THE impetus for the development of transfusion-sparing pharmaceuticals or techniques has been the risks associated with transfusion (infectious disease, transfusion-associated lung injury, transfusion reactions [hemolytic, anaphylactic], immunomodulation) and the need for augmenting oxygen delivery when compatible blood is not available or cannot be used. The blood supply of North America and Western Europe is now safer than ever; however, the need for blood and blood components at times is greater than their availability. 1
Acute normovolemic hemodilution (ANH), withdrawal of blood while maintaining isovolemia, was originally introduced to return coagulation factors to the patient after cardiopulmonary bypass. 2The concept that hemodilution might save erythrocytes per se appears to have been first envisioned by Messmer et al . 3The basis of this notion is that ANH reduces the circulating concentration of erythrocytes, thus, following ANH, the blood lost during surgery contains a lesser concentration of erythrocytes than if ANH had not been performed, resulting in a conservation of erythrocytes. ANH is one of the few procedures in medicine that lends itself to mathematical modeling. Analyses have shown that efficacy of hemodilution requires high initial hematocrits, 4,5low hematocrits after ANH (“target”), 4,5and a surgical blood loss that exceeds a specific minimum value 4,5and is also within a specific range. 5Recently, a more detailed analysis concluded that for ANH to be efficacious (conserve at least one unit of erythrocytes), surgical blood loss should be more than approximately 70% of the patient's blood volume. 6
Although the use of ANH to conserve erythrocytes and transfusion has been advocated by several organizations 7–9and transfusion experts, 10,11its use has remained controversial because of its uncertain efficacy and safety. The authors of a recent meta-analysis noted that their findings were inconclusive. 12The expected inconclusive nature of the totality of all publications rests on several important issues. Studies have failed to fulfill one or more of these critical criteria, all of which are required to provide a valid test of hemodilution: a relatively homogeneous population of patients so that blood loss is reasonably uniform; prospective random allocation of patients to groups with or without ANH; sufficiently high initial hematocrit; sufficiently low hematocrit (“target”) after ANH; withdrawal of a sufficient volume of blood; prospective transfusion criteria, uniformly and consistently applied; surgical blood loss that is within the range of potential efficacy; and a sample size sufficiently large to have a reasonable expectation of detecting a difference, should one exist. Indeed, of the publications analyzed, it appears that none satisfied these criteria.
In this issue of Anesthesiology, Matot et al. 13report the results of their study, which was uniquely designed to meet all criteria to provide an adequate test of the efficacy of ANH. They removed approximately 2 l of blood during ANH, reducing the patients’ hematocrit from 41% to 24%, and found the technique to be efficacious, significantly decreasing the fraction of patients requiring allogeneic transfusion (from 36% to 10%). Their findings are in accord with the prediction that more than 70% of a patient's blood volume must be lost for ANH to be efficacious, 6and that there is a range of blood loss, above and below which ANH will not result in avoidance of allogeneic transfusion. 5,6Also, as predicted by mathematical analysis, 6in the study conducted by Matot et al. 13, when blood loss was within the range of potential efficacy (70–90% of the estimated blood volume), ANH was exceedingly efficacious: none of the nine patients in whom ANH was performed required allogeneic blood, whereas all 10 patients in the control group did (P < 0.0001); and the groups had an equal incidence (100%) of allogeneic transfusion when blood loss exceeded 90% of the patient's estimated blood volume (numbers of patients not transfused supplied by personal communication with I. Matot, June 21, 2002). When surgical blood loss is exceedingly high, ANH can reduce the number of allogeneic units of blood transfused but not the fraction of patients requiring allogeneic transfusion. 5,6It is important to note that the efficacy of ANH depends not only on the amount of surgical blood loss but also on a sufficiently high initial hematocrit and the removal of a sufficient quantity of blood to achieve a sufficiently low “target” hematocrit. 4–6
The removal of such substantial quantities of blood and the reduction of the hemoglobin concentration to values sufficient to produce efficacy has the potential to threaten patient safety. Normovolemia must be maintained (to preserve oxygen delivery to and oxygenation of critical organs) not only during but also after performing ANH of a substantial fraction of the patient's blood, as the administered fluids leave the vascular space at different rates. 14,15Although the reduction of the hemoglobin concentration to a value as low as 5 g/dl does not produce systemic evidence of inadequate oxygenation in healthy, conscious humans, 16it does produce subtle, reversible deficits of cognitive function. 17,18Anesthetized patients can withstand hemodilution to hemoglobin concentrations at least as low as 8 g/dl (and likely substantially lower) without systemic evidence of inadequate oxygenation. 19,20It has been recommended that anesthetized healthy patients not be transfused until the hemoglobin concentration decreases to 7 g/dl 21or 6 g/dl. 8The safe limits of hemodilution for patients who cannot increase blood flow sufficiently to critical organs (e.g. , because of arterial stenosis, vasculitis, or impaired cardiac function) are not known. Hemodilution has not been associated with systemic or cardiac markers of inadequate oxygenation in patients undergoing coronary artery surgery. 22,23The coronary vasculature has a reserve dilatory capacity, which, in response to acute anemia, can increase blood flow by several fold. 24However, data from laboratory studies clearly demonstrate that at very low hemoglobin concentrations (below 3–5 g/dl), the myocardium becomes hypoxic, 25with decreased contractility, 26and that the hemoglobin value at which this occurs is higher when coronary artery blood flow is limited by a stenosis. 27
As other studies before it, the present study did not evaluate a sufficient number of patients to document the safety of ANH. For example, it would not have been expected to be able to detect an increase in myocardial infarction rate in this group of American Society of Anesthesiologists physical status I and II patients. Even in patients with or at high risk for coronary artery disease undergoing abdominal aortic surgery, the incidence of myocardial infarction is only approximately 4%. A study with an 80% likelihood of detecting a 25% increase of this incidence, to 5%, would require approximately 14,000 patients. Detection of a doubling of this rate, to 8%, would require a study population of approximately 1,200 patients.
Thanks to Matot et al. 13, we now have appropriate evidence that ANH, as predicted, can be efficacious if used correctly for the appropriate patient population. Unfortunately, its safety is unknown and must be weighed against the risks of the procedure and those of allogeneic transfusion. In addition to the potential risks of not maintaining normovolemia (hypovolemia or hypervolemia), theoretically, ANH could cause an increase in surgical blood loss because of increased blood flow (owing to increased cardiac output 16,28,29and lower blood viscosity 30); decreased concentrations of clotting factors 31(they are removed together with the erythrocytes); the effect of the fluid, such as hetastarch (on coagulation factors and platelet function 32), infused to replace the withdrawn blood; and perhaps because of altered margination or function of platelets. 33,34Although some investigators have reported increased surgical blood loss with ANH, Matot et al. 13did not. In addition, ANH requires expertise, takes time, and potentially could also divert the attention of the anesthesiologist from other patient care issues.
The potential for blood components to transmit viral disease is at an all-time low in Western Europe and North America, with the risk for transmission of human immunodeficiency virus (HIV), hepatitis C virus, and hepatitis B virus each being estimated as approximately 1 per 1,900,000, 1 per 1,600,000, and 1 per 180,000 units transfused, respectively. 35,36The risk of fatal hemolytic transfusion reactions (the majority occur in the operating room) exceeds that of transmission of HIV or hepatitis B virus. 37However, the incidence of transfusion-induced serious bacterial infection is at least as high as that of either HIV or hepatitis C virus, and it has been suggested that the incidence may be substantially underestimated. 38In some areas of the world, the incidence of transmission of parasitic disease is substantially higher than that of viral disease. 39New vectors of transmissible disease may, and likely will, appear. In addition to the infectious risk, the consequences of potential immunomodulation 40and the incidence of transfusion-associated lung injury 41are not clearly defined.
Thus, we are left with a not unusual clinical circumstance. We can quantify the efficacy of a technique or therapy, but its full risks are not clear. Is it worth the candle? 42It is better to light a single candle than to curse the darkness (attributed to Confucius): with the efficacy having been shown, perhaps someone will now address safety.