Perioperative Evolution of Sodium Levels in Cirrhotic Patients Undergoing Liver Transplantation: An Observational, Prospective Study

Background & Aims: Hyponatremia is an important predictor of early death among cirrhotic patients in the orthotopic liver transplantation (OLT) waiting list. Evidence exists that prioritizing OLT waiting list according to the MELD score combined with plasma sodium concentration might prevent pretransplantation death. However, the evolution of plasma sodium concentrations during the perioperative period of OLT is not well known. We aimed to describe the evolution of perioperative sodium concentration during OLT and its relation to perioperative neurohormonal responses. Methods: Twenty-seven consecutive cirrhotic patients who underwent OLT were prospectively included in the study over a period of 27 months. We studied the evolution of plasma sodium levels, the hemodynamics, the neurohormonal response and other biological markers during the perioperative period of OLT. Results: All patients with a hyponatremia before OLT were in the Child C group. None of the patients had acute or chronic renal impairment. Interestingly, in patients with hyponatremia, plasmatic sodium reached normal levels as soon as the injured livers were removed during surgery and the plasma sodium concentration remained within normal ranges 1 day, 7 days, as well as 6 months after surgery. Conclusions: Further investigation of rapid correction and stabilization of sodium levels after OLT, as observed in our study, would be of interest in order to fully understand the mecanisms involved in cirrhosis related hyponatremia, its prognostic value and clinical implications. gradient of PO 2 [AaPO , mmHg]; Lactates [mmol/L]; Hematocrit [%], Central Venous Pressure (PVC, mmHg), Mean Pulmonary Arterial Pressure (MPAP, mmHg), Cardiac Index (CI, L/min/m2). p from MANOVA with repeted mesures. p1 represents changes in parametres over time within both groups whereas p2 represents interaction between time and Child.


Introduction
Patients with compensated cirrhosis have an imbalance of sodium excretion. While the glomerular ltration rate remains within normal values, an increased sodium tubular reabsorption decreases urinary sodium excretion. Following a sodium load, kidneys are incapable to excrete additional sodium overload. Activation of the mineralocorticoid receptor by aldosterone [1,2] or cortisol (in patients with impaired 11βhydroxysteroid dehydrogenase activity) [3] would explain mentioned sodium reabsorption. In contrast to patients without end-stage liver disease who can excrete a solutefree water overload, most cirrhotic patients with ascites cannot. The impaired solute-free water excretion is the consequence of increased secretion of arginine vasopressin from the neurohypophysis stimulated by arterial under lling [1,4]. In normal subjects, osmotic stimulus controls arginine vasopressin secretion while in patient with advanced cirrhosis, arterial under lling by decreasing the stretch on arterial baroreceptors, overrides the osmotic stimulus and stimulates the synthesis and release of the hormone. In this group of patients who retain solute-free water in an amount disproportionate to sodium retention, hyponatremia occurs concomitantly to an increase in extracellular uid volume. Additionally, patients with cirrhosis and ascites may develop a rapidly progressive (type 1) or more insidious (type 2) functional renal failure also known as hepatorenal syndrome [5,6].
Low plasma sodium concentration is an important predictor of early death in cirrhotic patients on the waiting list for orthotopic liver transplantation (OLT) [7][8][9][10][11][12] and is also associated with an increased number of complications following OLT regardless of the clinical condition causing hyponatremia [13][14][15]. Whether the prognostic value of hyponatremia is due to the fact that it is a marker of the disease's stage, it's direct negative effects on several organs's function or both is unknown [8]. Kim et al. [16] had shown, through a retrospective analysis of adult candidates registered within the Organ Procurement and Transplantation Network in 2005 and 2006, that prioritizing OLT waiting list according to the MELD score combined with plasma sodium concentration could have prevented deaths within patients in the list.
In addition, Sharma [17] and al. have also shown an increase in survival rate related to hyponatremia among patients with MELD scores > 11 providing more evidence to the importance natremia control in the pre-OLT period.
While it seems clear that hyponatremia is a risk factor for mortality in the pre-OLT period, the evolution of sodium levels and its relation to neurohormonal response during the perioperative period of OLT is unknown. In this observational, prospective study, we monitored plasma sodium concentration and multiple biological markers including the neurohormonal response in the perioperative period of OLT in 27 consecutive patients who underwent OLT in our institution.

Patient selection
Twenty-seven consecutive patients (age 56  years) were prospectively included in the study over a period of 27 months. Sample size correlates to the annual average number of liver trasnplatations in our institution. The ethics committee of our institution (Hôpitaux Universitaires de Genève, Geneva, Switzerland) approved the protocol and written informed consent was obtained from each patient at the time of inclusion in the waiting list.
The study protocol was registered in the ISRCTN register (N° 43452247) with the objective to describe the renal and neurohormonal responses during liver transplantation. In current study we analysed data concerning variations in serum sodium levels. In fact, there is increasing evidence showing the importance of hyponatremia as an independent predictor of mortality after listing for OLT [7,8,12,14,15], and as so, its potential value in prioritizing patients for OLT.

Preoperative evaluation (PRE OLT )
Patients were included at the time they were admitted in the waiting list for OLT. Assessment of liver disease was based on medical history, physical examination, laboratory tests, hemodynamic parameters, liver imaging and histology. The evolution of the disease during the waiting period was assessed every three months. PRE OLT parameters were recorded during the last patient evaluation preceding OLT. Two groups of patients were compared according to the severity of liver disease assessed by the Child-Pugh score: patients with mild or moderate liver dysfunction (child A or B scores, Child AB group) and patients with severe liver dysfunction (Child C score, Child C group).

Anesthetic management and surgery
Anesthesia was induced with continuous infusion of midazolam, sulfentanyl, and atracurium with conformity to our institution's protocol. Each patient has a radial artery catheter for continuous arterial blood pressure monitoring and blood sampling, a triple-lumen central venous catheter (HydrocathA, Ohmeda, Swindon, UK) for uid perfusion and drug administration, and a ow-directed thermodilution pulmonary artery catheter (CCOmbo, Edwards Lifesciences, Irvine, CA, USA) for pulmonary artery pressure, mixed venous oxygen saturation (SvO 2 ), and cardiac output mesuring (Vigilance, Edwards Lifesciences, Irvine, CA, USA). Perioperative normothermia [35.5-37 °C] was maintained with warming blankets. Following resection of the native liver, graft implantation was performed with end-to-end anastomosis. To avoid congestion of splanchnic organs, renal hypertension, and decreased venous return induced by the concomitant cross-clamping of portal vein and inferior vena cava, a bypass was placed according to the hemodynamic tolerance to portal and inferior vena cava clamping. Indications for porto-femoral to axillary vein bypass were a decreased (> 30%) cardiac index or decrease in mixedvenous hemoglobin oxygen saturation below 70% when clamping both the portal vein and the inferior vena cava. The extracorporeal system included a nonocclusive centrifugal pump (model 540 Bio-Medicus, Medtronic, Kerkrade-West, Holland), heparin-coated tubings (Medtronic, Kerkrade-West, Holland) and catheters (Carmeda AB, Töby, Sweden).
Liver grafts that had been preserved at 4 °C in University of Wisconsin solution were ushed after the completion of the lower caval anastomosis with 4% human albumin (ZLB Behring AG, Bern, Switzerland) to obtain potassium concentrations in the out ow perfusate that steadily decreased below 12 mmol/l. After upper caval and portal anastomosis, the graft was revascularized with portal ow and the venovenous bypass terminated. Hepatic artery anastomosis followed by bile duct reconstruction was then performed.
To maintain stable hemodynamic parameters, crystalloids (Ringer lactate containing 6 g/L sodium salts) were infused at a rate of 5 to 10 mL/kg/hour to maintain the following criteria: heart rate < 100 beats per min, mean arterial pressure > 60 mmHg, right atrial pressure between 5 and 10 mmHg and pulmonary artery occlusion pressure between 8 and 14 mmHg. Red blood cells (in saline adenine glucose-mannitol solution) were transfused when serum hemoglobin concentration was ≤ 8 g/dl during surgery and ≤ 9 g/dl in the postoperative period. Fresh-frozen plasma was administered according to coagulation status as documented by thromboelastography and according to surgical clinical assessement of bleeding during surgery. Platelets were transfused during per and post -operative periods when platelets count decreased below 30,000 per mL. No other coagulation factors were administered, and no intraoperative blood salvage was used. All intravenous uids except platelets were administered through a uid warming device.
Hemodynamic measurements and blood collection in the perioperative period Blood samples (20 mL) and urine samples (5 mL) were collected 1 hour after induction of anesthesia (ANE sample), at the end of the anhepatic phase (ANH sample), one hour after graft reperfusion (POST 1h sample), and in the ICU at 24 hours (POST 24h sample) after liver reperfusion. At each time point, systemic and pulmonary hemodynamic parameters and urine ow were measured. The creatinine clearance was calculated according to the Cockcroft-Gault formula. Plasma Cystatin C was also measured in the perioperative period. Sodium exctretion fraction was calculated according to following formula: FE NA = ([urine sodium] x [plasma creatinine]) / ([plasma sodium] x [urine creatinine]). Sodium concentration was measured within plasma, serum and urine using indirect potentiometry. [18] Hormone measurements Blood samples were collected into tubes containing ethylene diamine tetra acetic acid. Tubes were placed on ice and thereafter centrifuged for 15 min at 2000 g (2-8 °C). Plasma samples were stored at -80 °C until analysis.
Plasma renin activity was assessed at 37 °C by direct radioimmunoassay measurement of angiotensin I production from endogenous substrates. In our institution, renin activity in supine position ranges from 0.23 to 1.55 ng/mL/hour. Free catecholamines were measured by liquid chromatography with amperometric detection. Blood (1 mL) was extracted on activated alumina at pH 8.6 and catecholamines were eluted in acidic media before injection into the chromatographic system. The detection limit was 10 pg/injection. Arginine vasopressin was determined by a radioimmunoassay after an extraction procedure using a reversed-phase column extraction according to the manufacturer's instructions (Bühlmann Laboratories AG, Allschwil, Switzerland). Relative a nity of the vasopressin antiserum was given as follows: 100% for arginine vasopressin, 0.25% for lysine vasopressin, and 0.001% for oxitocin and vasotocin. The detection limits were 0.35 pg/mL or 0.32 pmol/L of plasma. Plasmatic NT-proBrain Natriuretic Peptide (pro-BNP) was assessed on Elecsys (Roche, Basel, Switzerland) and DXI (Beckman Coulter, Brea, CA, USA) auto-analysers. NTpro-BNP < 125 pg/mL are considered as normal values.

Postoperative evaluation (POST OLT )
Biological parameters were evaluated in all patients 1 hour, 24 hours and 6 months after OLT (POST OLT ).

Characteristics of patients and surgery
Between August 2010 and November 2015, twenty-seven patients consisting of 22 males and 5 females who underwent OLT were prospectively included at our institution. Characteristics of patients are summerized in Table 1. The etiologies of cirrhosis were: alcohol abuse (n = 9), hepatitis (n = 13), biliary (n = 2), and others (n = 3). Two groups of patients were compared: Child AB and and Child C groups. One patient obtained a living donor, and two had split livers. The remaining patients received whole cadaveric liver graft, including one for retransplantation. Among all patients, ten had venovenous bypass ( ve in each group). Before surgery, the time spent on the waiting list was similar in Child AB and Child C groups. During surgery, the anhepatic phase and cold ischemia did not differ between the two groups. Red blood cells and platelet requirements were similar in Child AB and Child C groups, but the Child C group received more fresh frozen plasma units.
One patient died during the study period in each group. The time spent in ICU was similar in the two groups.
Plasma sodium concentration and OLT Hyponatremia was de ned by a sodium levels < 135 mmol/l as per standard de ntion [19].
All patients with Pre OLT hyponatremia were in the Child C group and the preoperative plasma sodium concentration was signi cantly lower in Child C than in Child AB groups ( Table 2 and Fig. 1A, p < 0.02). In Child C patients, four out of thirteen (30%) had a plasma sodium concentration of less than 135 mEq/l.
Interestingly, plasma sodium concentrations varied during the observation period and hyponatremia occurred concomitantly to complications (digestive hemorrhage, bacterial peritonitis and decompensated ascites) as illustrated in Fig. 2. During the perioperative surgical period, the sodium concentration did not change signi cantly in Child AB patients while within Child C sodium levels reached normal levels during the anhepatic phase (ANH) (p = 0.02, Fig. 1B). To be noted, correction of hyponatremia in the Child C group was not associated with adverse outcomes.
Speci cally, no central pontine myelinolysis was observed. Plasma sodium concentrations remained in normal range and similar in both groups seven days (p = 0.63, Fig. 1C) and six months (POST OLT , p = 0.56, Fig. 1D) after OLT. Four patients in the Child C group remained with a plasma sodium concentration lower than 135 mEq/L at day seven after OLT. The latter was not associated with worst outcomes in the observation period. Sodium levels normalized after in three out of these four patients. FE NA of Child C patients did not signi cantly increase from the PRE OLT to the POST OLT periods (

Systemic hemodynamic and biological parameters in the perioperative period
Pulmonary and systemic hemodynamics were measured after induction of anesthesia (ANE), during the ANH, 1 hour (POST 1h ) and 24 hours (POST 24h ) after graft reperfusion. There was no difference between the two patient groups for mean arterial pressure (p = 0.89), heart rate (p = 0.81), and cardiac index (p = 0.14) (Fig. 3). In contrast, Child AB patients had higher systemic vascular resistance values than Child C during the ANE period but, in the POST 1h and POST 24h periods, both groups had similar values (Fig. 3).
Pulmonary vascular resistance index was signi calty lower in the Child C group (p = 0.02) while mean pulmonary arterial pressures did not differ between groups during the observed period ( Table 3). The alveolo-arterial gradient of PO 2 (AaDO 2 , mmHg) transiently increased in the Child C group during the ANE and POST 1h periods but both groups had similar AaPO 2 24 h after graft reperfusion. Plasma lactate concentrations similarly increased in both groups in the ANE and POST 1h periods while arterial pH steadily decreased over time in both groups without any signi cant difference between groups. Hematocrit, which was lower in Child C group during the ANE period comparing to the Child AB , similarly decreased in both groups during the observed period (Table 3).

Renal and Neurohormonal responses during surgery and 24 h after graft reperfusion
During the perioperative period, plasma concentrations of creatinine (p = 0.11, Table 2) and the creatinine clearance (p = 0.64) did not differ between both groups, while cystatin C concentrations signi cantly but moderately increased in the Child C group at POST 24h . (P < 0.005, Table 2). Urine ow was higher during the POST 1h period in Child AB patients (p = 0.01).
The plasma renin activity and concentrations of epinephrine, vasopressin, and brain natriuretic peptide increased during the anhepatic period and returned to normal values in the consecutive periods for most patients (Table 4).

Discussion
Description of the perioperative evolution of sodium levels and neurohormonal response in patients with cirrhotic liver disease undergoing OLT was the main objective of our study. All patients with hyponatremia were in the Child C group. Interestingly, low plasma sodium concentration was corrected rapidly after cirrhotic liver was removed and the plasma sodium concentration remained within normal ranges 1 day, as well as 6 months after surgery. This correction was not associated with adverse outcomes within study's observation period. Speci cally, no central pontine myelinolysis was observed.
Hyponatremia in cirrhosis is primarily the result of solute-free water retention exceeding that of sodium [20]. The proposed mechanism is splanchnic arterial vasodilatation leading to reduced systemic vascular resistance and the release of various neurohormones. An arterial under lling is sensed by baroreceptors and activates a compensatory neurohumoral response through the sympathetic nervous system, the renin-angiotensin-aldosterone system, as well as the release of arginine vasopressin. The net result of this activation is solute-free water retention, intense renal sodium retention, and hypervolemic hyponatremia. The degree of activation of these neurohumoral mechanisms correlates directly to the degree of portal hypertension [21]. Serum sodium concentration is therefore an important parameter to be considered before OLT as it might re ect the severity of the disease.
Several studies emphasize the importance of sodium levels in the assessement fot OLT priority within potential candidates. Hyponatremia < 130 mmol/L was shown to be an independent predictor of mortality after listing, as lower sodium levels at inclusion in the waiting list were associated with 3-and 12-month decreased survival [7]. A clear cutoff for the association between hyponatremia and survival after listing in not yet well de ned and differ between studies: < 135 mmol/L [22], < 130 mmol/L [23], or 126 mmol/L [24] respectively. The mortality risk might also be inversely correlated to the plasma sodium concentrations [16,25]. Consequently, several MELD-derived scores that incorporate plasma sodium concentration proposed to include hyponatremia as a priority criterion for OLT. The prevalence of hyponatremia in OLT candidates varies from 11-34%. It is 21,6% (cutoff at 130 mmol/L) or 49.4% (cutoff at 135 mmol/L) in the overall population of cirrhotic patients. In our study, hyponatremia was present in four patients (30%, cutoff at 135 mmol/L). None of the patients had a sodium concentration lower than 120 mEq/L.
The predictive effect of sodium concenratation was shown to be greater within patients with low MELD SCORE. Thus, in the same way that serial MELD scores predict mortality better than a single value, serial measurements of plasma sodium could have a better predictive value than a single measurement [16].
While awaiting OLT, hyponatremia is likely to uctuate over time concomitantly with the complications of cirrhosis and the patient treatment [26,27]. This uctuation was reproduced in our study (see Fig. 2). The number of acute aggravations and the ability to recover would probably be a factor of interest at the time of OLT listing and should be futher studied. An increased risk of death is associated with a serum sodium concentration < 126 mmol/L (6.5fold increase) or < 131 mmol/L (5-fold increase) measured at any time during the waiting period for OLT [25]. In our study, a serum sodium concentration < 131 mmol/L at the time of listing was not associated with an increased risk of death at 6 monthes following OLT. The fact that sodium levels uctuate before OLT and that hyponatremia seem to have predictive effect for postOLT increased mortality might strengthen the importance of studying the effect of serial sodium concentration measuring before OLT as they could provide a fuller picture of postOLT mortality risk.
In addition, Hyponatremia prior to OLT was associated with an increased 90-day postoperative mortality [14]. Similarly, in patients with cirrhosis, hyponatremia at time of OLT (< 130 mmol/L) was shown to increase the risk of postoperative neurologic disorders, infectious complications, and renal failure during the rst month after OLT and reduced the 3-month survival rate [13]. However, the postOLT outcome highly depends on donor characteristics, cardiovascular evolution during liver replacement, and surgical complications. In our study, sodium levels were signi cantly lower in patients with severe liver failure before OLT. Sodium correction took place during OLT before the surgical anhepatic phase (Fig. 1), reaching similar normal values in both groups. None of the patients experienced neurological complications due to this rapid correction of sodium concentration.
Correction of hyponatremia following OLT and its outcomes seems to be a matter of debate. Few studies have focused on the uctuations of sodium levels during and after OLT. Both, Romanovski et al. [28] and Hudcova et al. [29], have shown that rapid correction of natremia after OLT in hyponatremic patients was associated with worse outcomes and complications in the post operative period of OLT. Similarly, Park et al. [30] has found that major shift of natremia during OLT with liver donor was a risk factor for prolonged mecanical ventilation post transplantation. In contrast, the study from Brandmann et al. [31] paying particular attention to the consequences of uctuations in sodium levels in patient with hepatic encephalopathy before OLT, showed that there was no in uence on neurological outcome or length of stay in ICU. Leise and al [32] have even found that hyponatremia was not associated with worse outcomes unlike hypernatremia before OLT. In our study, natremia correction was not associated with negative outcomes and the time spent in ICU after OLT was not signi cantly longer in patients with hyponatremia.
Another interesting observation of our study is that hyponatremia coexists with most complications of cirrhosis as previously reviewed by Schrier el al. [1,4] (Fig. 4). Activation of arterial baroreceptors occurs with low cardiac output (cirrhotic cardiomyopathy or decreased preload) as well as high cardiac output (septic-like syndrome/in ammation) and exacerbation of splanchnic vasodilatation with increased portal pressure [1,4,33]. This unifying hypothesis further reinforces the usefulness of hyponatremia as a marker of disease severity. When serum sodium concentrations were measured over time in a single patient (Fig. 2), hyponatremia was associated with decreased preload (digestive hemorrhage) or sepsis/in ammation (bacterial peritonitis). Recurrent ascites with hyponatremia frequently remains of unknown origin. Thus, besides rare neurological disorders induced by rapid modi cations of serum sodium concentration, hyponatremia is likely to be a surrogate marker of either decreased preload, aggravation of cardiomyopathy, aggravation of liver disease (increased splanchnic vasodilatation), or occurrence of a septic-like syndrome through arterial under lling. Guidelines have been published to treat episodes of hyponatremia in cirrhotic patients [34]. However, treatment of acute complications of cirrhosis and OLT are likely to be the best therapies when hyponatremia occurs [35].
Finally, the exact mechanism of hyponatremia correction during OLT is yet to be elucidated. In our study, normalization of serum sodium concentrations within the surgical period was unrelated to the systemic vascular resistance that remain unchanged in Child C patients during the perioperative period. Although, portal blood ow immediately increases with graft placement, correction of the splanchnic arterial vasodilatation is usually observed only within 3 to 6 months after OLT [36]. The systemic vascular resistance is unlikely to be re ected by the stimulation of arterial baroreceptors by arterial under lling as previously shown by Rector et al. [37]. RW Schrier [1,4] emphasizes that arterial under lling should be assessed by the release of neurohormones. In the ANH phase of OLT, serum renin activity and concentrations of epinephrine, vasopressin, and pro-BNP increased similarly in both Child AB and Child C patients (Table 4) as previously described [38]. However, the unstable peroperative conditions associated with a high neurohormonal response precludes correlating the normalization of serum sodium concentration to a decrease hormone release. Eyraud et al. have already shown in 2002 [39] that, in patients with normal liver function who had major hepatic resection with portal triad clamping and occlusion of the inferior vena cava below and above the liver, the sudden drop in venous return enhances the neurohormonal response to sustain the systemic hemodynamics. After unclamping, endogen catecholamines, vasopressin, renin and atrial natriuretic peptide return to baseline values within ve minutes. These observations suggest that this acute neurohmoral response, which is similarly observed in our study, do not solely explain the progressive normalization of sodium levels observed after OLT.
Interstignly, Cystatin-C values were moderately but signi caly higher in the Child C group 24 hours after OLT. This increase is consistent with recent studies showing that Cystatin-C is a more sensitive indicator for renal impairment in cirrothic patients than serum creatinine and creatinine clearance [40]. Moreover, higher values of Cystatin-C were correlated with a more severe cirrothic disease [41] and has been shown to be an independent predictor of post OLT survival [42]. Thus, Cystatin-C measurement might be of interest for early detection of kidney disease and as so, help prevent renal related complications in cirrothic patients. In addition, the sensitivity of Cystatin-C for early detection of kidney disease in cirrothic patients, its correlation with the severity of the disease and its prediction value for post OLT mortalilty could place the cystatin-C as an interesting factor in priotizing patients in the OLT waiting list and should be further studied.
Our study has signi cant limitations. The number of patients was limited and the number of patients with a very low natremia (only 3 ou of 27 patients had sodium levels < 130 mmol/l) even smaller, the sample size however, correlates to our institution's annual average liver trasnplatations. Furthermore, the choice of using the Child-Pugh classi cation to assess severity of cirrhosis could be a matter of debate. Since the MELD score was adopted in 2002 to allocate priority to candidates, the mortality on the waiting list has decreased. The MELD score predicts the survival for patients with cirrhosis more acurately than the Child score. Interestingly, in patients with hyponatremia, the MELD-Na score might further decrease the 90-days mortality on the waiting list [14]. However, to assess the severity of cirrhosis in our study, we chose the Child-Pugh score that re ects clinical symptoms (encephalopathy and ascites) better than the MELD score, which is based on the logarithmic transformation of bilirubin, creatinine, and INR.

Conclusions
In summary, hyponatremia at listing is now recognized as an independent factor of mortality within 90 days after OLT. Moreover, in OLT candidates with cirrhosis, episodes of hyponatremia are likely to be associated with disease complications and the importance of serial determination should be investigated. A score including sodium concentration variations may be an interesting parameter to consider before OLT. Interestingly, in our study, hyponatremia was rapidly corrected after successful liver transplantation without observed complications. Short-and long-term clinical outcomes and prognostic value of this correction in the post-operative period remains to be better understood. Finally, the fact that sodium levels remain stable after post-operative correction despite a persistant splanchnic vasodilatation suggests that probably other mecanisms are implicated in hyponatremia associated with cirrhosis and futher studies on the subject would be of interest.

Declarations Ethics Statement
The ethics committee of our institution (Hôpitaux Universitaires de Genève, Geneva, Switzerland) approved the protocol and written informed consent was obtained from each patient at the time of inclusion in the waiting list.

Consent for publication
All authors have approved the manuscript and agree with its submission.

Competing interests
None declared.

Funding
The project was funded by departemental funds of the anesthesiology service of the University Hospital of Geneva, Switzerland.

Authors' Contributions
Dr. Ido Zamberg -Wrote the manuscript, took part in the analysis of results and literature review.
Prof. Eduardo Schiffer -Team leader and main investigator, wrote the manuscript, performed statistical analysis.
Dr. Julien Maillard -Participated in writing rst darft, data collection and criticaly revised the manuscript.
Dr. Simon Tomala -Participated in writing rst darft and criticaly revised the manuscript.
Dr. Benjamin Assouline -Criticaly revised the manuscript and took part in statistical anaylisis.
Dr. Thomas Mavrakanas -Criticaly revised the manuscript and performed statistical analysis.
Dr. Axel Andres -Participated in patients' inclusion, criticaly revised the manuscript.
Dr. FlorenceAldenkortt -Participated in patients' inclusion, criticaly revised the manuscript.  Serial determination of plasma sodium concentrations in a single patient from the ChildC group. The PREOLT value of plasma sodium concentration was 136 mmol/L (arrow).