Secondary Logo

Share this article on:

Time to Reconsider the Role of Relative Blood Volume Monitoring for Fluid Management in Hemodialysis

Keane, David F.*,†,‡; Baxter, Paul; Lindley, Elizabeth*,†; Rhodes, Laura; Pavitt, Sue§

doi: 10.1097/MAT.0000000000000795
Renal/Extracorporeal Blood Treatment

Relative blood volume (RBV) monitoring during hemodialysis has been used to help guide fluid management for decades, although with little supporting evidence. The technique relies on the assumption that variation in RBV during fluid removal reflects the capacity for vascular refilling and that efficient refilling is related to fluid overload. This study investigated the relationship between RBV variation and bioimpedance-based fluid overload in 47 patients on stable hemodialysis. Mean treatment ultrafiltration volume (UFV) was 1.7 L and RBV reduction was 3.2%/hour. Relative blood volume slopes were grouped based on trajectory: flatline (no decrease), linear decrease, or linear decrease followed by flatline. Fluid overload was similar (p > 0.05) across groups pre-dialysis (1.0, 2.2, and 1.6 L, respectively) and post-dialysis (−0.8, −0.1, and −0.1 L), whereas UFV was higher in patients with a linear decrease (1.8, 2.5, and 1.6 L; p = 0.02). Specific ultrafiltration rate, but not fluid overload, was associated with RBV change over dialysis. At least half the patients in each group finished dialysis fluid depleted based on bioimpedance, suggesting that the link between refilling and fluid overload is not as straightforward as previously assumed. These results question the assumptions that the absence of an appreciable decrease in RBV indicates fluid overload, and a rapid fall suggests fluid depletion.

From the *Department of Renal Medicine, Leeds Teaching Hospitals NHS Trust, Leeds, United Kingdom

NIHR Devices for Dignity Healthcare Technology Co-Operative, Sheffield, United Kingdom

Leeds Institute for Cardiovascular and Metabolic Medicine, University of Leeds, Leeds, United Kingdom

§School of Dentistry, University of Leeds, Leeds, United Kingdom.

Submitted for consideration August 2017; accepted for publication in revised form February 2018.

Disclosure: The authors have no conflicts of interest to report.

This work was supported by the National Institute for Health Research Healthcare Technology Cooperative Devices for Dignity.

Correspondence: Dr David F. Keane, Department of Renal Medicine, 1st Floor Lincoln Wing, St. James’ Hospital, Leeds LS97TF, UK. Email:

Fluid management is one of the principal functions of hemodialysis. Clinical assessment of fluid status has been the basis of deciding how much fluid to remove during each treatment, but it is accepted that this approach is inadequate.1 A number of technologies have been proposed for objective assessment of fluid status, including measurement of relative blood volume (RBV). Relative blood volume devices measure changes in intravascular fluid status of the blood passing through the dialysis lines by monitoring the concentration of constituents of whole-blood, such as hemoglobin or hematocrit. These hemoconcentration markers can be measured by a number of techniques, including optical absorbance or transmission, the speed of ultrasound, or conductivity, but all effectively monitor relative changes in blood water concentration.2 These simple, noninvasive measurements can detect reductions in blood volume in real time, offering the potential for prevention of intradialytic hypotension (IDH) and improved fluid management. However, despite being used in hemodialysis for almost 30 years, there is still no robust evidence as to how the measurements can be used in practice.

There are two major assumptions underpinning the use of RBV in hemodialysis. First, that the hemoconcentration observed in blood passing through the dialysis lines reflects the relative change in the concentration of the whole blood volume. This is valid as long as both the amount of the marker being measured and the distribution of this marker are constant throughout the measurement session.3 Hemolysis, blood leaks, or blood transfusions during hemodialysis could affect the total amount of the marker, but these are not common. Because capillaries and the central circulation have different concentrations of red blood cells (known as the F-cell ratio), changes in capillary blood flow during a dialysis session could alter the distribution of the hemoconcentration marker.4

Second, for the technology to be used for fluid management, there must be a relationship between RBV changes and fluid status. A constant or increasing RBV is widely interpreted as a sign that the rate of refilling of the vascular space from the interstitium matches or exceeds the rate of fluid removal, indicative of interstitial fluid overload, whereas reductions in RBV suggest that vascular refilling cannot compensate for fluid removal, indicative of reduced interstitial fluid volumes. When these principles are applied to real-time monitoring of RBV, often as part of a feedback-controlled mechanism, reducing ultrafiltration rates in response to RBV reductions can reduce the number of hypotensive episodes.5 When applied to target weight management, RBV values that are relatively constant during a dialysis session are assumed to indicate that the patient is fluid overloaded, typically leading to reductions in target weight until there is an acceptable drop in RBV over a dialysis session.6 However, there are no clear definitions of the different trajectories, and good evidence in support of this approach is lacking. The only randomized controlled trial looking at the effect of RBV-based fluid management on hard outcomes actually showed increased mortality in the group managed with RBV.7

Bioimpedance is another simple, noninvasive technology that can provide information on fluid status. The Body Composition Monitor (BCM; Fresenius Medical Care, Bad Homburg, Germany) uses a model specifically designed for renal patients8 to give an estimate of fluid overload, which the device names “overhydration” (OH; we will use OH to specifically describe the parameter measured by BCM and “fluid overload” to describe the physiological state). Body Composition Monitor–measured OH has been well validated9 and shown to be directly associated with morbidity and mortality in patients on hemodialysis.10 Although experience and clinical judgment is required when interpreting BCM data, it is less subjective than the interpretation of changes in RBV.

Relative blood volume and BCM are measuring two distinct compartments—relative fluid volumes in the circulation and fluid in the tissue respectively—and they have the potential to be complementary techniques. This study aimed to exploit the greater evidence base underpinning BCM-based fluid management to try to improve our understanding of how RBV can inform fluid management.

Back to Top | Article Outline

Materials and Methods


This study is a subanalysis of a study analyzing the validity of alternative BCM measurement protocols11 and is exploratory in nature, so no formal sample size calculation was performed. A cohort of 47 patients on stable hemodialysis was recruited, being more than 18 years of age and having no apparent localized fluid accumulations. Hemodialysis prescriptions were for regimes of three sessions of 4 hours per week, dialysate temperature was 36°C, sodium was 137 mmol/L as standard, and patients were free to eat and drink as they desired.

Back to Top | Article Outline

Data Collection

Bioimpedance measurements with the BCM were made before dialysis according to manufacturer’s instructions. Measurements were checked visually for artifacts and repeated until the difference in BCM-measured OH was not greater than 0.2 L between readings. Post-dialysis OH was calculated as pre-dialysis OH minus the change in weight of the patient over the dialysis session.

Relative blood volume measurements were made using the Crit-Line III Monitor (Hema-Metrics, Kaysville, UT). The RBV results from each hemodialysis session were downloaded to allow analysis. Device calibration was checked monthly using a verification filter. Planned and achieved ultrafiltration volumes were recorded.

Back to Top | Article Outline

Data Analysis

Relative blood volume was defined using the percentage reduction in RBV normalized for time in hours (ΔRBV/hour). Each RBV slope was characterized based on the approach set out by Lopot et al.6 (Figure 1) using a value of the ΔRBV/hour slope cutoff (S cutoff) to distinguish slopes.

Figure 1

Figure 1

  1. Flatline: “A” slopes are characterized by a flat line throughout a hemodialysis session, with a maximum slope of S cutoff.
  2. Late reduction: “B” slopes are characterized by a flat slope over an initial period of the hemodialysis session (with a maximum slope of S cutoff for at least 1 hour) followed by a more rapid reduction in blood volume for the remainder of the session (with a minimum slope of S cutoff).
  3. Linear reduction: “C” slopes are characterized by a linear reduction in blood volume throughout the hemodialysis session (with a minimum slope of S cutoff).
  4. Early reduction: “D” slopes are the inverse of B slopes and are characterized by an initial rapid blood volume slope (with a minimum slope of S cutoff for at least 1 hour) followed by a flat slope for the rest of the session (a maximum slope of S cutoff).

Manufacturer’s guidance for distinguishing between slope groups A to D use a value for S cutoff of 3%, and this was also the basis of the fluid management strategies in the Crit-Line Intradialytic Monitoring Benefit (CLIMB) trial.7 Other ways of distinguishing groups have been used that vary quite significantly from the manufacturer’s proposal, notably the use by Sinha et al.12 of a conservative value of S cutoff of 1.5%.

In addition to using ΔRBV/h as the basis of categorizing each treatment session, associations between the value itself and other variables were explored.

Back to Top | Article Outline

Statistical Analysis

Pre- and post-dialysis OH, body mass index (BMI), ΔRBV/hour, and ultrafiltration parameters (ultrafiltration volume [UFV], UFV normalized by body weight—specific UFV—and ultrafiltration rate [UFR] normalized to body weight—specific UFR) were compared between the RBV groups by one-way analysis of variance. Because there are numerous criteria for categorizing RBV trajectories, sensitivity analysis was undertaken to reclassify all the data based on the a conservative definition of S cutoff as a maximum fall of 1.3% per hour.12

The relationship between ΔRBV/hour and pre- and post-dialysis OH and UFV were investigated using the Pearson correlation coefficient (r).

Back to Top | Article Outline


The characteristics of the included subjects can be seen in Table 1. Treatment sessions were all completed without any recorded symptoms or interventions. There was only one patient in the B group, so for analysis B and C slope groups were combined as representing blood volume dynamics following classic Guyton physiology.13

Table 1

Table 1

There was no difference in pre- or post-dialysis OH between the different slope groups, but UFV was higher in the B and C groups, both in the primary and sensitivity analyses (Table 2). There was no association between trajectory and BMI.

Table 2

Table 2

The plots of pre- and post-dialysis BCM-measured OH by RBV category confirm that there is no discernible difference in OH pattern between the groups (Figure 2). Across all categories, there were subjects who finish dialysis fluid depleted as measured by BCM.

Figure 2

Figure 2

There was no association between the rate of reduction of the RBV slope (ΔRBV/hour) and either pre- or post-dialysis OH, but the specific UFR was positively associated with ΔRBV/hour (r = 0.29; p = 0.045; Figure 3).

Figure 3

Figure 3

Back to Top | Article Outline


These results question some commonly held views on the association between changes in RBV during dialysis and fluid status. Although there is no gold-standard assessment of excess fluid, BCM assessments of fluid status have well-described measurement characteristics and reproducibility.14 Furthermore, patients on hemodialysis with BCM-measured OH that is too high or too low15 have reduced survival. Maduell et al.16 have previously demonstrated a relationship between RBV and BCM-measured OH. Here, we build on Maduell et al.’s work by evaluating the impact of common RBV-based fluid management strategies on fluid status measured by BCM.

A slopes (or flatlines) in response to ultrafiltration are generally assumed to suggest that fluid excess is driving vascular refilling and the maintenance of RBV. Based on this assumption, a reduction in target weight would normally be indicated. The data here suggest that subjects finishing dialysis up to 2 L fluid depleted, as measured by BCM, were classed as having an A-shaped curve. Even using conservative definitions of the curves, more than half the subjects with an A slope finished hemodialysis fluid depleted (Figure 2). Reducing target weight in individuals with A slopes would lead to excessive post-dialysis fluid depletion and risk of IDH. There are studies that have demonstrated benefit from reducing target weights in patients with a flatline,17 , 18 but the outcomes presented are limited, such as achievement of reduced weight, which, in itself, does not necessarily translate to better outcomes. It is possible that fluid management based on these principles is not dissimilar to probing for dry weights.

The A slopes in patents finishing dialysis fluid depleted could be explained by a nonconstant F-cell ratio. Hemodialysis is associated with fluid shifts from the microcirculation to the macrocirculation to maintain central blood volume.19 Mitra et al. showed that this increases RBV values, in their study by approximately 8%, and this effect could mask real reductions in absolute blood volume. Postural changes and eating have also been shown to affect the F-cell ratio.3

B and C slopes are thought to indicate that an individual is close to target weight, although clear guidance on how this is translated into practice is lacking. This could be based on defining individual RBV limits for a patient, below which patients have previously become symptomatic.20 Alternatively, the appearance of an acute reduction in the RBV trajectory21 has been suggested as indicating failing vascular refill and therefore proximity to target weight. However, with both these methods, there is significant uncertainty in these markers. The data from this study underlined this uncertainty. Although the rate of change in RBV is often used to adjust target weight, Figure 3 shows there was no significant association between ΔRBV/hour and pre- or post-dialysis OH. However, patients in groups B and C did have greater UFV and specific UFR than the other groups, and ΔRBV/hour was associated with specific UFR, suggesting that the rate of fluid removal has a greater influence on RBV than the interstitial fluid volume driving vascular refilling.

D-shaped curves have been associated with increased ultrafiltration volume,22 higher fluid overload,23 and treatments where patients became symptomatic.24 There are conflicting reports about how common these measurements are, from less than 1% in an adult unit25 to 91% in a pediatric unit.26 The data here demonstrate clearly that these changes cannot always be physiological. Figure 4 shows the RBV data from four subjects in this study with D-shaped curves. Despite the apparent large reduction in RBV, the treatments were completed without complications and stable blood pressures. In each of these cases, the reduction in RBV over the initial hour would suggest much greater fluid loss from the circulation than was removed by ultrafiltration, using estimated absolute blood volume from anthropometry.27 It is notable that in all these cases, the baseline hematocrit measured by the Crit-line was very low. Although we did not have reference blood samples from the same session for hematocrit comparison, routine, laboratory monthly blood data from sessions preceding and following the sessions monitored by Crit-line show that all these patients’ hematocrit was stably in the normal range, suggesting a measurement artifact. An artifactually low initial hematocrit would correspond with overestimation of reductions in RBV, as observed here.

Figure 4

Figure 4

A common feature of all of the groups is high variation. Plasma-refilling coefficients vary markedly between individual patients undergoing hemodialysis,22 , 28 with a removal of 2 L of fluid over 1 hour giving anywhere between 0.7% and 21.9% reduction in RBV.29 It then follows that critical RBV limits will also vary between subjects.30 Removing the intersubject variability by defining individual critical RBV limits improves the reliability, and this has, in one study, been shown to predict IDH events with a variation of less than 5% RBV,31 but there still remains significant intraindividual variation in the RBV response to ultrafiltration.32 Concurrent use of BCM and RBV allows the opportunity to account for some of the variability between different measurement sessions, and absolute blood volume (ABV) measurements could further explain intrasubject variation.

This uncertainty may, in part, explain the lack of good interventional studies supporting the use of RBV-based fluid management. The one published randomized controlled trial that used mortality as an outcome, the CLIMB study, actually showed a negative result.7 Noninterventional studies have been more promising. Sinha et al.12 analyzed patients enrolled in the dry-weight reduction in hypertensive hemodialysis patients (DRIP) trial, investigating probing for dry weight. They reported a number of observations in support of the use of RBV slope for assessment of dry weight, including the fact that RBV slopes steepen upon dry weight probing and that baseline RBV slope is associated with weight loss and reduction in blood pressure after probing. These findings support the physiological basis of RBV monitoring but provide no evidence for the benefit of using the technology to guide decision making. It is also worth pointing out that dry weight determined from probing can be much lower than the weight at normal fluid status.

Relative blood volume measurements can also be used to automatically adjust the ultrafiltration rate using a feedback loop to attempt to avoid IDH. Results from randomized trials using this approach have been mixed. Relative blood volume–guided ultrafiltration in hypotensive-prone patients has previously been shown to reduce the number of treatments with IDH,33 whereas more recently Leung et al.34 demonstrated no benefit from feedback control technology. When Antlanger et al.35 enrolled fluid-overloaded patients and applied a rapid dry weight reduction protocol, they observed a lower rate of complications in the study group treated by ultrafiltration regulation and temperature regulation combined than in the conventional dialysis group, although ultrafiltration regulation and conductivity regulation was worst.

Measurement of ABV36 offers the possibility of removing some of the inherent uncertainty in the use of RBV measurements. Absolute blood volume measurements have been used to demonstrate that vascular refilling is dependent on UFV but not fluid overload, consistent with the results presented here,37 and a small pilot study has demonstrated that use of an ABV threshold to guide target weight management reduces IDH.38

Our study was exploratory in nature and was not powered for outcomes; the relatively small numbers of participants in each group is a weakness. However, at an individual level, our data show that the commonly held view assumption that a low ΔRBV/hour (an A slope) is not a reliable indicator of fluid overload. There is a real need for further studies, using well-defined approaches to RBV-based fluid management, to evaluate the impact on outcomes.

Back to Top | Article Outline


These data call into question the assumption that patients with a flatline RBV are fluid overloaded and require a reduction in target weight. Current use of RBV for fluid management could be leaving patients at risk of complications associated with low BCM-measured OH.15 , 39 There is a need for further observational studies that use objective and reproducible classifications in the management of RBV. The complementary nature of BCM, ABV, and RBV supports further studies into how the information from all tests can be combined.

Back to Top | Article Outline


D.K. is funded by a Healthcare Science Research Fellowship from the National Institute for Health Research (HCS-D12-06). The views expressed in this publication are those of the authors and not necessarily those of the National Health Service (NHS), the National Institute for Health Research, or the Department of Health.

Back to Top | Article Outline


1. Covic A, Onofriescu M. Time to improve fluid management in hemodialysis: Should we abandon clinical assessment and routinely use bioimpedance? Clin J Am Soc Nephrol 2013.8: 1474–1475.
2. Schneditz D, Pogglitsch H, Horina J, Binswanger U. A blood protein monitor for the continuous measurement of blood volume changes during hemodialysis. Kidney Int 1990.38: 342–346.
3. Dasselaar J, Huisman R, De Jong P, Franssen C. Measurement of relative blood volume changes during haemodialysis: Merits and limitations. Nephrol Dial Transplant 2005.20: 2043–2049.
4. Schneditz D, Ribitsch W, Schilcher G, Uhlmann M, Chait Y, Stadlbauer V. Concordance of absolute and relative plasma volume changes and stability of cells in routine hemodialysis. Hemodial Int 2016.20: 120–128.
5. Santoto A, Mancini E, Basile C, et al. Blood volume controlled hemodialysis in hypotension-prone patients: A randomised, multicentre controlled trial. Kidney Int 2002.62: 1034–1045.
6. Lopot F, Nyiomnaitham V, Svárová, Polakovic V, Svára F, Sulková S. Continuous blood volume monitoring and “dry weight” assessment. J Ren Care 2007.33: 52–58.
7. Reddan D, Szczech LA, Hasselblad V, et al. Intradialytic blood volume monitoring in ambulatory hemodialysis patients: A randomized trial. J Am Soc Nephrol 2005.16: 2162–2169.
8. Chamney PW, Wabel P, Moissl UM, et al. A whole-body model to distinguish excess fluid from the hydration of major body tissues. Am J Clin Nutr 2007.85: 80–89.
9. Wabel P, Chamney P, Moissl U, Jirka T. Importance of whole-body bioimpedance spectroscopy for the management of fluid balance. Blood Purif 2009.27: 75–80.
10. Wizemann V, Wabel P, Chamney P, et al. The mortality risk of overhydration in haemodialysis patients. Nephrol Dial Transplant 2009.24: 1574–1579.
11. Keane DF, Baxter P, Lindley E, et al. The body composition monitor: A flexible tool for routine fluid management across the haemodialysis population. Biomed Phys Eng Express. 2017.25: 3.
12. Sinha A, Light R, Agarwal R. Relative plasma volume monitoring during hemodialysis AIDS the assessment of dry weight. Hypertension 2010.55: 305–311.
13. Guyton A, Hall J. Textbook of Medical Physiology. 2006.Philadelphia, PA, Elsevier.
14. Wabel P, Chamney P, Moissl U, et al. Reproducibility of bioimpedance spectroscopy (BIS) in health and disease (abstract). Nephrol Dial Transplant 2007.22(suppl 6): 137.
15. Dekker MJ, Marcelli D, Canaud BJ, et al; MONDO Initiative: Impact of fluid status and inflammation and their interaction on survival: A study in an international hemodialysis patient cohort. Kidney Int 2017.91: 1214–1223.
16. Maduell F, Arias M, Massó E, et al. Sensitivity of blood volume monitoring for fluid status assessment in hemodialysis patients. Blood Purif. 2013.35: 202–208.
17. Steuer RR, Germain MJ, Leypoldt JK, Cheung AK. Enhanced fluid removal guided by blood volume monitoring during chronic hemodialysis. Artif Organs 1998.22: 627–632.
18. Rodriguez H, Domenici R, Diroll A, Goykhman I. Assessment of dry weight by monitoring changes in blood volume during hemodialysis using Crit-Line. Kidney Int 2005.68: 854–861.
19. Mitra S, Chamney P, Greenwood R, Farrington K. The relationship between systemic and whole-body hematocrit is not constant during ultrafiltration on hemodialysis. J Am Soc Nephrol 2004.15: 463–469.
20. Steuer RR, Leypoldt JK, Cheung AK, Senekjian HO, Conis JM. Reducing symptoms during hemodialysis by continuously monitoring the hematocrit. Am J Kidney Dis 1996.27: 525–532.
21. Mitra S, Chamney P, Greenwood R, Farrington K. Linear decay of relative blood volume during ultrafiltration predicts hemodynamic instability. Am J Kidney Dis 2002.40: 556–565.
22. Santoro A, Mancini E, Zucchelli P. Ultrafiltration behaviour with different dialysis schedules. Nephrol Dial Transplant 1998.13(suppl 6): 55–61.
23. Bonello M, House AA, Cruz D, et al. Integration of blood volume, blood pressure, heart rate and bioimpedance monitoring for the achievement of optimal dry body weight during chronic hemodialysis. Artif Organs 2007.30: 1098–1108.
24. Hothi D, Harvey E, Goia C, Geary D. Blood-volume monitoring in paediatric haemodialysis. Pediatr Nephrol 2008.23: 813–820.
25. Lopot F, Nejedlý B, Sulková S. Continuous blood volume monitoring and ultrafiltration control. Hemodial Int 2000.4: 8–14.
26. Dheu C, Terzic J, Menouer S, Fischbach M. Importance of the curve shape for interpretation of blood volume monitor changes during haemodiafiltration. Pediatr Nephrol 2009.24: 1419–1423.
27. Nadler S, Hidalgo J, Bloch T. Prediction of blood volume in normal human adults. Surgery 1962.51: 224–232.
28. Iimura O, Tabei K, Nagashima H, Asano Y. A study on regulating factors of plasma refilling during hemodialysis. Nephron 1996.74: 19–25.
29. Koomans HA, Geers AB, Mees EJ. Plasma volume recovery after ultrafiltration in patients with chronic renal failure. Kidney Int 1984.26: 848–854.
30. Andrulli S, Colzani S, Mascia F, et al. The role of blood volume reduction in the genesis of intradialytic hypotension. Am J Kidney Dis 2002.40: 1244–1254.
31. Barth C, Boer W, Garzoni. Characteristics of hypotension-prone haemodialysis patients: Is there a critical relative blood volume? Nephrol Dial Transplant 2003.18: 1353–1360.
32. Krepel HP, Nette RW, Akçahüseyin E, Weimar W, Zietse R. Variability of relative blood volume during haemodialysis. Nephrol Dial Transplant 2000.15: 673–679.
33. Gabrielli D, Krystal B, Katzarski K, et al. Improved intradialytic stability during haemodialysis with blood volume-controlled ultrafiltration. J Nephrol 2009.22: 232–240.
34. Leung KCW, Quinn RR, Ravani P, Duff H, MacRae JM. Randomized crossover trial of blood volume monitoring-guided ultrafiltration biofeedback to reduce intradialytic hypotensive episodes with hemodialysis. Clin J Am Soc Nephrol 2017.12: 1831–1840.
35. Antlanger M, Josten P, Kammer M, et al. Blood volume-monitored regulation of ultrafiltration to decrease the dry weight in fluid-overloaded hemodialysis patients: A randomized controlled trial. BMC Nephrol. 2017.18: 238.
36. Schneditz D, Schilcher G, Ribitsch W, Krisper P, Haditsch B, Kron J. On-line dialysate infusion to estimate absolute blood volume in dialysis patients. ASAIO J. 2014.60: 436–442.
37. Kron S, Schneditz D, Leimbach T, Aign S, Kron J. Vascular refilling is independent of volume overload in hemodialysis with moderate ultrafiltration requirements. Hemodial Int. 2016.20: 484–491.
38. Kron S, Schneditz D, Czerny J, Leimbach T, Budde K, Kron J. Adjustment of target weight based on absolute blood volume reduces the frequency of intradialytic morbid events. Hemodial Int 2018.22: 254–260.
39. Hur E, Usta M, Toz H, et al. Effect of fluid management guided by bioimpedance spectroscopy on cardiovascular parameters in hemodialysis patients: A randomized controlled trial. Am J Kidney Dis 2013.61: 957–965.

bioimpedance; blood volume; fluid management; hemodialysis

Copyright © 2018 by the American Society for Artificial Internal Organs