Hypernatremia predicts poor survival in patients with terminal cancer: a retrospective cohort study

Background: Although palliative care providers, patients, and their familes rely heavily on accurate prognostication, the prognostic value of electrolyte imbalance has received little attention. Methods: As a retrospective review, we screened inpatients with terminal cancer admitted between January 2017 and May 2019 to one hospice-palliative care unit. Clinical characteristics and laboratory results were obtained from medical records for multivariable Cox regression analysis of independant prognostic factors. Results: Of the 487 patients who quali�ed, 15 (3%) were hypernatremic upon admission. Median survival time was 26 days. Parameters associated with shortened survival included male sex, advanced age (> 70 years), lung cancer, poor performance status, elevated in�ammatory markers, azotemia, impaired liver function, and hypernatermia. In a multivariable Cox proportional hazards model, male sex (hazard ratio [HR]=1.53, 95% con�dence interval [CI]: 1.15–2.04), poor performance status (HR=1.45, 95% CI: 1.09– 1.94), leukocytosis (HR=1.98, 95% CI: 1.47–2.66), hypoalbuminemia (HR=2.06, 95% CI: 1.49–2.73) and hypernatremia (HR=1.55, 95% CI: 1.18–2.03) emerged as signi�cant predictors of poor prognosis. Conclusion: Hypernatremia may be a useful gauge of prognosis in patients with terminal cancer. Further corroborative studies of large scale and prospective design are needed.


Background
Accurately predicting the prognosis of patients with terminal cancer is essential, because it aids in clinical decisions and management plans of palliative care workers.Over the past few decades, a number of studies have been conducted in this setting to identify patient-related prognostic factors of potential use, such as performance status, anorexia-cachexia, delirium, and dyspnea [1]; and various laboratory abnormalities have similarly been identi ed.Serum markers of systemic in ammation (i.e., leukocytosis, lymphopenia, C-reactive protein (CRP) elevation, and in ammatory cytokines) are well-known indices of poor survival [2].Biomarkers of hepatic dysfunction (elevated lactate dehydrogenase [LDH], prolonged international normalized ratio [INR], hypoalbuminemia) and renal impairment (serum urea, creatinine, and uric acid elevations) have been implicated as well [2,3].
Electrolyte abnormalities are common among the terminal cancer patients and represent yet another signi cant means of predicting survival [4].In an earlier study, 79% of such patients showed at least one electrolyte abnormality upon referral for palliative care [5].Hyponatremia is the most frequent electrolyte disorder of patients with terminal cancer, having dour consequences [6].In fact, its early detection and proper correction may actually prolong median survival time [7].Although hypernatremia is otherwise a rarity in these patients, offering little opportunity for study, there is evidence to suggest that the prognostic rami cations again are negative [8,9].Potassium imbalance is also a highly prevalent electrolyte disorder.Especially in cancer patients, hypokalemia often presents in conjunction with hyponatremia and hypomagnesemia [4].According to some, hyperkalemia is a prognostically unfavorable determinant, but this view remains controversial [10,11].Research on the prognostic utility of other electrolyte disorders including hypercalcemia and hypermagnesemia, has likewise proved inconclusive [5,12].
The present study was undertaken to assess the prevalence of electrolyte imbalance in terminally ill cancer patients, investigating the potential prognostic signi cance.

Methods
As a retrospective review, we examined medical records of 515 patients admitted to the palliative care unit of Incheon regional cancer center between January 2017 and May 2019.All participants were terminally ill with cancer and were not expected to survive beyond 6 months.Ultimately, 10 patients who lacked serum electrolyte data and 18 patients transferred from other medical institutions were excluded, leaving 487 patients eligible for nal analysis.The institutional review board of Gachon University Gil Medical Center approved this study (GCIRB2019-149), waiving the need for informed patient consent as designed.
Electronic medical records provided the following patient data: age, sex, primary cancer site, Eastern Cooperative Oncology Group Perfomance Status (ECOG-PS), evidence of active infection, survival times, and laboratory diagnostic results.ECOG-PS is a scale (0-4) used to assess physical ability.On day of admission, an experienced member of the palliative care team, including physicians and registered nurses, scored each patient functionally.Survival time was de ned as the period from admission-day blood testing (electrolytes included) to day of death.Laboratory testing on day of admission included total white blood cell (WBC) count with differential, hemoglobin, platelet count, serum creatinine, serum albumin, total bilirubin, international normalized ratio (INR), CRP, serum sodium, and potassium.

Statistical analysis
Descriptive data were expressed as medians and numbers of subjects.Kaplan-Meier method and log-rank test were invoked to measure differences in survival times across all patient characteristics, using Cox proportional hazards models to identify independent predictors of survival in univariable and multivariable analyses.Variables showing signi cance (p<0.05) in univariable analysis served for nal multivariable analysis.A two-tailed p<0.05 was deemed signi cant.All computations were drived by standard software (Stata SE v9.2; StataCorp, College Station, TX, USA).

Results
A total of 487 patients (men: 268, 55%; women: 219, 45%) with terminal cancer quali ed for study participation.Baseline clinical characteristics of the study population are shown in Table 1.Median age was 70 years.The most common type of malignancy was cancer of the gastrointestinal tract (127 patients, 26.1%), followed by hepatopancreatobiliary cancer (118 patients, 24.2%).ECOG-PS scores 3 and 4 were recorded in 38.8% and 37.6% of patients, respectively.Median values of abnormal laboratory parameters were as follows: hemoglobin 10.1 g/dL, CRP 5.2 mg/dL, and sodium 134 mEq/L.The median survival time overall was 26 days.
Survival time is shown according to clinical and laboratory characteristics in Table 2. Advanced age (>70 years), male sex, lung cancer, poor performance status, leukocytosis, neutrophilia, lymphopenia, thrombocytopenia, azotemia, hypoalbuminemia, hyperbilirubinemia, prolonged prothrombin time (PT)-INR, elevated CRP, and hypernatremia were factors associated with signi cantly shorter median survival time.
Kaplan-Meier survival curves plotted by serum sodium level are depicted in Figure 1.In patients with hypernatremia (vs.eunatremia or hyponatremia), survival was signi cantly shorter (p<0.001),whereas survival in eunatremic and hyponatremic patients appeared similar.

Discussion
Findings of the present study indicate that in terminally ill cancer patients with hypernatremia (vs.eunatremia or hyponatremia), the prognosis is demonstrably poor.These results are aligned with outcomes of previous studies, offering added support.Based on a group of 259 cancer patients referred for palliative care, Alsirafy et al. encountered shorter median survival (8 days) and higher mortality (68%) in those with hypernatremia than in hyponatremic or eunatremic counterparts [9].However, multivariable analysis of well-known prognostic factors in terminal cancer patients was not done.Our data indicate that the association between hypernatremia and poor survival remains robust after controlling for other predictors of survival.
Salahudeen et al. have also reported poor clinical outcomes involving higher mortality, longer hospitalization, and greater hospital expense in patients with hypernatremia [8].There were some differences from our cohort of terminal patients whose life expectancies were roughly 6 months.They examined subjects admitted to a comprehensive cancer center with any stage of disease; and their focus was on iatrogenic hypernatremia, because baseline hypernatremia contributed so few patients.In our investigation, laboratory testing took place on day of admission, aimed at existing rather than acquired hypernatremia.Hence, this is perhaps the rst effort to explore the prognostic utility of spontaneous hypernatremia in terminally ill cancer patients.
The prevalence of hypernatremia in patients with terminal cancer is unclear.Salahudeen and colleagues found that hypernatremia in cancer patients increased from 0.2% on admission to 2.6% during the course of hospitalization [8].Another study has also indicated that 8.5% of adult cancer patients referred for palliative care are hypernatremic [9].Similar to prior studies, we recorded a 3% prevalence of hypernatremia.
Little is known of the speci c mechanism by which hypernatremia worsens survival, but there is at least one plausible explanation.Hypernatremia is generally induced by the loss of electrolyte-free water.Physiologic feedback mechanisms, such as thirst and antidiuretic hormone (ADH) release, are then promptly activated to increase water intake and minimizing additional free water loss.In a healthy population, elevated serum sodium levels return to normal range accordingly [11].However, feedback dysfunction in patients with terminal cancer may impede or prevent normalization of serum sodium concentrations, and many patients with terminal cancer are faced with non-replenishment of water lost through excessive sweating, vomiting, diarrhea, and nasogastric drainage [13].Impaired response to thirst due to diminished mental faculties or poor oral intake and subsequent dehydration may induce hypernatremia in such patients outside hospital environments.Still, there is virtually no research on hypernatremia in cancer patients.Retrospective studies of older adult patients and critically ill patients admitted to intensive care units would be helpful to understand this problem in the context of terminal cancer [14,15].Mental debilitation and poor oral intake create rapid declines in their general conditions [16].Although causality between hypernatremia and deteriorating general conditions remain in question, hypernatremic patients are extremely ill and less inclined to survive.
Certain prognostic factors, namely poor functional status, leukocytosis, and hypoalbuminemia, are well documented in past reports, but the data on effects of hyponatremia have been inconsistent.Several earlier endeavors have shown the negative prognostic aspect of hyponatremia in a variety of cancers [17,18].Yoon et al. have also demonstrated a relation between hyponatremia and shorter survival time in terminally ill cancer patients [19].However, another study has failed to support this relation in Korean patients with terminal cancer entering a hospice unit, although hampered by a relatively short median survival time (9.5 days) [20].To our knowledge, the present analysis offers the most fully controlled results in a comparable setting, adjusted for potential confounders.Also, our subjects survived longer than those in previous studies.Even so, we did not nd hyponatremia predictive of poor survival under these conditions.Further prospective studies to explore the prognostic signi cance of hyponatremia are nevertheless warranted.
At present, evidence of the prognostic value of potassium disorders is sparse.Cui et al. have observed an association between serum potassium level and survival time, but signi cance was not reached in multivariable analysis [21].In our study, hyperkalemia similarly did not emerge from multivariable analysis as a signi cant predictor of poor prognosis.One retrospective study conducted in Taiwan cites a serum potassium level > 5 mg/dL as an objective index of short-term survival (7 days) in patients with advanced cancer [10].Still other researchers have found no prognostic signi cance attached to potassium imbalance [2,20].
There are acknowledged limitations to interpreting the results of our study.First, the cohort is representative of a single center only, and the number of hypernatremic patients was small.Given the prevalence of hypernatremia in terminal cancer patients, large-scale multicenter investigations would be helpful to determine the actual prognostic import of hypernatremia under such circumstances.Another drawback is the lack of sequential or interventional data on serum sodium levels.A previous study report does indicate that hyponatremia normalization is prognostically bene cial in patients with advanced nonsmall cell lung cancers [22].One may thus infer that without normalization, patients with persistent hypernatremia will fare worse.

Conclusion
We have shown that hypernatremia on admission for palliative care of terminal cancer is predictive of shorter patient survival.Despite its low prevalence (3%), greater clinical attention to the prognostic utility of hypernatremia is needed.

Declarations
-Ethics approval and consent to participate: The institutional review board of Gachon University Gil Medical Center approved this study (GCIRB2019-149), waiving the need for informed patient consent as designed.
Consent for publication: Not applicable.
-Availability of data and materials: The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.
-Competing interests: The authors declare that they have no competing interests.
-Funding: This work was supported by the Gachon University Gil Medical Center (Grant number: FRD2019-11) -Authors' contributions: All authors have participated su ciently in this work to take public responsibility for appropriate portions of the content and have read and approved the nal manuscript.Speci c author contributions were: MSS: analysis of data, interpretation, drafting the manuscript, and nal revision.ICH: study design, analysis of data, interpretation, and critical revision of the manuscript.JHJ: study conception, design, interpretation, and revision the statistical session.HHL: study design, acquisition of data, and analysis of data.JHC: study conception, acquisition, interpretation, and drafting sections of the manuscript.JYS: interpretation, critical revision the discussion session.

Table 2 .
Survival time in relation to patient characteristics a Median value applied b Log-rank test ECOG-PS, Eastern Cooperative Oncology Group performance status; PT/INR, prothrombin time/international normalized ratio; CI, confidence interval

Table 3 .
Independent prognostic indices of survival (Cox proportional hazards model) a Based on variables of significance (p<0.05) in univariable analysis PT/INR, prothrombin time/international normalized ratio; CRP, C-reactive protein Figures Page 12/12