Introduction

       Bipolar affective disorder is a severe psychiatric illness which can be life-threatening. At least 1% of population were affected and fluctuations between manic and depressed mood states were its character. Mood stabilizers such as lithium and sodium valproate have had a remarkable beneficial effect on the lives of millions suffered from bipolar affective disorder. Yet, the cellular and molecular basis for mood stabilizers' therapeutic effects remains to be fully elucidated. One widely accepted hypothesis for lithium's mechanism of action is the inositol-depletion theory ^[1]. This theory proposes that lithium blocks inositol monophosphatase and causes an accumulation of inositol monophosphates and a corresponding depletion of myo-inositol. Details of this theory have been described in literature ^[1,2,3,4,5]. According to this theory, there should be a decrease of myo-inositol concentration after lithium medication.

       The inositol-depletion theory is supported by in vitro MR spectroscopy (MRS) studies in rat brain extracts, especially with high resolution MR spectroscopy ^{[4, 6]}. In vitro cell studies in rat brain also found a decrease of myo-inositol concentration after lithium administration ^[7,8,9]. However, in vitro human MRS studies to date have been less clear, showing higher ratio of myo-inositol/creatine in the temporal lobe ^[10] and the basal ganglia region ^[11], no significant difference in the occipital lobe ^{[5, 11]} and parietal lobe ^[5], and lower myo-inositol concentrations in right frontal lobe ^[5] after lithium medication. Moreover, inositol depletion in lithium-treated rats was not detectable in vivo MR spectroscopy also ^[6].

       In addition to lithium, sodium valproate is now also commonly used as an effective mood stabilizer as sodium valproate can be well-tolerated by the patients. Nonetheless, it was felt to have a different mechanism of action since sodium valproate does not appear to inhibit inositol monophosphatase ^{[12, 13]}. However, interestingly, in a recent animal study we have shown that both sodium valproate and lithium affected the phosphoinositol cycle (PI-cycle) in the same manner, therefore possibly sharing a common mechanism of action in the treatment of bipolar disorder ^[4].

       The purpose of the present study was to quantitatively measure the concentration of metabolites in both frontal and temporal lobes after sodium valproate medication in euthymic bipolar patients to determine if these were altered in any way from controls.

Materials and Methods

Subjects

       We studied 9 adult patients with bipolar affective disorder and 11 healthy volunteers. This study was approved by the local ethics committee and an informed consent form was signed by each subject after the nature of the procedure had been fully explained. The subjects were excluded from the study if they were not suitable for MR spectroscopy study, such as cardiac pacemake, implanted neurostimulator, implanted drug delivery system, surgery within the last two months, metal fragments in the body, and claustrophobia. The patients were euthymic outpatients and met the diagnostic criteria for bipolar affective disorder. Among the patients, there were seven women and two men with a mean age of 42 years (SD 9, range = 34~58 years). All the patients took sodium valproate 1000 mg daily as a the sole medication continuously. The healthy volunteers were recruited by advertisement. The volunteers were screened in detail to make sure they were healthy. There were five female and six male volunteers aged 30~50 years (mean 37 and SD 7). None of the volunteers was taking any medication.

Magnetic Resonance Spectroscopy

       Magnetic resonance experiments were performed by using a Magnex 3 T scanner and a quadrature birdcage coil, and spectrometer control was provided by an Surrey Medical Imaging System (SMIS) console. The external standard method was used in this study. The external standard was a 125ml glass sphere filled with physiological saline containing 5 mmol NAA, 5 mmol GABA, 2.5 mmol Glutamine, 2.5 mmol Glutamate, 4mmol creatine, 1 mmol choline chloride, 2.5 mmol myo-inositol. The sphere was placed beside the subject's head and no discomfort to the subject was caused due to its small size. First, axial and coronal scout brain images were collected using gradient echo imaging sequences. The position resolved spectroscopy (PRESS) sequence was used to acquire proton MRS data with TE1=25 ms, TE2=25 ms, TR=3000 ms, and 128 scan averages. The MRS data were acquired from three square voxels (2 cm×2 cm×2 cm) placed in the cortex of left frontal lobe, the cortex of left temporal lobe, and external standard solution (Fig. 1). Special care was taken to place the voxel in same location for each subject by a neuroradiologist. From the scout images of each subject, the average coordinates ^{[14, 15]} of the centers of the two brain voxels were determined: x = 0.5 mm (SD=1.6), y = 63.5 mm (SD=12.1), z = -25.5 mm (SD=4.2) in the frontal lobe, and x = 32.2 mm (SD=6.3), y = 20.5 mm (SD=3.9), z = 0.7 mm (SD=2.6) in the temporal lobe. In order to measure T1 and T2 values of the metabolites in the brain and external standard solution, MRS data were collected with different TE values at a constant TR and different TR values at a constant TE both for two healthy volunteers and two patients and also from external standard solution ^[15].

点击查看大图

Fig 1.  The voxels (white box) were localized based on series of scout coronal and axial images. Voxels (2 cm×2 cm×2 cm) were placed in the left frontal lobe (1A), left temporal lobe (1B), and external standard solution (1C), respectively.

MRS Data Analysis

       The MRS data were transferred to another computer for analysis. After phase and baseline correction, Marquardt functions were used to fit the resonance peaks of N-acetylaspartate (NAA), total creatine, choline-containing compounds, and myo-inositol, and their respective peak areas were measured using Peak Research (PERCH) program (distributed by PERCH project, Department of Chemistry, University of Kuopio, Kuopio, Finland). An average area of both 3.56 ppm and 3.65 ppm myo-inositol peaks was used to quantify myo-inositol concentration.

       For quantitative measurement of brain metabolite concentrations, the following equation of MR signal intensity, S (represented by resonance peak area), for MRS sequence ^[15,16,17] was used (equation 1):

       Where N represents the number of metabolite molecules per unit voxel. By fitting S and TR (or TE) values to equation 1 at constant TE (or TR), metabolite T1 (or T2) values were obtained for the two healthy volunteers, the two patients, and the external standard solution.

       From equation 1 we can derive equation 2 to calculate any metabolite molecule numbers per unit voxel in either patient or healthy volunteer using PRESS sequence:

       where the subscripts s and b represent standard solution and brain, respectively ^{[15, 17]}.

       To obtain accurate brain metabolite concentrations, [Met]b, in millimoles per kg of wet brain, the CSF volume fraction, fcsf, in the spectroscopic voxels must be corrected. Thus, brain metabolite concentrations were caculated as in equation 3:

       where Vvoxel is the volume of a 8 cm³ spectroscopic voxel ^{[15, 18]}.

点击查看大图

点击查看大图

点击查看大图

Statistical Analysis

       The two-tailed unpaired Student t test was used for determining the significance of difference between the means of metabolite concentrations. Differences were deemed significant if P value was less than 0.05.

Results

       MR scans were completed for all subjects. All the spectra were judged to be of adequate quality for the purpose of quantitative analysis.

       Table 1 lists the concentrations of brain metabolites in both frontal and temporal lobes for healthy volunteers. In the healthy volunteer group, the mean concentration of myo-inositol was 7.42±0.66 mmol/kg frontally and 7.47±0.67 mmol/kg temporally. Table 2 lists the concentrations of brain metabolites in both frontal and temporal lobes for the patients. The mean concentration of myo-inositol was 7.33±0.89 mmol/kg frontally and 7.44±1.02 mmol/kg temporally in the patient group. Compared with the healthy volunteers, the patients had no significantly lower levels of myo-inositol (P = 0.80 in frontal lobe and P = 0.95 in temporal lobe). Also, there were no significant differences for myo-inositol concentration between frontal and temporal lobes, no matter whether they were volunteers or patients.

       With regard to other brain metabolites, the concentrations of choline-containing compounds, total creatine and NAA were similar among healthy volunteers and bipolar disorder patients (Table 1 and Table 2). Through statistical analysis, there were no significant differences for choline-containing compounds (P = 0.42 in frontal lobe and P = 0.85 in temporal lobe), total creatine (P = 0.88 in frontal lobe and P = 0.84 in temporal lobe) and NAA (P = 0.14 in frontal lobe and P = 0.65 in temporal lobe) between volunteer and patient groups.

       Figure 2 shows examples of proton magnetic resonance spectra for metabolites in frontal lobe of volunteer (2A), frontal lobe of patient (2B) and standard solution (2C). Similar spectra were found among volunteers and patients in our study. In some spectra obtained from both volunteer and patient, the myo-inositol 3.65 peak (contains inositol monophosphates) was higher than 3.56 peak (contains inositol monophosphates and glycine), although higher 3.56 peak was common in human brain. But in standard solution, the myo-inositol 3.56 peak is always higher than 3.65 peak. The contributions of macro molecules should also be taken into consideration.

点击查看大图

Fig 2.  Proton magnetic resonance spectra in a healthy volunteer (2A), a patient with bipolar affective disorder (2B), and the standard solution (2C).

点击查看表格

Tab 1.  Metabolite concentrations (mmol/kg wet brain) in frontal and temporal lobes in healthy volunteers

点击查看表格

Tab 2.  Metabolite concentrations (mmol/kg wet brain) in frontal and temporal lobes in the patients with bipolar disorder

Discussion

       The inositol depletion theory is widely accepted to explain lithium's therapeutic effectiveness. For many years, measuring concentration of brain myo-inositol after lithium administration has been a major issue in an attempt to illustrate the mechanism of therapeutic action ^{[3,4,5,6, 10, 19,20,21,22]}. However, no overwhelming results have been reached, especially in human MR spectroscopy studies. Both increase ^{[10, 11]} and decrease ^[5] of myo-inositol concentration after lithium medication were observed.

       The therapeutic effects of sodium valproate in psychiatric diseases are most substantially recognized in biporlar affective disorder. This well-tolerated medication has shown the promising efficacy for the patients with bipolar affective disorder. One study demonstrates that sodium valproate inhibits the activity of glycogen synthase kinase-3 ^[23]. Our previous findings using high resolution MR spectroscopy in rat extracts suggest that both lithium and sodium valproate may share a common mechanism of action in the treatment of bipolar disorder via actions on the PI-cycle since sodium valproate administration produced exactly the same results as lithium administration ^[4].

       Our present results show a small, but non-significant, decrease in the mean values of brain myo-inositol concentration in the patients with bipolar disorder compared to in healthy volunteers. There were also no statistically significant differences for choline-containing compounds, total creatine and NAA. These results do not suggest that sodium valproate administration significantly alters baseline concentrations of myo-inositol in euthymic bipolar patients. These findings are consistent with those from previous studies of lithium, in which myo-inositol concentrations have been altered only in patients who have an active mood disorder.

       One of the potential problems with measuring myo-inositol with MRS is that both the 3.65 ppm and 3.56 ppm myo-inositol peaks contain contribution from inositol monophosphates and 3.56 ppm peak is also contaminated with signal from glycine. In our opinion, the statement that inositol monophosphates and glycine contribute a minor component (<5%) to the total myo-inositol resonance ^[5] was too arbitrary. According to Scholtz et al ^[24], numbers of protons contributing to in vivo 3.6 ppm peak are 4:2:2:2 for myo-inositol, myo-inositol-1-P, myo-inositol-3-P, myo-inositol-1, 4-P2, respectively. In human brain, the myo-inositol peaks can be influenced by the signal contribution from macro molecules also ^[25]. The accuracy of myo-inositol concentration may thus be obscured by the contributions from inositol monophasphates, glycine and macro molecules, especially when there is an increase of inositol monophasphates and glycine after mood stabilizer medication.

       Localized proton spectra from the frontal part of the human brain are often distorted by "ghost" artifacts, especially with short echo time PRESS sequence ^[26]. Our initial experiments also showed that the quality of short echo time PRESS sequence in frontal lobe was unsatisfactory. Ideally, myo-inositol measurements should be performed at short echo times since myo-inositol resonances exhibit relatively short T2 relaxation times. As a compromise between a little signal loss and eddy current or artifact problems at very short echo time ^[20], we used TE1 25 ms and TE2 25 ms for PRESS sequence in this study. The peaks of myo-inositol is of good quality in all subjects.

       In summary, brain metabolites in frontal and temporal lobes were quantitatively measured by in vivo proton MR spectroscopy. No significant difference of brain myo-inositol concentration was observed between healthy volunteers and euthymic bipolar patients after sodium valproate medication. Sodium valproate administration does not significantly alter baseline concentrations of myo-inositol in euthymic bipolar patients. However, more subjects are necessary to reach a reasonable result in future studies.