Effects of dilutional hyponatremia on brain organic osmolytes and water content in patients with cirrhosis

Authors

  • Tea Restuccia,

    1. Liver Unit, Hospital Clínic, Barcelona, Spain
    2. Institut d'Investigacions Biomèdiques August Pi-Sunyer (IDIBAPS), Barcelona, Spain
    3. Instituto Reina Sofía de Investigación Nefrológica, Barcelona, Catalunya, Spain
    Search for more papers by this author
  • Beatriz Gómez-Ansón,

    1. Department of Radiology, Hospital Clínic, Barcelona, Spain
    2. Institut d'Investigacions Biomèdiques August Pi-Sunyer (IDIBAPS), Barcelona, Spain
    Search for more papers by this author
  • Mónica Guevara,

    1. Liver Unit, Hospital Clínic, Barcelona, Spain
    2. Institut d'Investigacions Biomèdiques August Pi-Sunyer (IDIBAPS), Barcelona, Spain
    3. Instituto Reina Sofía de Investigación Nefrológica, Barcelona, Catalunya, Spain
    Search for more papers by this author
  • Carlo Alessandria,

    1. Liver Unit, Hospital Clínic, Barcelona, Spain
    2. Institut d'Investigacions Biomèdiques August Pi-Sunyer (IDIBAPS), Barcelona, Spain
    3. Instituto Reina Sofía de Investigación Nefrológica, Barcelona, Catalunya, Spain
    Search for more papers by this author
  • Aldo Torre,

    1. Liver Unit, Hospital Clínic, Barcelona, Spain
    2. Institut d'Investigacions Biomèdiques August Pi-Sunyer (IDIBAPS), Barcelona, Spain
    3. Instituto Reina Sofía de Investigación Nefrológica, Barcelona, Catalunya, Spain
    Search for more papers by this author
  • M. Elena Alayrach,

    1. Institut d'Investigacions Biomèdiques August Pi-Sunyer (IDIBAPS), Barcelona, Spain
    Search for more papers by this author
  • Carlos Terra,

    1. Liver Unit, Hospital Clínic, Barcelona, Spain
    2. Institut d'Investigacions Biomèdiques August Pi-Sunyer (IDIBAPS), Barcelona, Spain
    3. Instituto Reina Sofía de Investigación Nefrológica, Barcelona, Catalunya, Spain
    Search for more papers by this author
  • Marta Martín,

    1. Liver Unit, Hospital Clínic, Barcelona, Spain
    2. Institut d'Investigacions Biomèdiques August Pi-Sunyer (IDIBAPS), Barcelona, Spain
    3. Instituto Reina Sofía de Investigación Nefrológica, Barcelona, Catalunya, Spain
    Search for more papers by this author
  • Magda Castellví,

    1. Department of Neurology, Hospital Clínic, Barcelona, Spain
    Search for more papers by this author
  • Lorena Rami,

    1. Department of Neurology, Hospital Clínic, Barcelona, Spain
    Search for more papers by this author
  • Aitor Sainz,

    1. Institut d'Investigacions Biomèdiques August Pi-Sunyer (IDIBAPS), Barcelona, Spain
    Search for more papers by this author
  • Pere Ginès,

    Corresponding author
    1. Liver Unit, Hospital Clínic, Barcelona, Spain
    2. Institut d'Investigacions Biomèdiques August Pi-Sunyer (IDIBAPS), Barcelona, Spain
    3. Instituto Reina Sofía de Investigación Nefrológica, Barcelona, Catalunya, Spain
    • Liver Unit, Hospital Clínic, Villarroel, 170, 08036 Barcelona, Spain
    Search for more papers by this author
    • fax: 34-73-451-55-22

  • Vicente Arroyo

    1. Liver Unit, Hospital Clínic, Barcelona, Spain
    2. Institut d'Investigacions Biomèdiques August Pi-Sunyer (IDIBAPS), Barcelona, Spain
    3. Instituto Reina Sofía de Investigación Nefrológica, Barcelona, Catalunya, Spain
    Search for more papers by this author

Abstract

In advanced cirrhosis there is a reduction in the brain concentration of many organic osmolytes, particularly myo-inositol (MI). Hyponatremia could theoretically aggravate these changes as a result of hypo-osmolality of the extracellular fluid. The aim of this study was to determine the effects of hyponatremia on brain organic osmolytes and brain water content in cirrhosis. Brain organic osmolytes, measured by 1H–magnetic resonance spectroscopy, and brain water content, as estimated by magnetization transfer ratio (MTR) and measurement of brain volume were determined in 14 patients with dilutional hyponatremia, 10 patients without hyponatremia, and eight healthy subjects. Patients with hyponatremia had remarkable lower levels of MI compared with values in nonhyponatremic patients and healthy subjects. Brain MI levels correlated directly with serum sodium and osmolality. Serum sodium was the only independent predictor of low brain MI levels. Serum sodium also correlated directly with other brain organic osmolytes, such as choline-containing compounds, creatine/phosphocreatine, and N-acetyl-aspartate. By contrast, brain glutamine/glutamate levels were higher in patients with cirrhosis compared with values in healthy subjects and correlated with plasma ammonia levels but not with serum sodium or osmolality. No significant differences were found in MTR values and cerebral volumes between patients with and without hyponatremia. In conclusion, dilutional hyponatremia in cirrhosis is associated with remarkable reductions in brain organic osmolytes that probably reflect compensatory osmoregulatory mechanisms against cell swelling triggered by a combination of high intracellular glutamine and low extracellular osmolality. These findings may be relevant to the pathogenesis of encephalopathy in hyponatremic patients. (HEPATOLOGY 2004;39:1613-1622.)

Several lines of evidence indicate that hepatic encephalopathy is related to abnormalities in astrocyte function, which may subsequently result in alterations of neuronal function.1–4 This astrocyte dysfunction appears to be the consequence, at least in part, of the intracellular accumulation of glutamine secondary to hyperammonemia, because ammonia can only be eliminated from the brain through the conversion of glutamate to glutamine by the enzyme glutamine synthetase, which is located within the astrocytes. Because of glutamine's osmotic effect, intracellular glutamine levels cause the astrocyte to swell and activate homeostatic mechanisms to prevent/reduce brain edema. These mechanisms include the reduction in the intracellular content of several osmotically active substances known as organic osmolytes, mainly myo-inositol (MI). Either the astrocyte swelling, the cell changes that take place to prevent this swelling, or a combination thereof causes astrocyte dysfunction.1–4 Investigations in experimental models and in patients with cirrhosis using 1H–magnetic resonance spectroscopy (1H-MRS) have consistently revealed a characteristic pattern of increased brain glutamine as well as reduction of MI and other organic osmolytes.5–9 This pattern is more marked in patients with overt hepatic encephalopathy but also occurs in patients without encephalopathy.9, 10

Advanced cirrhosis is frequently complicated with hyponatremia, a condition characterized by a reduction in the osmolality of the extracelullar fluid.11 It has been hypothesized that hyponatremia in cirrhosis can contribute to hepatic encephalopathy by causing a further impairment in astrocyte function.3 However, no studies have been reported in this setting. In patients with hyponatremia but without liver disease and in animal models of hyponatremia, the brain concentration of organic osmolytes, particularly MI, is markedly reduced, probably owing to a homeostatic mechanism to prevent cell swelling due to the passage of fluid from the extracellular to the intracellular space.12–16 On the other hand, in an experimental model of acute liver failure, hyponatremia exacerbates brain edema.17 Therefore, it is possible that hyponatremia could aggravate astrocyte dysfunction in cirrhosis. The current study was undertaken to investigate the effects of hyponatremia on brain concentration of organic osmolytes, as evaluated by 1H-MRS, in patients with cirrhosis with and without hyponatremia. To assess the existence of brain edema, brain water content was estimated by measuring the magnetization transfer ratio (MTR) and cerebral volume.

Abbreviations

MI, myo-inositol; MTR, magnetization transfer ratio; 1H-MRS, 1H–magnetic resonance spectroscopy; MR, magnetic resonance; MRI, magnetic resonance imaging; glutamate/glutamine, Glu; Cho, choline-containing compounds; NAA, N-acetyl-aspartate; Cre, creatine/phosphocreatine.

Patients and Methods

Patients.

Twenty-four patients with cirrhosis seen at the Liver Unit of the Hospital Clínic of Barcelona from April 2002 to May 2003 were enrolled. A control group of eight healthy subjects with similar age was also studied. The study was approved by the Institutional Review Board of the Hospital Clínic and patients gave written informed consent to participate. Exclusion criteria were:

  • 1age lower than 18 or greater than 75;
  • 2active alcoholism during the 3 months prior to the study;
  • 3hepatic encephalopathy grades II to IV;
  • 4neurological or psychiatric disease;
  • 5treatment with psychotropic drugs or other drugs known to alter neuropsychological function;
  • 6gastrointestinal bleeding or infection within 1 week before the study;
  • 7previous treatment with transjugular introhepatic portal systemic shunts (TIPS) or surgical portosystemic shunts;
  • 8severe extrahepatic disease; and
  • 9contraindication for a magnetic resonance (MR) examination.

Patients with true hyponatremia (low serum sodium levels associated with signs of dehydration in the absence of ascites and edema) were excluded from the study.

Study Design.

The investigational protocol consisted of 1H-MRS and MR imaging (MRI) to determine: (1) the intracerebral concentration of MI, glutamate/glutamine (Glu), choline-containing compounds (Cho), N-acetyl-aspartate (NAA), and creatine/phosphocreatine (Cre); (2) the brain water content, as estimated using MTR; and (3) the normalized total cerebral volume and segmented grey and white matter volumes. In addition, blood samples were obtained to determine liver and renal tests, serum and urine osmolality, serum and urine electrolytes, plasma ammonia concentration, plasma renin activity, and the plasma levels of vasopressin, aldosterone, and norepinephrine.18 To exclude any possible effect of diuretics on brain organic osmolytes and water content, all patients were studied under identical conditions of low sodium diet and without diuretic therapy for at least 5 days.

MRI and Spectroscopy: Acquisition and Postprocessing.

MR studies were obtained using a GE 1.5T MR-Unit (GE Medical Systems, Milwaukee, WI) and the head coil. All studies included oblique (parallel to the corpus callosum) FLAIR (fluid attenuated inversion recovery) and T2 fast spin echo images, to exclude pathology. Oblique T1-weighted spin echo (repetition time = 600, echo time = 9, field of view 24, thickness 5, spacing 1.5, matrix 256 × 256, 1 number of excitations) images with and without transverse magnetization were then obtained in exactly the same locations as the inversion recovery images. These two sets of images were postprocessed using GE software (Functool 2000, GE Medical Systems), and pixel-by-pixel MTR maps were generated. On these maps, four regions of interest including right and left parietal and frontal white matter, as well as four regions of interest including right and left parietal and frontal cortex, respectively, were placed, and MTRs quantified as a percentage of signal loss [MTR = 100 (S0 − Ss) / S0, where S0 is the mean signal intensity for a particular region obtained from the T1-weighted spin echo sequence without the saturation pulse, and Ss is the mean signal intensity for the same region with the saturation pulse].

Two repeated measurements of MTRs in the white matter and cortex were obtained in each of 10 subjects included in the study for reproducibility purposes. Repeatibility coefficients19 were calculated, resulting in 4.42 and 6.74 for MTRs in the parietal white matter and parieto-occipital cortex, respectively.

Axial T2 fast spin echo images were then obtained for localization purposes for 1H-MRS. 1H–single voxel spectroscopy was obtained using a volume of interest (2 × 2 × 2 cm) in the right parietal white matter and point-resolved spectroscopy. Care was taken not to include the skull, subcutaneous fat, or ventricles in the volume of interest. A first spectrum (repetition time = 1,500, echo time = 30, 8 number of excitations) was obtained to quantify the four major brain metabolites: MI, NAA, Cre, and Cho. In addition, a second spectrum (repetition time = 6,000, echo time = 30, 8 number of excitations) was obtained from the same volume of interest as previously, to look specifically at Glu levels.

Automated prescanning was performed in all cases, and additional manual prescanning only when needed. This resulted in all spectra having 6 Hz or less of full width at half height of the unsuppressed water peak, and percentages of water suppression higher than 96%. All spectra were assessed for general quality by an experienced observer (B. Gómez-Ansón). Spectra were automatically postprocessed using GE software (PROBE-Quantool 2000, GE Medical Systems). Briefly, curve fitting and line width normalization are automatically performed, and four major peaks are identified: NAA, Cre, Cho, and MI. The fit amplitudes for each peak are reported as “machine numbers,” which are proportional to metabolite peak areas. Peak amplitudes are referenced to water as an internal standard and are therefore proportional to metabolic concentrations. Metabolite ratios (NAA/Cre, Cho/Cre, and MI/Cre) are then calculated taking the concentration equivalents of NAA, Cho, and MI to Cre.

Raw data of those spectra acquired after February 2003 (when the LC model was acquired) were transferred to a separate Ultra 60 SUN-WS (Sun Microsystems Inc., Santa Clara, CA) and postprocessed using the LC model version 620 to determine absolute metabolite concentrations. This is an automated, external reference method that uses a basis set of known metabolite concentrations for the different metabolites as a reference. Peak amplitudes in the acquired spectrum are compared to this basis set, and apparent absolute concentration metabolite determinations (mM) and their standard deviations are provided by the software. For the purposes of this study, MI and Glu values having a standard deviation below 40% were accepted. The LC model obviates most of the limitations of a semiquantitative method (Quantool), which is mainly dependent on the magnet's performance and allows determination of absolute metabolite concentrations.

Quality assurance was performed on a weekly basis during the entire length of the study to assess the stability of external conditions. In addition, for absolute metabolite quantitation (LC model), an NAA phantom of a known concentration (50 mM) was built, and a calibration factor was determined on a weekly basis.

Three-dimensional MR images (inversion recovery preparation time spoiled gradient, min echo time, preparation time 450, field of view 25, thickness 1.5 mm, 128 locs, 1 number of excitations) of the entire brain were also obtained, converted into ANALYZE format on a separate workstation, and then postprocessed using SIENAX version 2.2 (FSL version 3.0, Oxford, UK). Briefly, images are segmented into peripheral gray matter, white matter, and ventricular cerebrospinal fluid (Images 1–3), and then normalized to a standard template to provide normalized quantitative data (mm3).21

Statistical Analysis.

Comparisons between groups were performed using the nonparametric Mann-Whitney test for continuous data and the chi-square and Fischer tests for categorical data. Multiple comparisons were perfomed using the Kruskal-Wallis test. Correlation among variables was analyzed by Pearson correlation. Statistical analyses of the data were performed by using SPSS 10 statistical software (SPSS Inc. and Microsoft Corp., Chicago, IL). Results are given as mean ± SD. P < .05 was considered statistically significant.

Results

Characteristics of Patients.

Table 1 shows the demographic data, liver and renal tests, and hormones of patients included in the study. Most patients had advanced cirrhosis as indicated by ascites and high mean Child-Turcotte-Pugh score. Grade I hepatic encephalopathy was present in 7 of the 24 patients (29%) at the time of the study. Fourteen patients (58%) had dilutional hyponatremia, as defined by serum sodium concentration below 130 mEq/L associated with ascites and/or edema.11 The individual values of serum sodium concentration are shown in Fig. 1. In patients with hyponatremia, the median time elapsed between the first available value of serum sodium below 130 mEq/L and the inclusion in the study (minimum period of hyponatremia) was 8 days (range: 2–58 days), while the median time between the last avaible value of normonatremia (serum sodium >135 mEq/L) and the inclusion in the study (maximum period of hyponatremia) was 92 days (range: 22–256 days). The presence of hepatic encephalopathy grade I was more frequent in patients with hyponatremia (43%) than in those without hyponatremia (10%), yet the difference did not reach statistical significance (P = .08). Neuropsychological tests to evaluate the existence of minimal hepatic encephalopathy were not performed.

Table 1. Clinical and Laboratory Characteristics of All Patients
 Mean ± SDRange
  • *

    Normal values: 9–33 μmol/L.

Age (y)58 ± 236–75
Sex (M/F)17/7
Alcoholic cirrhosis (yes/no)11/13
Ascites (yes/no)22/2
Hepatic encephalopathy grade I (yes/no)7/17
Bilirubin (mg/dL)5 ± 1.20.7–30
Albumin (g/L)27 ± 0.921–35
Prothrombin time (%)55 ± 328–87
Child-Turcotte-Pugh class (B/C)10/14
Serum creatinine (mg/dL)1 ± 0.10.6–2.3
Serum sodium (mEq/L)129 ± 2118–143
Serum osmolality (Osm/kg)278 ± 4255–315
Urine sodium (mEq/L)11 ± 32–56
Plasma ammonia (μmol/L)*45 ± 59–97
Mean arterial pressure (mm Hg)79 ± 264–99
Plasma renin activity (ng/mL/h)6.4 ± 10.5–19
Aldosterone (ng/dL)148 ± 388–704
Norepinephrine (pg/mL)462 ± 5668–913
Arginine vasopressin (ng/L)1.5 ± 0.20.6–3.7
Figure 1.

Individual values of serum sodium concentration in patients with cirrhosis and healthy subjects. Hyponatremia was defined as serum sodium ≤130 mEq/L.9

Brain Organic Osmolytes.

Table 2 shows the values of brain organic osmolytes in patients with cirrhosis and in healthy subjects. Representative spectra are shown in Fig. 2. The concentration of MI, the predominant organic osmolyte in the human brain, was significantly reduced in patients with cirrhosis compared with that in healthy subjects, both estimated as relative values or in absolute concentrations. Considering all subjects together, there was a highly significant direct correlation between brain MI and serum sodium concentration and osmolality (Fig. 3). Similar correlations were also observed when only patients with cirrhosis were considered (r = 0.60, P = .002 and r = 0.7, P = .001, respectively). Patients with hyponatremia had significantly lower values of MI compared with those of patients without hyponatremia (15 ± 1 vs. 24 ± 2, respectively; P < .05) (Table 3). The mean reduction of brain MI in patients with hyponatremia with respect to the mean value in healthy subjects was 71% (range: 56%–100%) compared with 55% (range: 19%–69%) in patients without hyponatremia (P = .003). No relationship was found between the duration of hyponatremia and brain MI levels. Patients with grade I hepatic encephalopathy had lower MI levels compared with patients without hepatic encephalopathy (13.2 ± 6 vs. 20.8 ± 7, respectively; P = .025).

Table 2. Brain Organic Osmolytes in Patients With Cirrhosis and in Healthy Subjects
 Healthy Subjects (n = 8)PAll Cirrhotic Patients (n = 24)
  • *

    As determined by the Quantool method (see Patients and Methods).

  • Quantitative determination available in 14 cirrhotic patients and in all healthy subjects.

Myo-inositol*52 ± 8<.00119 ± 8
Myo-inositol (mM)4 ± 1<.0010.6 ± 0.2
Choline-containing compounds62 ± 8<.00140 ± 7
Choline-containing compounds (mM)1 ± 0.2<.0010.6 ± 0.1
Creatine/phosphocreatine*73 ± 8.269 ± 6
Creatine/phosphocreatine (mM)5 ± 1.044 ± 0.4
N-acetyl-aspartate*114 ± 12.9113 ± 15
N-acetyl-aspartate (mM)7 ± 1.026 ± 0.7
Glutamine/Glutamate (mM)8 ± 3.110 ± 3
Figure 2.

Representative 1H-MRS spectra from (a) a healthy subject, (b) a patient with cirrhosis without hyponatremia, and (c) a patient with cirrhosis and hyponatremia. Note the reduction in myo-inositol (Ino) and choline (Cho) peaks in patients with cirrhosis, particularly in the presence of hyponatremia, compared with peaks in the healthy subject. By contrast, glutamine/glutamate (Glx) peaks are increased in patients with cirrhosis compared with that of healthy subjects. Ino, myo-inositol; Cho, choline; Cr, creatine; Glx, glutamine/glutamate; NAA, N-acetyl-aspartate.

Figure 3.

Relationship between brain myo-inositol concentration and serum sodium and osmolality in patients with cirrhosis (boxes) and healthy subjects (open circles).

Table 3. Brain Organic Osmolytes in Cirrhotic Patients With and Without Hyponatremia
 Patients With Hyponatremia (n = 14)PPatients Without Hyponatremia (n = 10)
  • *

    As determined by the Quantool method (see Patients and Methods).

  • Quantitative determination available in 14 cirrhotic patients and in all healthy subjects.

Myo-inositol*15 ± 5.00624 ± 8
Myo-inositol (mM)0.5 ± 0.2.010.9 ± 0.2
Choline-containing compounds*39 ± 8.242 ± 5
Choline-containing compounds (mM)0.6 ± 0.1.40.7 ± 0.7
Creatine/phosphocreatine*65 ± 6.00273 ± 4
Creatine/phosphocreatine (mM)3 ± 0.2.0024 ± 0.2
N-acetyl-aspartate*109 ± 16.2117 ± 11
N-acetyl-aspartate (mM)6 ± 0.8.76 ± 0.4
Glutamine/Glutamate (mM)9 ± 3.0512 ± 2

To assess whether the reduction in brain MI was related to changes in serum sodium concentration or other factors (e.g., the severity of liver failure), univariate and multivariate analyses were performed in patients with cirrhosis. In the univariate analysis, brain MI levels were related to serum sodium, bilirubin, and albumin concentrations; prothrombin time; and Child-Turcotte-Pugh score. However, in a multivariate analysis, including serum sodium and Child-Turcotte-Pugh score, serum sodium concentration was the only independent predictor of low brain MI levels (P = .019).

Besides MI, the brain concentrations of four other substances (Cho, Cre, NAA, and Glu) also known to act as organic osmolytes in the setting of hyponatremia were assessed. Patients with cirrhosis had significantly lower concentration of Cho compared with that of healthy subjects (see Table 2). The concentrations of Cre and NAA were also lower in patients with cirrhosis but reached statistical significance only when absolute measurements of these two parameters were considered (see Table 2). As with MI, the brain concentrations of Cho, Cre, and NAA correlated directly with serum sodium concentrations (r = 0.73, P = .001; r = 0.63, P = .002; and r = 0.4, P = .05, respectively). However, when values of these parameters were compared in patients with and without hyponatremia, only differences in Cre reached statistical significance (see Table 3). In keeping with the results of previous studies,4, 5, 7, 9, 10 the intracerebral concentration of Glu in patients with cirrhosis was greater than that of healthy subjects (see Table 2). Patients with hyponatremia had values of Glu lower than those of patients without hyponatremia (see Table 3). In contrast to the existence of direct correlations between serum sodium concentration or serum osmolality and MI and other brain organic osmolytes, there was no correlation between brain Glu and serum sodium (Fig. 4) or osmolality. However, brain Glu correlated directly with plasma ammonia concentration (see Fig. 4). Plasma ammonia concentration did not correlate with the presence of hepatic encephalopathy or severity of liver failure.

Figure 4.

Relationship between brain glutamate/glutamine and serum sodium and plasma ammonia in patients with cirrhosis (boxes) and healthy subjects (open circles). Quantitative determinations of glutamine/glutamate were available only for 14 patients with cirrhosis.

MRI and MTR.

Table 4 shows MTR values measured in different areas of the brain in patients with cirrhosis and healthy subjects. No significant differences were observed between the two groups in any of the eight different areas studied. Moreover, there were no differences in MTR values between patients with and without hyponatremia (16.3 ± 3 vs. 14.8 ± 3, respectively; P = .3) (Table 5), and no correlation was found between MTR values and serum sodium or osmolality (r = 0.2 for both; P values were not significant). Plasma ammonia levels did not correlate with MTR values in patients with cirrhosis. Moreover, patients with high ammonia levels (above the upper normal limit) and hyponatremia had MTR values similar to those of patients with high ammonia levels without hyponatremia.

Table 4. MTR in Patients With Cirrhosis and in Healthy Subjects
MTRHealthy Subjects (n = 8)All Cirrhotic Patients (n = 24)
  1. NOTE: No P values were considered significant.

Right frontal grey matter14.9 ± 414.6 ± 4
Left frontal grey matter15.4 ± 315.3 ± 4
Right parietal grey matter13.6 ± 214.8 ± 4
Left parietal grey matter14.3 ± 414.2 ± 4
Mean grey matter14.6 ± 314.7 ± 3
Right frontal white matter17.6 ± 216.9 ± 4
Left frontal white matter17.8 ± 217.5 ± 4
Right parietal white matter17.3 ± 314.9 ± 3
Left parietal white matter17.0 ± 315.3 ± 3
Mean white matter17.4 ± 216.2 ± 3
Mean grey and white matter16.3 ± 315.4 ± 3
Table 5. MTR in Cirrhotic Patients With and Without Hyponatremia
MTRCirrhotics With Hyponatremia (n = 14)Cirrhotics Without Hyponatremia (n = 10)
  1. NOTE: No P values were considered significant.

Right frontal grey matter13.5 ± 415.9 ± 3
Left frontal grey matter15.0 ± 316.0 ± 4
Right parietal grey matter13.9 ± 416.0 ± 4
Left parietal grey matter12.9 ± 415.8 ± 5
Mean grey matter13.8 ± 415.8 ± 3
Right frontal white matter16.6 ± 417.2 ± 2
Left frontal white matter17.5 ± 517.4 ± 3
Right parietal white matter14.2 ± 415.8 ± 3
Left parietal white matter14.7 ± 315.9 ± 4
Mean white matter15.8 ± 316.6 ± 3
Mean grey and white matter14.8 ± 316.3 ± 3

Cerebral Volumes.

Patients with cirrhosis had lower total cerebral volume and white matter volume compared with values in healthy subjects, although the differences did not reach statistical significance. Cerebrospinal fluid values were significantly higher in patients with cirrhosis (Table 6). There were no significant differences in cerebral volumes between patients with and without hyponatremia (Table 7).

Table 6. Normalized Cerebral Volumes in Patients With Cirrhosis and in Healthy Subjects
 Healthy Subjects (n = 8)PAll Cirrhotic Patients (n = 24)
  1. NOTE: Total cerebral volume: grey matter volume + white matter volume + cerebrospinal fluid volume.

Total cerebral volume (cm3)1.273 ± 30.061.217 ± 81
Peripheral grey matter volume (cm3)539 ± 32.2519 ± 39
Cerebrospinal fluid volume (cm3)31 ± 6.0447 ± 23
Total grey matter volume (cm3)661 ± 37.2635 ± 42
White matter volume (cm3)611 ± 17.09581 ± 49
Table 7. Normalized Cerebral Volumes in Cirrhotic Patients With and Without Hyponatremia
 Cirrhotics With Hyponatremia (n = 14)PCirrhotics Without Hyponatremia (n = 10)
  1. NOTE: Total cerebral volume: grey matter volume + white matter volume + cerebrospinal fluid volume.

Total cerebral volume (cm3)1.217 ± 8311.213 ± 84
Peripheral grey matter volume (cm3)515 ± 43.9523 ± 36
Cerebrospinal fluid volume (cm3)51 ± 18.144 ± 29
Total grey matter volume (cm3)630 ± 47.7642 ± 37
White matter volume (cm3)587 ± 45.3574 ± 55

Discussion

In this study, 1H-MRS and MRI were used to investigate the effects of hyponatremia on brain organic osmolytes and brain water content in cirrhosis. Our findings in nonhyponatremic patients with cirrhosis are in keeping with previous studies showing a characteristic pattern of increased glutamine and reduced MI and Cho compared with values in healthy subjects.4–7, 9, 10 The presence of hyponatremia was associated with a further and important reduction in the already low levels of organic osmolytes, particularly MI, which is the main organic osmolyte in human brain.14 Of note, hyponatremia was the only independent predictor of low MI levels. Moreover, in the whole group of patients with cirrhosis studied, serum sodium concentration and serum osmolality correlated directly with the brain concentration of MI, Cre, NAA, and Cho. The changes observed in brain organic osmolytes in patients with cirrhosis and hyponatremia included in the current study are qualitatively similar to those described in patients with hyponatremia without liver disease.14 The difference is that brain glutamine, an organic osmolyte that is reduced in the setting of hyponatremia without liver disease,14 is not reduced in patients with cirrhosis despite the existence of hyponatremia and hypo-osmolality. Thus high glutamine levels are probably the consequence of an increased exposure of the brain to high ammonia levels.1–4 This contention is further supported by the observation of a direct correlation between brain glutamine and plasma ammonia levels in the present study, a finding also reported in a previous investigation.9

The possible relationship between the marked changes of brain organic osmolytes in hyponatremic patients and disturbances in neuropsychological function is not completely understood. Experimental studies in animal models and isolated brain cells suggest that organic osmolytes might have important cellular functions besides their prominent role in osmoregulation.3, 4, 22 Therefore, the exhaustion of these components in the setting of hyponatremia may contribute to astrocyte dysfunction and disturbances in neurological function in patients with liver disease. Along these lines, it has been shown that hyponatremia predisposes to brain edema in an experimental model of acute liver failure and hyperammonemia.17 Moreover, a recent study has shown that administration of hypertonic saline reduces intracranial pressure in patients with acute liver failure and cerebral edema.23 These findings are in keeping with the clinical observation that patients with cirrhosis and hyponatremia frequently develop hepatic encephalopathy in conditions associated with an increased ammonia synthesis, such as diuretic therapy, gastrointestinal bleeding, or infections. Moreover, hepatic encephalopathy associated with the development of brain edema has been reported in patients with cirrhosis and marked hyponatremia treated with transjugular intrahepatic portosystemic shunts.24 Although the present study was not specifically designed to investigate a possible relationship between hyponatremia and hepatic encephalopathy, there was a trend for a higher incidence of clinically significant hepatic encephalopathy in patients with hyponatremia compared with those without hyponatremia. Nevertheless, unequivocal proof of this possible relationship would require a study with a different design (e.g., longitudinal rather than cross-sectional) and larger sample size.

To assess whether the combination of hyperammonemia due to cirrhosis and low extracellular osmolality due to hyponatremia was associated with the development of brain edema, brain water content was estimated by measuring MTR and cerebral volume. None of these two methods is specific for the detection of changes in brain water content, because they may be altered by many other factors. Nevertheless, they are probably the best methods currently available to evaluate brain water content in humans. MTR consists of quantifying the transfer of magnetization between free protons in water and bound protons associated with macromolecules, and has been used mainly to assess demyelination and axonal loss in several pathological conditions, particularly multiple sclerosis.25, 26 It has been suggested that MTR may also detect small degrees of brain inflammation and edema.27 In the case of edema, the lower the MTR values, the higher the amount of water in the brain.

In the present study, MTR values obtained in white and gray matter of different brain regions in patients with cirrhosis and hyponatremia were lower than those of patients without hyponatremia and healthy subjects, yet the differences did not reach statistical significance. The lack of differences between patients with cirrhosis and healthy subjects are at odds with those of a recent study by Córdoba et al. performed in a series mainly composed by normonatremic patients with cirrhosis.28 In this latter study, patients with cirrhosis showed significantly lower MTR values compared with those of a control group of healthy subjects, a finding that was interpreted as suggestive of low-grade cerebral edema.28, 29 The reason for the discrepant findings between the two studies is not clear. In the present study, patients were investigated in the absence of diuretic therapy, while Córdoba et al.28 reported that the majority of patients were studied under diuretic therapy. Although the effect of diuretics on brain organic osmolytes and water has not been investigated, it could be speculated that diuretic therapy could increase brain glutamine synthesis through an increased exposure of the brain to ammonia from renal origin, further disrupt brain osmoregulation, and lead to an increase in brain water content. Alternatively, the differences among the two studies may be due to the different methods used to measure MTR: spin echo was used in the present study, while Córdoba et al.28 used gradient echo. Although there are no comparative studies between these two methods, a different sensitivity in the detection of brain water content cannot be ruled out completely. Furthermore, other factors in the method of measurement employed for MTRs (using only four regions of interest in our study, and obtaining an average value) or the threshold of acceptability for motion-degraded images (which may have been different in both studies) may also be partly responsible for the different results, and more evidence in this regard is needed. The other method used in the present study to estimate brain water content was the quantification of normalized, segmented volumes of grey and white matter and cerebrospinal fluid using three-dimensional MRI. This method has neither been used previously in patients with cirrhosis nor applied to the assessment of brain water, but it has been validated to estimate total brain volume.21 The assumption was that any significant increase in brain water would result in an increase of the volume of some regions of the brain or total brain volume. No differences were found in white and grey matter or in total cerebral volumes in patients with hyponatremia compared with those without hyponatremia. Taken together, the results of the MTR studies and measurements of cerebral volume suggest that under the conditions of the current study and with the possible limitations of the methods used, the presence of hyponatremia in patients with cirrhosis does not seem to be associated with an increased brain water content. However, the existence of a low-grade cerebral edema not detectable with the methods used cannot be ruled out completely.

Several other aspects of the present study merit a specific comment. First, all patients included had dilutional hyponatremia, which is the most common cause of hyponatremia in cirrhosis.11 Whether or not true hyponatremia (i.e., hyponatremia in the absence of ascites and edema)—which may also occur in cirrhosis—has similar effects on intracerebral osmolytes and water is unknown and would deserve specific investigation. Second, although the exact duration of hyponatremia could not be determined in our patients, most patients had been hyponatremic for at least several days or weeks before the study. Therefore, in many cases the brain had had quite a long period of adaptation to hypoosmolality, and the findings observed probably reflect a situation of chronic hyponatremia. In this regard, the lack of increased estimated brain water content observed in our patients with hyponatremia compared with those without hyponatremia is in keeping with results of experimental studies showing that after an increase in brain water content immediately after the development of hyponatremia, brain water returns to normal values after several days of hyponatremia.12, 15, 30 It is possible that signs of brain edema were found if patients had been studied earlier in the course of hyponatremia. Third, it would have been interesting to perform a second MR study in hyponatremic patients after correction of hyponatremia. However, this was not possible because spontaneous correction of dilutional hyponatremia did not occur in any of the patients included, and the V2 vasopressin receptor antagonists that are effective in improving serum sodium concentration are not available yet for use in clinical practice.31 Fourth, the very low values of MI found in patients with cirrhosis and hyponatremia could theoretically enhance the risk for the development of neurological complications (i.e., central pontine mielynolysis) after correction of hyponatremia compared with that of noncirrhotic hyponatremic patients. Therefore, treatment of cirrhotic patients with hyponatremia with newly developed V2-receptor antagonists should be done with great caution when they become available for use in clinical practice. Nevertheless, no single case of central pontine mielynolysis has been reported in phase 2 studies published so far.32, 33 Finally, patients with cirrhosis included in the current study showed changes in keeping with some degree of cerebral atrophy (decreased brain volumes and increased cerebrospinal fluid spaces), this being mainly a result of white matter volume reduction. Although the presence of brain atrophy in cirrhosis has been reported,34 the main contribution of reduction of white matter to this atrophy had not been previously demonstrated and would be worth further investigation.

In conclusion, hyponatremia in cirrhosis is associated with a remarkable reduction in the intracerebral concentration of organic osmolytes other than glutamine. This reduction is probably the result of a homeostatic osmoregulatory mechanism to prevent an increase in brain water content. The remarkable changes in brain organic osmolytes may contribute to further impairment of astrocyte function and represent a pathogenic link between hyponatremia and hepatic encephalopathy.

Acknowledgements

We thank Raquel Cela for her technical assistance. We also thank Dr. Wladimiro Jiménez and the nursing staff of the Hospital Clínic Liver Unit for their help.

Ancillary

Advertisement