Anything Goes? High Time for Smart Blood Volume Monitors

In this issue of the ASAIO Journal, Keane et al.1 present the results of their exploratory study on the combined use of whole body bioimpedance (WBB) and relative blood volume (RBV) monitoring in stable hemodialysis (HD) patients. Although both techniques deal with fluid volume (status), they are quite different (Table 1). Whole body bioimpedance measurements with the Body Composition Monitor (BCM) used in this study are usually done predialysis to obtain an equilibrated and absolute value for global fluid overload.2 This information is helpful when setting the required (absolute) ultrafiltration volume on the dialysis machine. For many dialysis caretakers, bioimpedance has revolutionized their way to view fluid overload and target weight prescription,3,4 especially in light of the association between fluid overload and mortality.5–7Table 1.: Bioelectrical Body Composition Compared with Relative Blood Volume MonitoringRelative blood volume monitoring, on the other hand, records the (relative) volume dynamics of the actual fluid removal in one “organ,” the blood volume, during HD. Blood volume serves as the link between the extracorporeal device, where fluid is removed, and the microvasculature, where fluid is exchanged with the extravascular space. Relative blood volume at the beginning of HD is always 100% (by definition), independent of hypo- or hypervolemia, and it does not imply euvolemia. The degree of hypo- or hypovolemia remains unknown without additional measurements of absolute blood volume.8–10 The step from “relative” to “absolute” blood volume could make a major difference in volume management if the manufactures implemented such measurement into their dialysis machines. Relative blood volume typically declines during ultrafiltration and rebounds after the end of ultrafiltration, similar to HD solute kinetics during and after the end of extracorporeal clearance. The RBV response, however, is quite variable and among other factors such as body position, metabolic demands, and digestive activity depends on the rate of extracellular fluid shifting from the tissue to the intravascular compartment termed “vascular refilling.”11 From the hemodynamic point of view, the interstitial compartment can be seen as some extended (venous) capacitance, protecting against excessive changes in (central and arterial) blood volume. Everything else being equal it is therefore intuitive to assume that ultrafiltration-induced vascular refilling is larger with extracellular fluid overload.12,13 Unfortunately, and contrary to the success of solute kinetic analysis in HD, RBV kinetic analysis did not live up to expectations. As a consequence, some have reverted to a simple analysis of RBV slopes and patterns of RBV slopes as measures for volume overload or depletion (Figure 1).14–16Figure 1.: Schematic time course of relative blood volume (RBV, in %) as a function of ultrafiltration (UF), vascular refilling (Ref), time, and volume overload. A: flat-line with large volume overload; B: constant decrease with moderate volume overload; D: large initial decrease reaching a constant level suggesting an increasing refilling rate. Notice the relative sizes and shapes of refilling and ultrafiltration symbols.Ideally, information from RBV monitoring curves has been thought to be useful in setting target weights,14,17 but also to improve hemodynamic stability,18,19 as the fluid and electrolyte adjustments (only to name a few) which arise from HD and ultrafiltration are not without acute and chronic complications.20,21 The quantification of ultrafiltration-induced changes in blood volume through RBV monitoring has thus been expected to help in detecting and preventing the incidence of acute treatment-induced complications, but unfortunately, some previous expectations18,19 were not completely met.22–25 Keane et al.2 apparently were not interested in evaluating the effect of RBV monitoring on symptoms but they were interested in “improving our understanding of how RBV can inform fluid management,” which essentially refers to target weight determination. As a study prerequisite, the authors acknowledged that there was “greater evidence base underpinning BCM-based fluid management,” meaning that BCM-measurements were taken as the reference, and RBV slopes were compared with them. In their cohort of 47 patients, the relative change in blood volume per hour (ΔRBV/h) was larger when the specific ultrafiltration rate (ml/kg/h) increased. This finding is consistent with expectations because the response of RBV not only depends on volume overload but also on the magnitude of the perturbation (the ultrafiltration volume removed per unit time) and the extension of the perturbed compartment, i.e., the absolute blood volume, usually assumed as 7% of (euvolemic) body mass. The specific ultrafiltration rate accounts for effects of time and size, and this is one of the mechanistic reasons why an upper limit has been recommended to prevent intradialytic morbid events, which have been shown to increase with higher ultrafiltration rates.26 Keane et al.,1 however, observed no association between the RBV slope and either pre- or postdialysis fluid overload confirming recent results from absolute blood volume measurements.27 The present data therefore rightfully question the assumption that patients with flat RBV monitoring slopes are fluid overloaded, as patients with “A-slopes” (flat-liners) were the least fluid overloaded (Figure 1). However, failure to establish a relationship between volume overload and RBV changes could be because of simplifications, misunderstandings, as well as to experimental problems of both RBV and WBB measurements: For example, the RBV change was smaller in volume-depleted patients. Could this also be a problem of WBB measurements? For geometric reasons, WBB and volume status is dominated by impedance and the extracellular volume measured in the legs (Figure 2).28 Volume added to or removed from the trunk such as during peritoneal dialysis is almost invisible with WBB.29 However, trunk volume overload is more easily mobilizable and refilled into the blood compartment because of favorable hydrostatic pressure gradients, resulting in a “flat-line” RBV trajectory and a small RBV change at the end of HD. By contrast, volume sequestered in the legs is an important contribution to volume overload derived from WBB, but this volume is more difficult to mobilize because of orthostatic effects resulting in a larger RBV change in spite of volume expansion. Changes in regional fluid distribution during HD are the main reason why WBB as measured continuously and immediately postdialysis underestimates actual ultrafiltration volume.30Figure 2.: Difference of blood volume and whole body bioimpedance homunculi. A: Blood is distributed throughout the whole body, but when the body is rest in a thermoneutral environment and in a supine position, most of the blood is located in the trunk and thorax. B: The impact of fluid on total volume estimates using whole body bioimpedance analysis is highest in the extremities (excluding hands, feet, and the head) and lowest in the trunk and thorax.The comparison of equilibrated WBB to dynamic RBV values is in itself problematic. Relative blood volume recorded at the end of treatment is not equilibrated because refilling continues when ultrafiltration is terminated.12,31 It may take an hour and more until a new equilibrium between intra- and extravascular volumes is reached under stable conditions.12 Finally, there appears to be a major problem with so-called D curves of the current study, i.e., trajectories starting with a large RBV drop and flattening during later phases of the treatment (Figure 1). Lopot et al.17 already commented that this “response lacks a clear physiologic interpretation.” Some of these curves are implausible for several reasons: The reported RBV drop by 20% (corresponding to an RBV of 80%) within 10 minutes leading to a further reduction of 30% (RBV = 70%) within 1 hour, without any sign of hemodynamic complication, as reported by the authors is implausible. Moreover, a 20% drop caused by 80 ml of ultrafiltration volume (UFV) implies an absolute blood volume (Vb) of only 400 ml (Vb = (UFV − RV)/ΔRBV) in the absence of refilling (RV = 0) and even lower in the presence of refilling. It is just impossible that absolute blood volume including the volume in the extracorporeal circulation (about 100 ml) was only 300 ml. Could this be a measuring error? The baseline hematocrit (H0) in the curves was very low (as low as 18%), while routine laboratory hematocrit was within the normal range. If H0 is inadvertently low because of some unnoticed dilution, all subsequent RBV changes (ΔRBV = 1 − H0/H) are overestimated. For plausible RBV estimation, H0 must represent an undiluted, systemic, mixed venous, or arterial blood sample. At the beginning of treatment, this sample is diluted because of recirculating priming solution, and a plausible H0 is inferred from a stable reading. However, if blood flow is stopped or slowed at some point after connecting the patient to the extracorporeal circulation, a stable reading of diluted blood will be mistaken for a plausible systemic concentration H0. It is almost impossible to check for plausibility in this very crucial period without information on machine status. Could treatments have been done in the presence of some infusion? Did access recirculation develop during dialysis? The list of possible handling errors is long, some of which have been discussed elsewhere (Table 2).32 In any case, an unrealistically low baseline hematocrit is suspicious of an initial dilution effect, which might have led to the subsequent excessive drop in RBV, as observed in this study.Table 2.: Sources of Error in Relative Blood Volume Monitoring32Currently, there is no mechanism to prevent possible RBV monitoring errors and to conduct a thorough plausibility check using stand-alone RBV devices. While a stand-alone device provides flexibility, systems with measuring cells incorporated into blood lines and receiving information about the machine status such as blood flows, ultrafiltration rates, and recirculation rates either by means of thermodilution38,39 or by measuring effective on-line clearance,40 are likely to provide more reliable data and hence improved safety. Not only might unreliable data lead to dangerous situations and decisions but inconclusive data have the potential to discredit the whole technology. The question also arises to which degree such errors contributed to the negative outcome of previous clinical studies. The manufacturers are therefore urged to address the plausibility of RBV data by integrating useful treatment information from the machine, especially in times of smart devices abounding. If everything is relative, absolute information could provide a clarification. As of now, anything seems to go with RBV monitoring, which too frequently seems to provide inconclusive information.