES as compared with those receiving ringer's acetate [9]. However, whether the increased mortality and morbidity risk were present in a more heterogeneous ICU population was still unknown. Whether other forms of HES carry a similar risk is also unclear from the 6S trial. The Crystalloid versus Hydroxyethyl Starch Trial (CHEST) sought to answer these questions.

CHEST was a well-conducted, blinded, randomized trial, that used a patient-centered outcome such as 90 day mortality as primary aim, and that was adequately powered to find differences between groups using the intention-to-treat principle. The authors were cautious to ensure that the intervention fluids of the two arms of the study had the expected composition by performing independent and random biochemical analyses. In addition, the study targeted a more general ICU patient population, as compared to other recent clinical trials evaluating crystalloids versus colloids, such as 6S. Importantly, the CHEST was designed to allow for a difference in fluid volumes between each arm, whereas 6S proscribed equal volumes. Limitations to this study include the following: (a) predefined criteria for the initiation of renal replacement therapy were absent, (b) the observed death rate was lower than the predicted death rate, and this may lead to difficulties in detecting mortality differences, (c) the patients who were less sick (than 6S trial) and elective surgical patients were included, and (d) the time to resolution of the objective parameter (heart rate, blood pressure, respiratory variation of systolic or mean arterial blood pressure, central venous pressure, capillary refill, and urine output) used to support a diagnosis of hypovolemia was not compared between the groups.

The CHEST found no difference in 90-day mortality between patients receiving 6% HES (130/0.4) and those receiving 0.9% saline. Interestingly, the use of renal replacement therapy was greater in patients receiving HES, even though by RIFLE (Risk, Injury, Failure, Loss, and End-stage kidney disease) criteria the saline group had more AKI. Post hoc analysis suggested that increases increatinine were more pronounced in the HES group, perhaps prompting the small but significant increased use of renal replacement therapy. Shaw and Kellum [10] (2013) theorize that this paradox may be explained by a reduction in glomerular filtration rate by HES despite better early urine output due to more effective volume expansion with the colloid. Furthermore, clinicians may be trading short-term improvements in hemodynamics and urine flow for long-term renal toxicity [10]. Whether some patients would benefit from better resuscitation efficiency (that is, achieving resuscitation goals faster with less fluid) even at the expense of some renal toxicity cannot be addressed by this or the 6S trial. Finally, the CHEST enrolled patients an average of 11 hours after ICU admission, and most 6S patients were already resuscitated prior to study entry. Thus, these aspects of the colloid-crystalloid debate rage on. It will be important to see the one-year outcome follow-up data yet to be published. In the meantime, the existing data confirm a renal toxicity signal from HES not only in patients with sepsis but also in the general ICU population.

Recommendations

Given this evidence of renal toxicity and in spite of the uncertainty of the effect on resuscitation efficiency, we believe that HES should be avoided in patients with severe sepsis as well as in other critically ill patients at high risk of AKI. There is no doubt that the colloid-crystalloid debate has been informed by these two trials. However, the remaining uncertainty on aspects such as resuscitation efficiency and timing of intervention just might gather enough rumble for a thirteenth round.