Secondary Logo

Journal Logo


Extravascular Lung Water and Peripheral Volume Status in Hemodialysis Patients With and Without a History of Heart Failure

Joseph, Geena*; MacRae, Jennifer M.; Heidenheim, A Paul*; Lindsay, Robert M.*

Author Information
doi: 10.1097/01.mat.0000221751.98144.03
  • Free


In hemodialysis patients, the maintenance of volume homeostasis is a constant battle in which the goal is to attain a target “dry” weight after dialysis. Current methods for determining dry weight are not accurate. Dry weight is determined clinically and often defined as the lowest weight a patient can tolerate without developing symptoms or hypotension. Many patients with cardiac dysfunction have low blood pressure, which makes the process of reaching their dry weight challenging. Hence, this population may be chronically underdialyzed and volume-overloaded.

The studies to date on volume status and fluid shifts have focused on stable hemodialysis patients without significant cardiac dysfunction. Congestive heart failure (CHF) and cardiac dysfunction are common problems in dialysis patients. Each year, among patients initiating dialysis, 36% have CHF and an additional 7% develop the condition while on dialysis therapy.1 Given the high prevalence of cardiac dysfunction and the fact that cardiovascular disease is the leading cause of mortality and morbidity in this population,2 the study of this group is important.

Krivitski and colleagues3 first described the measurement of extravascular lung water (EVLW) using blood ultrasound velocity and electrical impedance dilution. The technique was initially studied in animals and found to agree with gravimetric measurements.4,5 The technique has also been tested in hemodialysis patients and found to be feasible, safe, reproducible, and consistent with invasive isotopic methods.6,7 With this technique, isotonic saline is used as a nondiffusible indicator and hypertonic (5%) saline is used as a diffusible indicator. By injecting these solutions and following their transit through the cardiopulmonary circulation, EVLW is derived from the cardiac output, the amount of water transferred to blood, and the increase in blood osmolality measured at the moment of osmotic equilibrium. See the accompanying article by MacRae et al.7 for a thorough discussion of the present technique.

Bioimpedance spectroscopy is a noninvasive method of measuring total body water (TBW) and its distribution into intracellular fluid (ICF) and extracellular fluid (ECF). This technique measures the resistance of body fluid compartments and thus body hydration. The ratio of the resistance reflects the relative volume of the ICF and ECF compartments. This technique has been well studied in the hemodialysis population.8,9

The purpose of this study was to compare EVLW, ECF, and ICF volumes between hemodialysis patients with a history of CHF and those without such a history. We hypothesized that hemodialysis patients with a history of CHF are volume overloaded and, therefore, have higher EVLW and ECF/ICF ratios. We also evaluated fluid volumes in two patients with CHF and pulmonary edema.

Materials and Methods

Study Description

Hemodialysis patients at the Adam Linton Dialysis Unit, London Health Sciences Centre, London, Ontario, were studied during routine hemodialysis treatments. The subjects had all been receiving conventional thrice weekly hemodialysis therapy for end-stage renal failure for at least 1 year. A total of 29 patients were included in the study, 12 with a history of CHF and 17 without a known history. The information regarding their history was taken from the patients' charted problem list, implying that the patients had or had not a hospital admission for CHF. Six patients had polytetrafluoroethylene bridging grafts (Gore-Tex inc, Newark, DE) and 23 had arteriovenous fistulae, all known to be well functioning on routine monthly testing of access flow and recirculation in the dialysis unit. These patients were selected based on their known hemodynamic stability during dialysis, well-functioning arteriovenous access, and willingness to participate in the study. All patients were over the age of 18 and gave written informed consent. All patients had echocardiograms or myocardial perfusion scans (MIBI) within 21 months (range 1week to 21months) of being enrolled in the study to evaluate cardiac function including left ventricular ejection fraction. They were all judged to be clinically euvolemic at the time of the study.

Details regarding the patients' comorbidities, dialysis adequacy, and nutritional status were recorded. Urea kinetic modeling is done routinely every 3 months in the dialysis unit and all values given were within 41 days (range 1–41days) of the study. Urea kinetic modeling was used to derive Kt/V, a measure of dialysis adequacy where K is the dialyzing membrane clearance, t is the time of dialysis delivered in minutes, and V is the volume of distribution of urea. Dietary protein intake was estimated from the protein equivalent of total nitrogen appearance normalized to body weight (nPNA). Albumin and nPNA are considered markers of nutrition. Another marker of nutrition used was the percentage standard body weight (%SBW), which is a patient's actual weight (postdialysis weight) expressed as a percentage of normal body weight for healthy Americans of similar sex, height, age range, and skeletal frame size. The data for healthy Americans were derived from the second National Health and Nutrition Evaluation Survey (NHANES II) as per the Dialysis Outcomes Quality Initiative (DOQI) guidelines.10

In addition, two patients with a history of CHF were studied when they presented with clinical and radiographic evidence of pulmonary edema.

All TBW, ICF, and ECF measurements were made before the initiation of hemodialysis for all 29 patients. Cardiac output (CO) and EVLW were measured for each patient in the first 60 minutes of the dialysis session. Patients were in minimum ultrafiltration (100 ml/h) until measurements were done and the amount of ultrafiltration was recorded. The EVLW values obtained were recorded as an absolute value and normalized (EVLWn) to the patient's weight at the time of measurement (ml/kg). This was done by adjusting the predialysis weight for ultrafiltration and fluid balance at the measurement time. All measurements were taken twice and the average was used in the final analysis. Baseline blood pressure, heart rate, predialysis weight, and target body weight were recorded. Target weight is defined as the lowest weight the patient can tolerate without developing symptoms of volume overload or hypotension.

Description of the Technology

Measurements of CO and EVLW were performed using the Transonic Hemodialysis Monitor, HDO1 Plus (Transonic Systems Inc, Ithaca, NY). This method uses the ultrasound dilution principle, which has been previously described.3,6,7 This technique has been proven to be safe, easy to perform during dialysis, and gives reproducible (mean differences within replicates 8–12%) results.6,7

The ultrasound dilution technique for measurement of lung water is based on the same theory as the classic indicator dilution method using nondiffusible and diffusible indicators, but instead of measuring the concentration profiles, the change in blood density is measured by ultrasound.3,7 The basic principle is that lung water can be measured by passing a nondiffusible substance through the lung vasculature, which causes a transient efflux of lung water into the vascular space, which can be measured noninvasively with ultrasound. This dilution technique can only calculate lung water from perfused areas; therefore, the EVLW calculated is functional, not anatomical, lung water. A detailed description of the technique and theoretical basis for the EVLW calculation is given in the accompanying article.7 Lung water volume using the ultrasound technique has been shown to be consistent with animal and human studies using the classic indicator dilution method.3–7

The experimental protocol involved adding an adaptor (Transonic Flow-QC set, Transonic Systems) to the venous and arterial dialysis lines, and placing an ultrasound probe on each line (which had been calibrated for the tubing) to detect changes in the density of blood. Then 30 ml of normal saline prewarmed to 37°C was injected into the venous blood line over 5–7 seconds and the ultrasound dilution curve was recorded by the probe on the arterial line. Approximately 3 minutes later, 20 ml hypertonic saline at 37°C was injected into the venous line over 5–7 seconds and the arterial probe recorded the next dilution curve. The dilution curves were overlapped and the results applied automatically to the Krivitski equation3 using Transonic software version 1.03 to obtain a measurement of EVLW. This was done twice for each patient and the average value was used in the data analysis.

Measurements of ECF, ICF, and TBW were done using bioimpedance spectroscopy before the initiation of dialysis. Body water compartments were measured using bioelectric impedance technology using the Hydra ECF/ICF Spectrum Analyzer 4.200 (Xitron Technologies, San Diego, CA). Each subject's height and weight were recorded for the spectrum analyzer. Whole-body bioimpedance measurements were done using the standard distal tetrapolar lead arrangement, which involves attaching electrodes to the patient's right hand and right foot.11 Subjects were placed in a supine position for 5 minutes and the bioimpedance measurement was taken before the subject was started on hemodialysis.

The impedance spectrum is measured by a spectrum analyzer using 50 programmed frequencies from 5 kHz to 1 MHz. The analyzer then fits these data to the Cole-Cole model of impedance spectra in living tissue using nonlinear curve fitting software.12,13 The model calculates the resistance of the ECF and ICF (Re and Ri). Volumes are predicted from the Re and Ri using equations from Hanai emulsion mixture theory, and whole-body ECF and ICF volume data are generated.13 The measurement of TBW using isotopic dilution methods is the “gold standard,” but is not practical in the clinical setting. Bioimpedance spectroscopy has been shown to be simple, safe, precise, and reproducible when compared with anthropometric measurements.14 In addition, it has been well studied in the hemodialysis population.8,9

Statistical Analysis

Statistical analysis was done using SPSS version 12.0 (SPSS, In., Chicago, IL). Results are expressed as mean values and standard deviations. The differences between mean values were evaluated by Student's paired t test for unpaired data. A two-tailed p value ≤ 0.05 was considered statistically significant. Associations between variables were assessed by Pearson's correlations and considered significant at p ≤ 0.05.


The characteristics of the 29 subjects are given in Table 1. The group with a history of CHF was slightly older and had a mean ejection fraction of 46.7% compared with 56.3%, which was a significant difference (p < 0.05). There were no other significant differences in the comorbidities between the two groups. In terms of dialysis, all patients were receiving adequate dialysis with Kt/V clearances > 1 .2, which is recommended by the DOQI guidelines. The dietary protein intake, estimated from nPNA and albumin (which are markers of nutrition), were similar in both groups. The CHF group had a significantly lower %SBW than the group without CHF (p < 0.05). However, the means for both groups (96 ± 13.4 and 110 ± 15.0) were within the target of 90% and 110% of SBW, as recommended by the DOQI guidelines.10

Table 1
Table 1:
Patient Characteristics

The values for CO and EVLW for each patient are shown in Table 2. The values for EVLW normalized (EVLWn) to body weight were 3.55 ± 0.94 ml/kg in patients without a history of CHF and 3.88 ± 0.82 ml/kg in patients with a history of CHF, a difference that was not statistically significant. There was also no relationship between ejection fraction and normalized lung water, as depicted in Figure 1.

Table 2
Table 2:
Values for Ejection Fraction, Weight, Cardiac Output, EVLW, and ECF/ICF Ratio for Each Patient
Figure 1.
Figure 1.:
The relationship between normalized extravascular lung water (EVLWn) and ejection fraction (%). There was no relationship between normalized extravascular lung water and ejection fraction. (r = –0.127, NS)

The ECF/ICF ratio was significantly higher in patients with a history of CHF (1.27 ± 0.29) than in those without a history of CHF (1.04 ± 0.15) (p < 0.05). There was one patient in the CHF group whose bioimpedance measurements were not done because her right leg had been amputated.

There was a statistically significant positive correlation between normalized lung water and the ratio of ECF to ICF, as seen in Figure 2 (r = 0.543, p < 0.01). This suggests that patients whose ECF/ICF ratios are higher tend to have higher EVLW.

Figure 2.
Figure 2.:
The relationship between normalized extravascular lung water (EVLWn) and ECF/ICF ratio. There is a positive correlation between the ECF:ICF ratio and the mean normalized lung water measurement (r = 0.543, p < 0.01), suggesting that patients whose ECF/ICF ratios are higher tend to have higher EVLW.

Two patients were studied while they were clinically and radiographically in pulmonary edema; their values are given in Table 3. Patient 1 presented with bilateral crackles approximately one quarter up his lung fields, and pedal edema. His measurements were completed within first 40 minutes and the last 30 minutes of a 3-hour and 45-minute dialysis session. His predialysis weight, postdialysis weight, and target weight are given in Table 3. His EVLWn was 7.95 ml/kg early in dialysis and 6.37 ml/kg near the end of the dialysis session. He was studied some months later when he was euvolemic. At that time his EVLWn was 4.90 ml/kg, which was considerably lower than his previous measurements. Patient 2 presented with an elevated jugular venous pressure (JVP), bilateral crackles in approximately two thirds of her lung fields, pedal edema, and hypoxia with a Pao 2 of 64 mm Hg on 100% oxygen. Her first measurements were taken well into dialysis treatment (2 hours, 40 minutes) after approx. 2 l ultrafiltration had taken place and then repeated in the last 30 minutes of the 4-hour session. These two EVLWn measurements were similar (5.95 ml/kg and 6.02 ml/kg) and much higher than the value measured some months later, at a time when she was not in pulmonary edema (4.48 ml/kg). Both patients, while in pulmonary edema, had EVLWn values higher than the 95% confidence intervals for both the group with a history of CHF and the group with no such history (p < 0.05). In fact, both patients' EVLWn values were higher than the highest EVLWn value observed in either group. A graphical representation of the two individuals at baseline and in pulmonary edema compared with the others is given in Figure 3 and Table 4.

Table 3
Table 3:
Values for the Two Patients at Baseline and During an Episode of Pulmonary Edema
Figure 3.
Figure 3.:
A graphical representation of the two individuals studied during an episode of congestive heart failure (CHF) and at their baseline compared with all stable patients. Both patients in CHF with pulmonary edema had EVLWn values higher than the 95% confidence intervals for both the group with a history of CHF and no history of CHF (p < 0.05). In fact, both patients had EVLWn values that were higher than the highest EVLWn observed in either group. PE indicates patients when they were in CHF with pulmonary edema; Stable, baseline measurements when patients were euvolemic.
Table 4
Table 4:
Comparison of Patients With Overt Congestive Heart Failure and the Other Two Groups


Bioimpedance spectroscopy technology has been extensively reported and clinically tested, and is an accepted method for measuring ECF/ICF in this population. It has been shown that the ECF/ICF ratio increases with volume overload and other conditions that cause contraction of the ICF, such as wasting conditions, malnutrition, or severe illness. Both groups were nutritionally adequate and equivalent by the measurements done routinely in the hemodialysis unit. Although there was a statistically significant difference between the two groups in terms of %SBW, it is not clinically significant given that the mean in both groups (96 ± 13.4 and 110 ± 15.0) were within the target of 90% and 110% recommended by the DOQI guidelines.10 Patients below 90% are considered mildly to moderately malnourished, whereas individuals above 115% are considered obese. The similar albumin and nPNA levels in the two groups support the likelihood that the groups were equivalent from a nutritional point of view, making the difference in ECF/ICF ratio more the result of ECF volume expansion than of ICF contraction.

This study shows that hemodialysis patients are generally ECF volume expanded, likely secondary to volume overload, and to a greater extent if they have a history of CHF. Both groups were significantly ECF volume expanded with mean ECF/ICF ratios of 1.04 and 1.27 (p < 0.01), as compared with the expected ratio in healthy subjects. Standard physiology texts indicate that, in a male, 60% of body weight is water; two thirds of this water is ICF, and one third ECF, therefore, and the ECF/ICF ratio is approximately 0.50.15 Spiegel et al., using the same technology as this study, confirmed this in healthy males, and showed that the ECF/ICF ratio was significantly higher in hemodialysis patients compared with normal controls, despite being at their target weight.9 Thus, there appears to be a spectrum of ECF volume expansion among hemodialysis patients, and those with a history of CHF tend to be at the higher end of the spectrum even without overt peripheral edema. Previous studies have shown that dialysis ultrafiltration reduces the ECF/ICF ratio with fluid removal, further supporting the conclusion that the ECF expansion we are seeing in hemodialysis patients is related to volume overload.16

The group with a history of CHF had a lower ejection fraction than the group without a history of CHF. This difference was small but statistically significant. Clinical CHF is an outcome measure that has been correlated with mortality. In a prospective cohort study, Harnett et al.17 showed that the median survival of dialysis patients with clinical CHF or a history of the condition was 36 months compared with 62 months in those without CHF, a difference that was statistically significant. This study showed that a clinical history of CHF was a strong, independent, adverse prognostic indicator. Therefore, in our study, patients were assigned to the individual groups based on history of CHF, regardless of the measured ejection fraction. Independent of group assignment, no relationship was found between EVLWn or ECF/ICF and ejection fraction (Figure 1).

The mean EVLWn was higher in the group with a history of CHF (3.88 + 0.82 ml/kg) than in the other group (3.55 + 0.94 ml/kg); however, this difference was not statistically significant. There was, however, a correlation between the ECF/ICF ratio and EVLWn (r = 0.543, p < 0.01) across both groups as shown in Figure 2. This suggests that patients whose ECF/ICF ratios are higher tend to have higher EVLW. Both parameters are at least partially related to fluid status; however, they can be influenced by other factors (e.g., EVLW by cardiac function and ECF/ICF by body composition). This supports the concept that EVLW expansion is also a spectrum and that as the ECF volume expands so does the EVLW, even without overt pulmonary edema. It can be inferred that clinically evident pulmonary edema eventually develops with progressive volume overload, irrespective of any other insult (e.g., myocardial damage).

The values for EVLW normalized to body weight are similar to the results obtained by Garland et al.6 (3.29 ± 1.0 ml/kg) and McRae et al. (3.67 ± 1.47 ml/kg),7 who studied lung water in hemodialysis patients without a documented history of cardiovascular disease. Thus, the technique shows consistent results in this patient population.

We were able to study two patients while they were clinically and radiographically in pulmonary edema. In both cases, the EVLW volumes in early dialysis and near the end of dialysis were significantly higher than the values subsequently measured at a time when the patients were not in pulmonary edema. This indicates that although they clinically improved with fluid removal, they were still in EVLW volume overload. The EVLW values obtained during the episodes of pulmonary edema were significantly higher than the values otherwise obtained in the stable population. Unfortunately, we were not able to obtain bioimpedance measurements during the acute episodes of pulmonary edema because the patients were undergoing emergent dialysis.

Our findings are consistent with older studies that used the classic method of multiple indicator dilution and found that lung water volume is 2–3 ml/kg in stable normal nonrenal subjects18–20 and significantly higher in patients with overt CHF (4–5 ml/kg).20 In our study, stable hemodialysis patients had EVLWn values between 2.2 ml/kg and 5.4 ml/kg, whereas our two patients in CHF had values ranging from 5.9 ml/kg to 7.9 ml/kg. It may be that hemodialysis patients, in general, have greater variability in the degree of extravascular lung water expansion without overt pulmonary edema compared with normal subjects (depicted in Figure 4). Perhaps these patients somehow adapt to their chronic volume overload (ECF and EVLW), and a higher EVLW (and ECF) is needed before overt pulmonary edema occurs. However, it is possible that the inference from Figure 4 is purely an artifact of different methodologies. More research is needed to compare hemodialysis patients with normal subjects using the same techniques.

Figure 4.
Figure 4.:
Comparison of lung water in normal subjects and hemodialysis patients, when stable and when in overt congestive heart failure (CHF). A comparison of lung water in normal subjects and hemodialysis patients when they are stable and in overt CHF is shown. Data from the present study and six other studies are combined in this figure. Different techniques were used to investigate the two groups.

One potential limitation of our study is that the ejection fractions were not measured at the same time as the study measurements. Furthermore, we had no knowledge of the patients' clinical status at the time of the echocardiograms or MIBIs. If the ejection fractions had been evaluated at the same time as the study measurements, there may have been a correlation between EVLW and ejection fraction. The history of CHF was taken from each patient's problem list, but was not verified. Therefore, it is possible that some patients in the non-CHF group may have had unrecorded episodes of CHF. Also, the patients within the CHF group may not have had a hospital admission; however, this is less likely, if the information was in the patients' chart. This could have influenced our findings of a significant ECF/ICF ratio and nonsignificant EVLW difference between the two groups.

Another limitation of this study is that dialysis patients without arteriovenous access were excluded (an arteriovenous access is necessary for the measurement of EVLW), so results are only applicable to hemodialysis patients with arteriovenous access. The determination of patients' hydration status was done clinically by assessing the JVP, leg edema, and auscultating for crackles, but not with any other methods (e.g., radiographically). If a patient with pulmonary edema was missed and included in the study, our EVLW results may actually be higher than the true value. Given that only one investigator (GJ) assessed the patients, there is at least consistency.

In summary, the group of patients with a history of CHF who were not in heart failure did not have higher EVLW values, but did have higher ECF/ICF ratios. Furthermore, patients whose ECF/ICF ratios are higher tend to have higher EVLW. It can be inferred that ECF volume expansion eventually leads to an EVLW that exceeds the “normal range” at which point clinical and radiographic evidence of pulmonary edema manifests. In conclusion, we surmise that these developing technologies of volume measurement are of value in this challenging clinical area.


1.Schreiber BD: Congestive heart failure in patients with chronic kidney disease and on dialysis. Am J Med Sci 325: 179–193, 2003.
2.Foley R, Parfrey P, Surnak M: Clinical epidemiology of cardiovascular disease. Am J Kidney Dis 32: S112–S119, 1998.
3.Krivitski N, Kislukhin V, Dobson A, et al.: Volume of extravascular lung fluid determined by blood ultrasound velocity and electrical impedance dilution. ASAIO J 44: M535–M540, 1998.
4.Krivitski NM, Trouve R: Animal validation of extravascular lung water by sound velocity and electrical impedance dilution [abstract]. FASEB J 9:A352, 1996.
5.Wilkins PA, Gleed RD, Krivitski NM, Dobson A: Extravascular lung water in the exercising horse. J Appl Physiol 91: 2442–1250, 2001.
6.Garland JS, Kianfar C, Nesrallah G, et al.: Measurement of extravascular lung water in hemodialysis patients using blood ultrasound velocity and optical density dilution. ASAIO J 48: 398–403, 2002.
7.MacRae JM, Joseph G, Kislukhin V, et al.: Determining lung water volume in stable hemodialysis patients. ASAIO J 52: XXX–XXX, 2006.
8.Shulman T, Heidenheim P, Kianfar C, et al.: Preserving central blood volume: Changes in body fluid compartments during hemodialysis. ASAIO J 47: 615–618, 2001.
9.Spiegel DM, Bashir K, Fisch B: Bioimpedance resistance ratios for the evaluation of dry weight in hemodialysis. Clin Nephrol 53: 108–114, 2000.
10.NKF-K/DOQI 2000 Clinical Practice Guidelines for Nutrition in Chronic Renal Failure. Am J Kidney Dis 35: 6(Suppl 2): S17–S104, 2000.
11.Kushner RF: Bioelectrical impedance analysis: A review of principles and applications. J Am Col Nutr 11: 199–209, 1992.
12.Cole KS: Membranes, Ions, and Impulses: A Chapter of Classical Biophysics. Berkeley, University of California Press, 1972.
13.DeLorenzo A, Andreoli A, Mattie J, Withers P: Predicting body cell mass with bioimpedance using theoretical methods: A technology review. J Appl Physiol 82: 1542–1558, 1997.
14.Van der Ham EC, Kooman JP, Christiaans MH, et al.: Body composition in renal transplant patients: Bioimpedance analysis compared to isotope dilution, dual x-ray absorptiometry and anthropometry. J Am Soc Nephrol 10: 1067–1079, 1999.
15.Halperin ML, Goldstein M: Sodium & water physiology, in Fluid, Electrolyte & Acid-Base Emergencies. Philadelphia: WB Saunders, 1988: 140–142.
16.Prakash S, Reddan D, Heidenheim P, et al.: Central, peripheral, and other blood volume changes during hemodialysis. ASAIO J 48: 379–382, 2002.
17.Harnett JD, Foley RN, Kent GM, et al.: Congestive heart failure in dialysis patients: Prevalence, incidence, prognosis, and risk factors. Kid Int 47: 884–890, 1995.
18.Anthonisen P, Crone C: Transcapillary migration of ethyl alcohol in the pulmonary circulation: A method for determining the water content of the lungs in vivo. Acta Physiol Scand 37: 370–379, 1956.
19.Marshall BE, Teichner RL, Kallos T, et al.: Effect of posture and exercise on the pulmonary extravascular volume in man. J Appl Physiol 31: 375–379, 1971.
20.Lilienfield LS, Freis ED, Partenope EA, Morowitz HJ: Transcapillary migration of heavy water and thiocyanate ion in the pulmonary circulation of normal subjects and patients with congestive heart failure. J Clinical Invest 34: 000–8, 1954.
Copyright © 2006 by the American Society for Artificial Internal Organs