Crystalloids versus colloids during acute normovolemic anemia: the quest continues...

The optimal kind of fluid for fluid resuscitation during acute, severe hemorrhage is still discussed controversially. Of particular interest in this context is the choice of colloids versus crystalloids and their effect on the critical hemoglobin level. In a previous issue of Critical Care, Pape and colleagues describe the effect of four different volume replacement options on the critical hemoglobin concentration, and show marked differences for the different treatments. Even though some important pathophysiological issues remain unsolved, the current manuscript adds interesting evidence to an ongoing quest.

Pape and colleagues discuss the eff ect of diff erent volume replacements in a previous issue of Critical Care [1]. In light of the numerous risks associated with allogeneic blood transfusions, many eff orts have been made towards blood-sparing management of the perioperative phase [2]. By inducing normovolemic, dilutional anemia, the amount of blood transfusions given throughout surgery can be reduced; however, an adequate oxygen supply/ demand relationship and resulting adequate tissue oxygena tion are the limiting factors when applying this technique. When the hemoglobin concentration falls below the level where oxygen delivery does not meet the tissue oxygen demand anymore, a critical hemoglobin level (Hb crit ) has been reached and outcome is poor without intervention, that is, the administration of red blood cells [3]. Also during acute, accidental blood loss the Hb crit can rapidly ensue. Recommendations and guidelines for treatment of acute, accidental hemorrhage have changed over the last years and continue to evolve.
Pape and colleagues add evidence for the ongoing quest for optimal fl uid replacement strategies during acute hemor rhage and address the unresolved controversy of crystalloid versus colloid fl uid replacement strategies with the help of an established porcine hemorrhage model [1]. In this particular porcine model Pape and colleagues simultaneously withdrew blood and replaced the volume with 'plasma-expansion-rate-adjusted' amounts of Ringer's solution, gelatine, 6% hydroxyethyl starch (HES) 130/0.4 or 6% HES 450/0.7 and assessed the Hb crit by repetitively measuring oxygen uptake (VO 2 ) in conjunction with a computed regression analysis to detect deterioration indicating critical hemoglobin values. With this technique the authors could show that the choice of fl uid replacement impacts the onset and extent of the Hb crit . In conclusion, the authors propose tetrastarch preparations (for example, HES 130/0.4) as the 'most advantageous regarding mainte nance of tissue oxygenation during progressive anemia' [1]. However, this present study undoubtedly has some limitations that suggest a more careful approach regard ing the interpretation of its results.
Foremost, the induced blood loss was replaced with diff erent solutions at diff erent ratios, with the intention to account for the diff erent plasma expansion eff ects. However, the stroke volume variation, as indicator for volume responsiveness, decreased more in all but the Ringer's solution group, while intrathoracic blood volume (ITBV) increased in the colloid groups. Consequently, normovolemia seems to have been main tained best in the Ringer's solution group, while hyper volemic anemia was induced in the other three groups. Th e authors addressed this fact by discussing volume responsiveness and the possibility of having infused Ringer's solution at higher rates of 4:1 or 5:1. Yet, as normo volemic anemia was the goal, the colloidal infu sions could have been infused at a lower rate. As rheo logic eff ects and augmentation of regional blood fl ow during fl uid replacements play a major role in acute hemodilution, simply using a fi xed blood/fl uid exchange rate based on the theoretical plasma expansion eff ect to address individual fl uid demand may not have been the optimal approach. Using

Abstract
The optimal kind of fl uid for fl uid resuscitation during acute, severe hemorrhage is still discussed controversially. Of particular interest in this context is the choice of colloids versus crystalloids and their eff ect on the critical hemoglobin level. In a previous issue of Critical Care, Pape and colleagues describe the eff ect of four diff erent volume replacement options on the critical hemoglobin concentration, and show marked diff erences for the diff erent treatments. Even though some important pathophysiological issues remain unsolved, the current manuscript adds interesting evidence to an ongoing quest. a goal-directed fl uid therapy instead could have been more useful and addressed the actual fl uid demands of the individual animals during severe bleeding (for example, a diff erence in pulse pressure (dPP)-guided fl uid regimen) [4].
Furthermore, even though the critical hemoglobin concen tration is the absolute limiting factor of hemodilution, adverse eff ects could have probably been seen at earlier stages [5]. As the cardiac oxygen extraction cannot be augmented like in other tissues, oxygen delivery and cardiac index are the actual limiting factors of hemodilution. In mild hemodilution, coronary oxygen delivery is improved by a decreased plasma viscosity. Redistribution of blood fl ow from other organs to the heart accounts for the preservation of cardiac oxygenation during more pronounced hemodilution. During severe hemodilution, exhaustion of subendocardial fl ow reserves will cause local ischemia and is the main complication of further blood loss. As shown in the study by Pape and colleagues [1], cardiac index did signifi cantly increase in all groups until the Hb crit was reached in the Ringer's solution. Th ereafter, only small changes could be found. However, when looking at the oxygen delivery, an even decrease can be found in all groups, indicating no benefi cial eff ects of any of the solutions at the Hb crit of the Ringer's solution. Finally, the oxygen extraction ratio signi fi cantly increased from baseline in all but the Ringer's solution group at their respective Hb crit . Th erefore, as coronar blood fl ow and cardiac oxygenation were not assessed, the actual limits of hemodilution with the solutions used are still not entirely clear.
As the authors stated in their discussion, the animals were terminated after reaching Hb crit and only short-term eff ects were assessed. Th us, the displayed results do not account for local tissue hypoxia and other eff ects that would have resulted from redistribution of the infused fl uids into the extravascular space as these eff ects would have needed additional time to be detectable.
Some statistical issues might also be mentioned. Even though the authors assessed for normal distribution and thereafter used multiple testing, they did not correct their P-values or their results for this fact. Corrections were only performed within a parameter. Inference from the results could have been drawn more cautiously, considering the extensive multiple testing of all parameters.

Conclusion
Subsequent trials will be needed to pathophysiologically explain and validate the results of the present study. Eff ects on a microcirculatory level of the diff erent fl uids at a Hb crit need to be assessed in more detail and, as mentioned, implementation of a goal-directed regimen might be suggested. In summary, the authors are to be congratulated for adding evidence to the long-running debate on what type of fl uid is best to use in hemorrhage.