Collapse of Fluid Balance and Association with Survival in Patients with Advanced Cancer Admitted to a Palliative Care Unit: Preliminary Findings

Abstract

Background: Few studies have investigated water balance as a predictor of survival in cancer patients in the last days of life.

Objective: To examine the association between extracellular water (ECW), intracellular water (ICW), ratio of ECW to ICW (ECW/ICW), and survival in nonedematous and edematous patients with advanced cancer admitted to a palliative care unit.

Design: A prospective observational study.

Setting/Subjects: Patients with advanced cancer admitted to a palliative care unit.

Measurements: Upon enrollment, bioelectrical impedance analysis was used to assess ECW, ICW, and body composition. We conducted time-to-event analyses using the Kaplan–Meier method, log-rank test, and univariate and multivariate Cox regression analyses.

Results: A total of 190 of 204 patients who participated in this study had ICW and ECW measures available for analysis. The median survival was 10 days (95% confidence interval [CI] 9–12) days. The median ECW, ICW, and ECW/ICW were 18.9 L, 19.1 L, and 1.0 for 72 nonedematous patients, and 21.9 L, 20.1 L, and 1.1 for 118 edematous patients, respectively. In univariate Cox regression analysis, ICW ≤20 L was associated with a shorter survival in the nonedematous patients (hazard ratio [HR] 1.92, 95% CI 1.10–3.34, p = 0.02) and a longer survival in the edematous patients (HR 0.61, 95% CI 0.41–0.90, p = 0.01). In multivariable analysis, ICW (≤20 L vs. >20 L) remained an independent prognostic factor in edematous patients (HR 0.64, 95% CI 0.43–0.95, p = 0.03).

Conclusions: Greater ICW was an independent predictor of poorer survival in cancer patients with edema in the last days of life.

Introduction

The prognosis of patients with advanced cancer is a major determinant in clinical decision making and end-of-life discussion in palliative care settings.¹ However, clinician prediction of survival has been found to be inaccurate and optimistic.^2,3 Furthermore, although several prognostic models, for example, the Palliative Prognostic score (PaP score)⁴ and the Palliative Prognostic Index,⁵ have been developed, there are still major limitations related to their accuracy or complexity.^1,6 Measurements by bioelectrical impedance analysis (BIA), that is, phase angle reflecting cellular membrane integrity and hydration level, have been preliminarily explored as a simple objective prognostic marker for patients with advanced cancer.^7–10 Although phase angle has been shown to have prognostic utility, it was not significant in patients with edema.⁹

BIA is commonly employed to assess body composition, such as fat-free mass (FFM) and fat mass, in various diseases.^11–13 In addition, BIA has been used as a promising tool for the measurement of water balance, that is, extracellular water (ECW), intracellular water (ICW), total body water (TBW = ECW + ICW), and a ratio of ECW to ICW (ECW/ICW).^14–16 Increases in ECW, TBW, and ECW/ICW and decreases in ICW may be seen in patients with fluid shift from intracellular to extracellular space, that is, second space (interstitial fluid and edema) and third space (ascites and pleural effusion).^14–16 Such fluid shift confers a poor prognosis in critically ill patients with various diseases (chronic liver disease, chronic renal disease, and acute heart failure, in which edema is common).^17–22

Much remains unknown about the prognostic utility of ECW, ICW, and ECW/ICW in the cancer palliative care setting. A small observational study that enrolled 84 patients with advanced cancer and a median survival of 22 weeks reported that ECW and ECW/ICW were significant predictors of survival.²³ To our knowledge, no investigators have examined these prognostic factors in patients with days to weeks of survival. Furthermore, it remains unclear how edematous status may affect accuracy of BIA. A better understanding of the ECW, ICW, and ECW/ICW in patients with weeks or days of survival may inform the use of these noninvasive variables in this patient population to facilitate clinical decision making.

We recently completed a study examining the prognostic utility of phase angle in patients admitted to a palliative care unit.⁹ In this secondary analysis, we examined the association between ECW, ICW, ECW/ICW, and survival in nonedematous and edematous patients with advanced cancer admitted to a palliative care unit.

Methods

Study setting and criteria

This is a secondary analysis of a study to examine the prognostic value of phase angle. The methods have been described previously.⁹ In brief, we conducted a prospective longitudinal observational study in the palliative care unit at University of Texas MD Anderson Cancer Center between April 21, 2015, and August 24, 2016. Patients with advanced cancer, aged 18 years old or more, admitted to the palliative care unit within 3 days, and receiving parenteral hydration within the past 48 hours were eligible. The requirement for parenteral hydration was because dehydration could negatively impact on the accuracy of measurements by BIA.²⁴ Patients with defibrillator, cardiac pacemaker, any implanted electrical device, extensive local infection, or wound preventing placement of BIA pads were excluded. Patients unable to communicate in English were also excluded.

This study was approved by the Institutional Review Board at MD Anderson Cancer Center. Written informed consent was obtained from patients or their surrogates if consciousness of the patient was unclear.

Data collection

We collected patient demographics at baseline, that is, age, gender, race, primary cancer site, and PaP score. The PaP score is a prognostic tool validated for patients with advanced cancer. It consists of six variables, that is, dyspnea at rest, anorexia, Karnofsky Performance Status (KPS), clinician prediction of survival, total leukocyte count, and lymphocyte percentage. The total score range is 0–17.5 and a higher score indicates worse survival.⁴

We also assessed measurements by BIA, Edmonton Symptom Assessment System (ESAS),²⁵ and the Memorial Delirium Assessment Scale (MDAS)²⁶ from the time of enrollment. Measurements by BIA were obtained with the RJL Systems Quantum IV (Clinton Township, MI). An electrode was placed over the right foot between the medial and the lateral malleoli at the ankle and another over the middle of the dorsal surface of the right hand between the distal prominence of the radius and the ulnar styloid in supine position. A single 50 kHz frequency alternating low voltage electrical current was applied.

We evaluated the association between ECW, ICW, ECW/ICW, and survival in patients with advanced cancer who had a median survival of days to weeks. We elected to examine ECW, ICW, and ECW/ICW because previous studies consistently found that they were associated with poorer survival.^22,23 We also compared these parameters between patients with or without clinical evidence of peripheral edema. In addition to ECW and ICW, BIA also provided phase angle that has been reported previously.⁹

ESAS is a validated tool to examine the intensity of 10 symptoms, including pain, fatigue, nausea, depression, anxiety, drowsiness, shortness of breath, appetite, feeling of well-being, and sleep. The average intensity of each symptom over the past 24 hours is evaluated with a numeric rating scale ranging from 0 (none) to 10 (worst).²⁵ MDAS is a validated tool consisting of 10 items clinician rated for assessment of delirium in patients with cancer. It examines the level of consciousness, disorientation, memory, recall, attention, disorganized thinking, perceptual disturbance, delusions, psychomotor activity, and sleep. A score for each item is 0–3, and a total score is 0–30. A score of ≥13 indicates delirium.²⁶

Survival data from enrollment were obtained from institutional databases and electronic health records.

Statistical analyses

The sample size justification is provided elsewhere.⁹ We summarized patient baseline body weight, body mass index, ECW, ICW, and ECW/ICW using descriptive statistics, including mean, standard deviation (SD), median, and interquartile range (IQR).

We estimated overall survival using the Kaplan–Meier method and compared between ECW (≤20 L vs. >20 L), ICW (≤20 L vs. >20 L), and ECW/ICW (≤1 vs. >1) using the log-rank test. We prespecified the median value as cutoff. The univariate Cox proportional hazards model was used to assess the effect of ECW, ICW, ECW/ICW, patient characteristics (age, gender, race, and primary cancer site), and prognostic variables (PPS and MDAS) on overall survival. We then included age, gender, race, primary cancer site, PPS, MDAS, ECW, ICW, and ECW/ICW in a multivariate model with stepwise selection. We conducted post hoc subgroup analyses on patients with and without clinical evidence of peripheral edema, because edema is common among patients with advanced cancer in terminal phase and may affect measurements by BIA.⁹

All computations were carried out in SAS 9.4 (SAS Institute, Inc., Cary, NC) and S + 8.2 for Windows (TIBCO Software, Inc.). A p value of <0.05 is considered to be statistically significant.

Results

Patient characteristics

Among the 571 patients admitted to the palliative care unit during the study period, 356 (62.3%) were eligible and 204 (35.7%) participated in this study. The baseline characteristics have been reported previously.⁹ In brief, the mean age was 61.8 years (SD 13.3 years), the number of females was 110 (53.9%), the number of patients with MDAS ≥13/30 was 106 (52.2%), and the mean KPS was 30.2% (SD 15.8%). The median survival was 10 days (95% confidence interval [CI] 9–12 days) for all of the patients, 11 days (95% CI 9–20 days) for the patients without edema (n = 74, 36.3%), and 9 days (95% CI 6–10 days) for the patients with edema (n = 130, 63.7%).

A total of 190 patients had baseline ECW and ICW measures, therefore, serving as the base for this analysis. Mean values of body weight and body mass index, and medians and mean values of ECW, ICW, and ECW/ICW are given in Table 1. The median ECW, ICW, and ECW/ICW at the baseline were 20.2 L (IQR 16.9–25.4 L), 19.7 L (IQR 16.4–23.9 L), and 1.1 (IQR 0.9–1.2), respectively (Table 1). The mean values of water balance, that is, ECW, ICW, and TBW, in healthy subjects (A),¹⁴ the patients without edema (B), and the patients with edema (C) are shown in Figure 1, respectively.

Table 1.

Patient Body Weight, Body Mass Index, Extracellular Water, and Intracellular Water (n = 190)

Characteristics	All patients (n = 190)	Patients without edema (n = 72)	Patients with edema (n = 118)
Body weight (kg)
Mean (SD)	75.5 (20.4)	69.0 (16.9)	79.4 (21.4)
BMI (kg/m2)
Mean (SD)	27.4 (11.7)	24.5 (5.4)	29.2 (13.9)
ECW (L)
Median (IQR)	20.2 (16.9–25.4)	18.9 (15.3–22.2)	21.9 (17.6–27.5)
Mean (SD)	21.9 (7.1)	19.6 (6.0)	23.3 (7.4)
ICW (L)
Median (IQR)	19.7 (16.4–23.9)	19.1 (15.6–22.6)	20.1 (16.5–24.4)
Mean (SD)	20.7 (5.7)	19.9 (5.6)	21.2 (5.8)
ECW/ICW
Median (IQR)	1.1 (0.9–1.2)	1.0 (0.9–1.1)	1.1 (0.9–1.2)
Mean (SD)	1.1 (0.2)	1.0 (0.2)	1.1 (0.2)

BMI, body mass index; ECW, extracellular water; ICW, intracellular water; IQR, interquartile range; SD, standard deviation.

FIG. 1.

Water balance in patients with advanced cancer admitted to a palliative care unit in terminal phase. This figure shows mean values of water balance, that is, ECW, ICW, and TBW, in healthy subjects (A),¹⁴ the patients without edema (B), and the patients with edema (C), respectively. Both ECW and TBW in the patients with and without edema were higher than those in healthy subjects, and ECW and TBW in the patients with edema were greater than those in the patients without edema. However, although ICW in the patients with and without edema was lower than that in healthy subjects, ICW in the patients with edema was greater than that in the patients without edema. It seems that all of the patients in this study had overhydration in greater or less degree, because both ECW and TBW in all patients were higher than those in healthy subjects. Covert collapse of water balance already occurred in the patients without edema, whereas overt collapse occurred in those with edema. ECW, extracellular water; ICW, intracellular water; TBW, total body water.

Survival analysis

There were no significant differences in survival between ECW ≤20 L and ECW >20 L in patients without edema and the patients with edema (log-rank test p = 0.98 and 0.29, respectively). However, an ICW ≤20 L was associated with decreased survival compared with ICW >20 L in survival analysis for the patients without edema (log-rank test p = 0.02) (Fig. 2A). In contrast, an ICW >20 L was associated with decreased survival compared with ICW ≤20 L in the patients with edema (log-rank test p = 0.009) (Fig. 2B). There were no significant differences in survival between ECW/ICW ≤1 and ECW/ICW >1 in patients without edema, nor patients with edema (log-rank test p = 0.39 and 0.98, respectively).

FIG. 2.

Kaplan–Meier survival curves This figure shows Kaplan–Meier survival curves in ICW of patients without edema (A) and patients with edema (B). We prespecified the median value as cutoff, because the ideal cutoff has not been defined for this population. A lower ICW (≤20) was associated with shorter survival in the patients without edema, whereas a higher ICW (>20) was associated with shorter survival in the patients with edema.

Table 2 shows the findings of univariate and multivariable Cox regression analysis. ICW remained an independent prognostic factor after adjusting for established prognostic factors in multivariable analysis in the patients with edema (≤20 L vs. >20 L, HR 0.64, 95% 0.43–0.95, p = 0.03). The differential effect of ICW (≤20 L vs. >20 L) on overall survival between patients with or without edema was statistically significant (p = 0.008) when including an interaction between edema status and ICW (≤20 L vs. >20 L). Furthermore, phase angle was significantly correlated with ECW (Spearman correlation coefficient −0.29, p < 0.001), ICW (Spearman correlation coefficient 0.27, p < 0.001), and ECW/ICW (Spearman correlation coefficient −0.72, p < 0.001) in all patients.

Table 2.

Univariate and Multivariate Cox Regression Analysis

Characteristics	Patients without edema (death/n = 59/72)				Patients with edema (death/n = 105/118)
	Univariate analysis		Multivariate analysisa		Univariate analysis		Multivariate analysisa
	HR (95% CI)	p Value	HR (95% CI)	p Value	HR (95% CI)	p Value	HR (95% CI)	p Value
Age (per year increase)	1.00 (0.98–1.02)	0.79			1.01 (1.00–1.03)	0.07
Male gender (female vs. male)	1.41 (0.83–2.38)	0.21			0.87 (0.59–1.28)	0.48
Race
White vs. others	0.56 (0.21–1.47)	0.24			2.26 (0.55–9.24)	0.26
Black vs. others	1.26 (0.45–3.54)	0.66			4.78 (1.01–22.56)	0.05
Hispanic vs. others	0.34 (0.10–1.17)	0.09			2.06 (0.46–9.31)	0.35
Karnofsky Performance Status	0.95 (0.93–0.97)	<0.001			0.95 (0.94–0.97)	<0.001
PaP score (per point increase)	1.27 (1.15–1.39)	<0.001	1.21 (1.09–1.34)	<0.001	1.17 (1.09–1.26)	<0.001	1.12 (1.03–1.20)	0.002
Memorial Delirium Assessment Scale (per point increase)	1.07 (1.04–1.10)	<0.001	1.04 (1.01–1.07)	0.005	1.06 (1.04–1.08)	<0.001	1.05 (1.03–1.07)	<0.001
Phase angle (≤3° vs. >3°)	4.51 (2.12–9.61)	<0.001	2.47 (1.15–5.33)	0.02	1.07 (0.67–1.69)	0.79
ECW, median (≤20 L vs. >20 L)	0.99 (0.58–1.70)	0.98			0.82 (0.55–1.20)	0.31
ICW, median (≤20 L vs. >20 L)	1.92 (1.10–3.34)	0.02			0.61 (0.41–0.90)	0.01	0.64 (0.43–0.95)	0.03
ECW/ICW, median (≤1 vs. >1)	0.80 (0.48–1.34)	0.40			0.99 (0.67–1.49)	0.98

^aAll the mentioned clinical variables were included in multivariate Cox regression analysis except for Karnofsky Performance Status, because it was already included in the PaP score.

CI, confidence interval; HR, hazard ratio; PaP score, Palliative Prognostic score.

Discussion

To our knowledge, this study is the first to explore the association between parameters of water balance measured by BIA, that is, ECW, ICW, and ECW/ICW, and survival in patients with advanced cancer admitted to a palliative care unit who had a survival of days to weeks.

Both ECW and TBW were elevated in our patients than in healthy subjects (Fig. 1), suggesting that our patients had overhydration to a variable degree. This observation may be related to frequent use of parenteral hydration in hospitals and/or pathophysiologic changes at the end of life. Covert collapse of water balance occurred in the patients without edema, whereas overt collapse occurred in those with edema. It is natural that ECW and TBW in the patients with edema were greater than those in the patients without edema due to overt collapse of water balance. Although ICW in all patients were lower than that in healthy subjects, ICW in the patients with edema was greater than that in the patients without edema (Fig. 1). This implies that severe collapse of water balance may also induce intracellular overhydration in cancer patients with days to weeks of survival.

The results of this study demonstrated that lower ICW is an independent predictor for poor survival in the patients without edema. It is known that patients with advanced cancer tend to have low phase angle and that low phase angle is correlated with worse prognosis and nutritional status in patients with advanced cancer without edema.^7–10,27 In addition, a lower phase angle is associated with a higher ECW/ICW ratio.¹⁴ Moreover, malnutrition and inflammation could deplete body cell mass, which eventually leads to the decrease in ICW and the increase in ECW/ICW among clinical populations.²⁸

Paradoxically, our results also showed that the subjects with greater ICW in the patients with edema had poor survival. It can be interpreted that severe fluid retention in a whole body may induce intracellular overhydration when water balance is highly collapsed in cancer patients with days to weeks of survival. As control of ICW is generally related to osmotic factors, the acute electrolyte imbalance, for example, hyponatremia, in terminal phase might cause an uncontrolled edema and increase in ICW.

Moreover, there were no significant differences in ECW among the patients with and without edema, because all patients had more or less collapse of water balance due to terminal phase. Furthermore, since the changes of ECW and ICW counteracted each other in ECW/ICW, the differences in it could be weakened particularly in the patients with edema.

The negative impact of positive fluid balance on morbidity and mortality has been revealed in critically ill patients with various diseases.^17–23 However, little is known about the implication of fluid retention in prognoses and how to monitor and manage fluid status in patients with advanced cancer, who tend to develop changes in body fluid distribution with migration of fluid from the intravascular to the extravascular space due to hypoproteinemia or anemia. Furthermore, systemic inflammation due to cancer cachexia not only promotes leakage of water from the intravascular to second and third space, but also decreases FFM that includes TBW (ECW and ICW).

Therefore, changes in ECW and ICW may be more pronounced in cancer patients with days to weeks of survival. Since malnutrition (hypoproteinemia and anemia) and systemic inflammation (proinflammatory cytokines and C-reactive protein) have individually poor outcomes in such population,^29–32 the combination of these parameters and measurements by BIA should be assessed in the future.

There are several limitations to this study. First of all, the findings cannot be generalized and an unmeasured bias could have influenced the results due to a prospective longitudinal observational study in a single center. Second, a systematic review revealed that application of equations validated in healthy subjects to predict body composition performs less well in oncologic and surgical patients and suggested that BIA estimations, irrespective of the device, could only be useful when performed longitudinally and under the same standard conditions.³³ The results of this study may imply that there are some differences between terminally ill patients with cancer and critically ill patients with various diseases. In addition, it is needed to monitor changes in body composition within individuals over time.

Third, patients with edema, ascites, or pleural effusions were included in analyses of this study. The use of BIA is still unclear whether the measurements are sufficiently accurate for clinical use in edematous patients. The assessment of water balance is known to have a limitation of the applicability of predictive equations generated by BIA in patients with altered hydration because states of hypo- and hyperhydration affect the electrolyte balance, which can influence the measurements by BIA regardless of the changes in water balance.^14–16 Fourth, the assessment of water balance was calculated by a single frequency and whole-body BIA in this study. This type of impedance may have limitation in differentiating ECW and ICW from TBW compared with a multiple frequency and segmental BIA.^14–16

Finally, a whole-body BIA measures the bioelectrical factors under the assumption of a steady water distribution, although water is not evenly distributed in the human body. Furthermore, a whole-body BIA is much more sensitive to changes of water balance in the limbs than in the trunk, that is, the same volume of ECW in the leg or that in trunk leads to different results.^14–16 However, since the mean KPS of all the patients in this study was ∼30 (severely disabled; hospital admission is indicated although death not imminent)³⁴ and the measurement in this study was performed in the supine position, fluid collection in legs due to gravity might be negligible.

Conclusion

Our findings provided evidence of collapse of water balance in hospitalized patients with advanced cancer in the last days to weeks of life. Furthermore, lower ICW was associated with poor survival in patients without edema, whereas higher ICW was associated with shorter survival in patients with edema. Future studies are needed to examine the pathophysiologic changes at the cellular and tissue level in these patients. Upon further validation, ICW may be useful as an objective measure of impending death in edematous patients in this setting.

Funding Information

This study was supported, in part, by National Institutes of Health (Grant Nos. R21CA186000-01A1; 1R01CA214960-01A1; and R21NR016736-01 to D.H.), an American Cancer Society Mentored Research Scholar Grant in Applied and Clinical Research (Grant No. MRSG-14-1418-01-CCE to D.H.), and the Andrew Sabin Family Fellowship to D.H. This study was also supported by the National Institutes of Health Cancer Center Support (Grant No. CA016672 to D.L.).

Author Disclosure Statement

The authors have read and understood the journal's policy on the declaration of interests and declare that there is no conflict of interest.