. Study of chronic schizophren ics using 3 1 ~ magnetic reson an ce chemical shift imaging 4
e
Fujimoto T, Nakano T, Takano T, Hokazono Y, Asakura T, Tsuji T. Study of chronic schizophrenics using 3’P magnetic resonance chemical shift imaging. Acta Psychiatr Scand 1992: 86: 455-462.
1
Phosphorus-3 1 chemical shift imaging showed regional abnormalities of in vivo 31PNMR spectra in the brains of chronic schizophrenic patients. In the left temporal region, the level of % phosphodiesters (PDE) was increased and the level of % y rw P-ATP (obtained by summation of y-ATP, rw-ATP, and P-ATP) was decreased. In the basal ganglia, the levels of % P D E were decreased and the level of yo phosphomonoesters was increased. The levels of % y rw P-ATP were increased in the right basal ganglia. The level of % phosphocreatine was decreased in the frontoparietal region. These findings may represent different patterns of dysfunction of membrane phospholipid bilayers and high-energy phosphate metabolism in the specific cerebral regions.
Recent progress in in viva phosphorus-3 1 magnetic resonance spectroscopy (31PMRS) has made it feasible to obtain 3’PNMR spectra non-invasively from the brain (1,2). In vivo 31PNMR spectra show the resonance peaks representing phosphomonoesters (PME), inorganic phosphate (Pi), phosphodiesters (PDE), phosphocreatine (PCr), y-ATP, a-ATP and P-ATP. 31PMRS, therefore, has been highlighted for its application potential to mental disorders. Pettegrew et al. (3,4) have carried out investigations on first-episode, drug-naive schizophrenic patients using a dual-tune surface coil with 1.5 T and reported lower levels of PME and Pi and higher levels of PDE and ATP in the dorsal prefrontal cortex. Stanley et al. ( 5 ) investigated the left dorsal prefrontal cortex in first-episode, drug-naive schizophrenic patients and chronic schizophrenic patients using the surface coil and FROGS pulse sequence with a 2 T 31PMRS. These studies showed evidence of lower levels of PME and P-ATP and a higher level of PDE. In chronic schizophrenic patients, the PME level was observed to be lower while the PDE level presented practically no change as compared with the control subjects (5). We have studied chronic schizophrenic patients using 31P chemical shift imaging (31PCSI) and reported higher levels of PDE and lower levels of ATP in the bilateral temporal regions (6). These reports have suggested that in vivo 31 P MRS is sensitive enough to detect changes in 31P NMR spectra in the brains of patients with schizophrenia. 31PCSI is able to yield 31PNMR spectra simultaneously from an array of voxels in a single exam-
T. Fujimoto’, T. Nakano’, T. Takano’,
Y. Hokazono’, T. Asakura3, T. Tsuji4
’
Departments of Psychiatry and Biochemistry, South Japan Health Science Centre, Fujimoto Hospital, Kagoshima University, Tokyo Medical and Dental University, Japan
Key words: schizophrenia; magnetic resonance; chemical shift imaging Toshiro Fujimoto, 17-4 Hayasuzu, Miyakonojo, 885, Japan Accepted for publication August 15, 1992
ination. These voxels correspond to a T, weighted image (Fig. 1).The voxels can be shifted on the image to acquire 31PNMR spectra in the volume of interest. Compared with the surface coil method, 31PCSI is superior in that it detects 31PNMR spectra with an optimum signal-to-noise ratio from relatively deep-lying regions in the brain, such as the temporal and central regions (1). It has been noted in the 31 P CSI method, however, that a baseline distortion of the spectra is generated due to delay time when the free induction decay (FID) acquisition method is used, necessitating the application of phase-encoding pulses (7). With the newly developed baseline correction technique, however, it has become possible to correct this distortion. We have used the 31P CSI method to acquire 31P NMR spectra from the temporal, frontoparietal and central regions (basal ganglia, corpus callosum) for comparison with the equivalent spectra obtained from normal subjects. Material and methods
This study was carried out on 16 chronic schizophrenic patients (all men, mean age 39 & 5 years) and 20 normal healthy controls (15 men and 5 women, mean age 34 2 years). The age of onset of the disease ranged between 18 and 30 years (mean age 22 t 4 years). Fourteen patients were right-handed and 2 patients were left-handed. Patient diagnosis was in accordance with the DSM-111-R criteria: 13 patients were of the disorganized type and 3 patients were the residual type. The Positive and Negative Syndrome Scale (PANSS) (8) and the Brief Psychi455
Fujimoto et a].
Fig. 1. "P-NMR spectra were obtained from 10 brain regions corresponding to the T, weighted 'H image using chemical shift imaging. The voxel size was 3 x 3 x 4 cm.
atric Rating Scale (BPRS) were also performed. All of the patients were under anti-psychotic medication. The studies were initiated after obtaining the informed consent of all patients and controls. The MR unit (Siemens-Asahi Meditech) used for this study with a static magnetic field strength of 2.0 T (phosphorus resonance frequency: 34.2 MHz) has a built-in quadrature detection (QD) bead coil. The investigations were performed by the CSI method, using a repetition time of 2000 ms, a delay time of 1.72 ms, a vector size of 1024, an acquisition of 12, a matrix size of 8 x 8, a slice thickness of 4 cm, and a voxel size of 3 cm x 3 cm x 4 cm (36 cm'). The measurement time was 26 min. By these means, 3'P NMR spectra were obtained from 10 brain regions (Fig. 1). These regions were the right and left frontoparietal regions (mainly frontal) (voxels 1, 2), the right temporal regions (voxels 3, 7), the central regions (mainly the basal ganglia including the corpus callosum and lateral ventricle) (voxels 4, 5), the tem-
Table 1. Percentage of individual peak areas to total observable peak areas of 3'P NMR spectra obtained from the voxel of Fig. 1. Comparison between chronic schizophrenic patients and normal controls
%PME
Voxel
1
2
3
4
5
6
7
8
9
10
Patient
10.8 (1.8) 10.1 (1.9)
10.7 (2.0) 10.0 (2.0)
11.1 (1.7) 10.0 (2.0)
10.0 (2.3) 9.5 (2.2)
12.3** (1.81 9.7 (2.61
9.2 (1.8) 9.1 (2.1)
9.9 (2.6) 10.3 (2.0)
10.2 (1.9) 10.8 (2.6)
10.8 (2.1) 9.8 (3.0)
10.4 (2.11 9.4 (2.6)
4.5 (1.O) 5.6 (2.01
6.0 (2.0) 6.0 (2.0)
6.4 (1.2) 6.0 (1.9)
5.3 (0.8) 5.3 (1.7)
5.6 (1.8) 5.5 (2.0)
4.5 (1.7) 5.1 (1.7)
5.3 (1.9) 5.5 (2.1)
4.6 (1.2) 4.9 (1.6)
4.6 (1.61 5.7 (1.8)
5.0 (1.2) 5.1 (2.2)
41.8 (3.5) 40.3 (4.1)
40.1 (3.5) 40.0 (3.51
37.8 (3.5) 38.9 (4.81
38.6* (3.51 41.6 (3.0)
36.7** (4.0) 40.2 (3.61
43.2* (3.2) 41.0 (2.91
37.1 (5.31 36.5 (4.51
37.9 (3.8) 40.0 (3.1)
37.5 (4.4) 36.8 (4.7)
42.1 (3.3) 40.1 (4.1I
8.7 (1.2) 9.5 (14
(0.1) 9.7 (1.1)
10.0 (2.0) 10.6 (13)
10.5 (0.9) 10.0 (1.51
11.0 (0.1) 10.3 (1.7)
10.9 (1.2) 10.1 (1.5)
11.0 (2.9) 11.9 (1.8)
11.6 (2.2) 12.0 (2.3)
11.9 (1.81 11.9 (2.0)
10.5 (1.51 11.3 (1.4)
10.3 (1.4) 10.5 (1.5)
10.4 (1.71 10.6 (1.5)
10.2 (1.61 10.5 (1.51
11.7 (1.6) 10.4 (1.61
10.2 (2.1) 11.1 (1.9)
9.8 (1.3) 10.6 (1.6)
10.4 (2.7) 10.1 (1.9)
11.4 (2.2) 10.4 (2.1)
10.7 (1.8) 11.4 (2.0)
10.0 (1.5) 9.8 (2.1)
14.2 (3.11 14.7 (5.2)
14.3 (2.2) 14.1 (2.7)
14.4 (3.0) 13.8 (2.0)
14.9 (3.4) 13.0 (2.4)
13.4 (2.7) 13.0 (2.5)
13.4 (2.5) 13.9 (2.2)
15.2 (3.4) 14.8 (1.9)
14.9 (2.5) 13.1 (2.3)
13.6 (3.0) 13.8 (3.0)
12.7 (2.3) 13.6 (2.91
9.3 (2.31 9.3 (2.6)
9.3 (1.8) 9.6 (2.3)
9.9 (2.2) 10.2 (1.7)
9.1 (2.0) 10.2 (2.5)
10.5 (1.3) 10.2 12.1)
9.0 (1.8) 10.2 (1.5)
10.7 (1.8) 10.9 12.1)
9.0 (3.1) 8.9 11.8)
10.6 (2.7) 10.5 12.2)
9.2 (1.3) 10.7 13.0)
33.9 (3.0) 34.5 (5.1)
34.1 (2.8) 34.3 (3.3)
34.6 (3.5) 34.5 (2.7)
35.8* (3.6) 33.6 (2.8)
34.2 (4.1) 34.3 (4.11
32.3** (2.1) 34.7 (2.81
36.4 (3.7) 35.7 (2.6)
35.3** (3.7) 32.4 (2.6)
35.1 (4.6) 35.7 (3.5)
31.8 (2.3) 34.1 (5.3)
Control %Pi
Patient Control
%PDE
Patient Control
%PCr
Patient Control
%l-ATP
Patient Control
%Y,-ATP
Patient Control
%P-ATP
Patient Control
%yaP-ATP
Patient Control
* P 0 . 5 , ** P=O.Ol, U-test, (SD)
456
8.8*
Chronic schizophrenia poral regions plus brain stem (voxels 8, 9) and the left temporal regions (voxels 6, lo). The 31P NMR spectra thus obtained were subjected to phase correction first and then to baseline correction. After baseline correction, the curve-fitting parameters (peak height, peak position and peak width) were determined. With these parameters, it was possible to calculate the peak areas for the following 7 peaks: PME, Pi, PDE, PCr, y-ATP, a-ATP, and 0-ATP. The above 7 peaks were then added to obtain the total phosphorus peak area, and the peak ratio ( % ) of each individual peak to the total peak area was determined. These are given as %PME, %Pi, %PDE, "/,PCr, % y A T P , %a-ATP and XP-ATP, respectively. The total ATP peak ratio (obtained by summation of the a-ATP, P-ATP, and y-ATP peaks) has been expressed here as the %y a P ATP. The PCr/Pi ratio was calculated and the pH determined from the chemical shift difference between Pi and PCr. Results
Table 1 shows the results of a comparison between the individual peak/total phosphorus peak ratios of the 31PNMR spectra for the chronic schizophrenic patients and the healthy controls. There was no significant difference in %Pi, %y-ATP, %a-ATP and % P-ATP. % PME, however, was significantly higher in voxel 5 for the patients. % P D E was significantly lower in voxels 4 and 5 (Fig. 2). However, in voxel 6, % P D E was significantly higher for the patients
(Fig. 3). %PCr was significantly lower in voxel 2. % y 'Y P ATP was significantly lower in voxel 6, but conversely higher in voxels 4 and 8. There were no significant differences in the PCr/Pi ratio (Table 2) or the intracellular pHi (Table 3 ) . Discussion
With the 31P CSI method used in this study, the primary problem concerns the reliability of the in vivo 3'P NMR spectra. The 3 ' P NMR spectra obtained with a 2 T MRS unit are known to offer a superior signal/noise (S/N) ratio and an increased spectral dispersion as compared with a 1.5 T MRS system (9). The 31PNMR spectra achieved with the present 31PCSI method had a non-uniform signal intensity due to inhomogeneity in the magnetic field associated with patient head size and shape. It was therefore difficult to compare the spectra in terms of their signal intensities. When the individual peak/total phosphorus peak ratio was considered, however, it was found to be stable. For the 20 healthy controls, the individual peak/total phosphorus peak ratio remained within a constant standard deviation. When the 20 healthy controls were divided into two groups of 10, it was seen that the ratios did not significantly change between the two groups. The human 31P NMR spectra obtained by Stanley et al. ( 5 ) with a 2 T MRS presented practically the same trend for the total peak ratios, so that their reliability can be considered very satisfactory.
A1 0
0 0
I
e o 0
*
O
55
PCr
I
0 0 0 0
0 0
-0
Fig. 2. 31P-NMR spectra of the central region (voxel 5 ) (mainly basal ganglia). A. Chronic schizophrenic patient (male 46 years old) %PDE is smaller than for the control. B. Control subject (male, 35 years old).
457
Fujimoto et al.
m
PCr
i\
111
'Cr
A
r-ATP
I
PDE
;1
r-ATP
, * is 0
PME
xi
Pi
Fig. 3. 31P-NMR spectra of the left temporal region (voxel 6). A. Chronic schizophrenic patient (male, 39 years old) %PDE is higher than for the control. B. Control subject (male, 37 years old).
The %Pi for the chronic schizophrenic patients showed no significant difference from those of the normal healthy controls. Nor did the results of Stanley et al. ( 5 ) show a significant difference in the % Pi for their chronic schizophrenic patients. The in vivo % PDE levels for the chronic schizophrenic patients of this study were significantly lower in the central regions (voxels 4 and 5, mainly basalganglia) (Fig. 2). Conversely, the %PDE level in the left temporal region (voxel6) was significantly higher (Fig. 3), thus indicating that an opposite phenomenon from that in the basal ganglia region had occurred. In in vitro 31P NMR, there are mainly two components in the PDE regions. One is glycerol 3-
%PME was significantly higher in voxel5 (mainly basal ganglia) in the chronic schizophrenic patients. The main components of the PME peaks are phosphocholine, phosphoethanolamine and a-glycerophosphate (4, 10, 11). PME reflects the anabolic activity of phospholipids. The higher level of % PME possibly indicates a higher anabolic activity of membrane phospholipids in this region. Stanley et al. (5), however, reported a lower level of %PME. This difference in terms of %PME is associated partly with the fact that the signals were obtained from different regions of the brain and also that the patients selected for this study have different characteristics from those chosen by Stanley et al.
Table 2. PCr/Pi ratio in voxels. Comparison between chronic schizophrenic patients and normal controls Voxel
1
2
3
4
5
6
7
8
9
10
Patients
2.00 (0.73)
1.57 (0.48)
1.60 (0.37)
2.01 (0.40)
2.14 (1.01)
2.66 (0.98)
2.45 (1.65)
2.64 (0.96)
2.97 (1.73)
2.17 (0.60)
Controls
1.93 (0.84)
1.78 (0.55)
1.91 (0.50)
2.09 (0.74)
2.10 (0.88)
2.20 (0.85)
2.44 (0.88)
2.76 (1.18)
2.23 (0.59)
2.64 (1.18)
Table 3. lntracellular pH in voxels. Comparison between chronic schizophrenic patients and normal controls Voxel
1
2
3
4
5
6
7
8
9
10
Patients
7.03 (0.03)
7.04 (0.04)
7.04 (0.02)
7.01 (0.03)
7.04 (0.05)
7.03 (0.03)
7.06 (0.06)
7.04 (0.05)
7.03 (0.05)
7.05 (0.05)
Controls
7.04 (0.04)
7.05 (0.04)
7.04 (0.04)
7.01 (0.03)
7.03 (0.03)
7.04 (0.03)
7.06 (0.07)
7.03 (0.05)
7.04 (0.06)
7.05 (0.05)
458
Chronic schizophrenia
phosphoethanolamine and the other is glycerol 3-phosphocholine. These two components occur only in the catabolic pathways of phospholipids (10, 1I). It has been reported that the in vivo and in vitro PDE peaks are markedly different. According to Cerdan (12)’ the large peak in the PDE region of the dog brain in vivo accounted for more than 35 % of the total phosphorus peaks using 31PNMR with a 2.1 T (35.8 MHz for phosphorus). However, it was only 8.32% in vitro measured by the methano1:HCl extraction method using high resolution 31PNMR at 143.73 MHz for phosphorus. The reason for this significant difference is that the PDE resonance in the in vivo 31P NMR spectra is due to substantial contributions from a fraction of mobile brain phospholipids. However, the in vivo PDE peaks depend on the magnetic field. Murphy et al. (13) proposed the view that the PDE component of the in vivo 31P NMR spectra was primarily from the phospholipid bilayer with a small contribution from a motionally averaged macromolecule (s). This is different from Cerdan’s interpretation. Our findings can be interpreted as reflecting regional dysfunctions of the membrane phospholipids themselves, probably of the phospholipid bilayer. It has been reported that a chronic ethanol-induced increase in PDE in the rat liver may show an alteration in the fluidity, structure or amount of the membrane (14). This finding may make it possible to interpret the changes in the % PDE in this study. The physical state of phospholipid bilayer is normally liquid crystalline and is affected by temperature (15). Individual phospholipid molecules have high mobility within the bilayer plane. The protein constituents of membranes are embedded in and through the phospholipid bilayer (the fluid mosaic model: Singer & Nicolson). Fluidity of the phospholipid bilayer permits specific interactions among different integral proteins. Translational motion of membrane proteins are limited only by the phospholipid bilayer viscosity. Membrane phospholipids have a close relation to the mechanisms of detection, integration and transmissions of signals. Membrane transport mechanisms regulate concentrations of ions and molecules within the cell and extracellular environment (15). The dysfunction of the phospholipid bilayer in the brains of chronic schizophrenic patients may alter fluidities, enzyme activities, receptor enzyme couplings, neurotransmitters’ actions, ion transport ( N a + , K’, Ca’ +,ATP) and multiple regulatory mechanisms of signals (16, 17). Turnovers of catecholamines and the complex activity of multiple neurotransmitter interactions may also be affected. Inositol-containing phospholipids that give rise to a dual second-messenger system may be altered. What sorts of factors could relate to the dysfunc-
tion of the membrane phospholipid bilayer in the chronic schizophrenic patients? First, the dysfunction may be specific to chronic schizophrenia. The drug-naive schizophrenic patients showed a higher level of % PDE (3-5); the higher level of
e
Fujimoto T, Nakano T, Takano T, Hokazono Y, Asakura T, Tsuji T. Study of chronic schizophrenics using 3’P magnetic resonance chemical shift imaging. Acta Psychiatr Scand 1992: 86: 455-462.
1
Phosphorus-3 1 chemical shift imaging showed regional abnormalities of in vivo 31PNMR spectra in the brains of chronic schizophrenic patients. In the left temporal region, the level of % phosphodiesters (PDE) was increased and the level of % y rw P-ATP (obtained by summation of y-ATP, rw-ATP, and P-ATP) was decreased. In the basal ganglia, the levels of % P D E were decreased and the level of yo phosphomonoesters was increased. The levels of % y rw P-ATP were increased in the right basal ganglia. The level of % phosphocreatine was decreased in the frontoparietal region. These findings may represent different patterns of dysfunction of membrane phospholipid bilayers and high-energy phosphate metabolism in the specific cerebral regions.
Recent progress in in viva phosphorus-3 1 magnetic resonance spectroscopy (31PMRS) has made it feasible to obtain 3’PNMR spectra non-invasively from the brain (1,2). In vivo 31PNMR spectra show the resonance peaks representing phosphomonoesters (PME), inorganic phosphate (Pi), phosphodiesters (PDE), phosphocreatine (PCr), y-ATP, a-ATP and P-ATP. 31PMRS, therefore, has been highlighted for its application potential to mental disorders. Pettegrew et al. (3,4) have carried out investigations on first-episode, drug-naive schizophrenic patients using a dual-tune surface coil with 1.5 T and reported lower levels of PME and Pi and higher levels of PDE and ATP in the dorsal prefrontal cortex. Stanley et al. ( 5 ) investigated the left dorsal prefrontal cortex in first-episode, drug-naive schizophrenic patients and chronic schizophrenic patients using the surface coil and FROGS pulse sequence with a 2 T 31PMRS. These studies showed evidence of lower levels of PME and P-ATP and a higher level of PDE. In chronic schizophrenic patients, the PME level was observed to be lower while the PDE level presented practically no change as compared with the control subjects (5). We have studied chronic schizophrenic patients using 31P chemical shift imaging (31PCSI) and reported higher levels of PDE and lower levels of ATP in the bilateral temporal regions (6). These reports have suggested that in vivo 31 P MRS is sensitive enough to detect changes in 31P NMR spectra in the brains of patients with schizophrenia. 31PCSI is able to yield 31PNMR spectra simultaneously from an array of voxels in a single exam-
T. Fujimoto’, T. Nakano’, T. Takano’,
Y. Hokazono’, T. Asakura3, T. Tsuji4
’
Departments of Psychiatry and Biochemistry, South Japan Health Science Centre, Fujimoto Hospital, Kagoshima University, Tokyo Medical and Dental University, Japan
Key words: schizophrenia; magnetic resonance; chemical shift imaging Toshiro Fujimoto, 17-4 Hayasuzu, Miyakonojo, 885, Japan Accepted for publication August 15, 1992
ination. These voxels correspond to a T, weighted image (Fig. 1).The voxels can be shifted on the image to acquire 31PNMR spectra in the volume of interest. Compared with the surface coil method, 31PCSI is superior in that it detects 31PNMR spectra with an optimum signal-to-noise ratio from relatively deep-lying regions in the brain, such as the temporal and central regions (1). It has been noted in the 31 P CSI method, however, that a baseline distortion of the spectra is generated due to delay time when the free induction decay (FID) acquisition method is used, necessitating the application of phase-encoding pulses (7). With the newly developed baseline correction technique, however, it has become possible to correct this distortion. We have used the 31P CSI method to acquire 31P NMR spectra from the temporal, frontoparietal and central regions (basal ganglia, corpus callosum) for comparison with the equivalent spectra obtained from normal subjects. Material and methods
This study was carried out on 16 chronic schizophrenic patients (all men, mean age 39 & 5 years) and 20 normal healthy controls (15 men and 5 women, mean age 34 2 years). The age of onset of the disease ranged between 18 and 30 years (mean age 22 t 4 years). Fourteen patients were right-handed and 2 patients were left-handed. Patient diagnosis was in accordance with the DSM-111-R criteria: 13 patients were of the disorganized type and 3 patients were the residual type. The Positive and Negative Syndrome Scale (PANSS) (8) and the Brief Psychi455
Fujimoto et a].
Fig. 1. "P-NMR spectra were obtained from 10 brain regions corresponding to the T, weighted 'H image using chemical shift imaging. The voxel size was 3 x 3 x 4 cm.
atric Rating Scale (BPRS) were also performed. All of the patients were under anti-psychotic medication. The studies were initiated after obtaining the informed consent of all patients and controls. The MR unit (Siemens-Asahi Meditech) used for this study with a static magnetic field strength of 2.0 T (phosphorus resonance frequency: 34.2 MHz) has a built-in quadrature detection (QD) bead coil. The investigations were performed by the CSI method, using a repetition time of 2000 ms, a delay time of 1.72 ms, a vector size of 1024, an acquisition of 12, a matrix size of 8 x 8, a slice thickness of 4 cm, and a voxel size of 3 cm x 3 cm x 4 cm (36 cm'). The measurement time was 26 min. By these means, 3'P NMR spectra were obtained from 10 brain regions (Fig. 1). These regions were the right and left frontoparietal regions (mainly frontal) (voxels 1, 2), the right temporal regions (voxels 3, 7), the central regions (mainly the basal ganglia including the corpus callosum and lateral ventricle) (voxels 4, 5), the tem-
Table 1. Percentage of individual peak areas to total observable peak areas of 3'P NMR spectra obtained from the voxel of Fig. 1. Comparison between chronic schizophrenic patients and normal controls
%PME
Voxel
1
2
3
4
5
6
7
8
9
10
Patient
10.8 (1.8) 10.1 (1.9)
10.7 (2.0) 10.0 (2.0)
11.1 (1.7) 10.0 (2.0)
10.0 (2.3) 9.5 (2.2)
12.3** (1.81 9.7 (2.61
9.2 (1.8) 9.1 (2.1)
9.9 (2.6) 10.3 (2.0)
10.2 (1.9) 10.8 (2.6)
10.8 (2.1) 9.8 (3.0)
10.4 (2.11 9.4 (2.6)
4.5 (1.O) 5.6 (2.01
6.0 (2.0) 6.0 (2.0)
6.4 (1.2) 6.0 (1.9)
5.3 (0.8) 5.3 (1.7)
5.6 (1.8) 5.5 (2.0)
4.5 (1.7) 5.1 (1.7)
5.3 (1.9) 5.5 (2.1)
4.6 (1.2) 4.9 (1.6)
4.6 (1.61 5.7 (1.8)
5.0 (1.2) 5.1 (2.2)
41.8 (3.5) 40.3 (4.1)
40.1 (3.5) 40.0 (3.51
37.8 (3.5) 38.9 (4.81
38.6* (3.51 41.6 (3.0)
36.7** (4.0) 40.2 (3.61
43.2* (3.2) 41.0 (2.91
37.1 (5.31 36.5 (4.51
37.9 (3.8) 40.0 (3.1)
37.5 (4.4) 36.8 (4.7)
42.1 (3.3) 40.1 (4.1I
8.7 (1.2) 9.5 (14
(0.1) 9.7 (1.1)
10.0 (2.0) 10.6 (13)
10.5 (0.9) 10.0 (1.51
11.0 (0.1) 10.3 (1.7)
10.9 (1.2) 10.1 (1.5)
11.0 (2.9) 11.9 (1.8)
11.6 (2.2) 12.0 (2.3)
11.9 (1.81 11.9 (2.0)
10.5 (1.51 11.3 (1.4)
10.3 (1.4) 10.5 (1.5)
10.4 (1.71 10.6 (1.5)
10.2 (1.61 10.5 (1.51
11.7 (1.6) 10.4 (1.61
10.2 (2.1) 11.1 (1.9)
9.8 (1.3) 10.6 (1.6)
10.4 (2.7) 10.1 (1.9)
11.4 (2.2) 10.4 (2.1)
10.7 (1.8) 11.4 (2.0)
10.0 (1.5) 9.8 (2.1)
14.2 (3.11 14.7 (5.2)
14.3 (2.2) 14.1 (2.7)
14.4 (3.0) 13.8 (2.0)
14.9 (3.4) 13.0 (2.4)
13.4 (2.7) 13.0 (2.5)
13.4 (2.5) 13.9 (2.2)
15.2 (3.4) 14.8 (1.9)
14.9 (2.5) 13.1 (2.3)
13.6 (3.0) 13.8 (3.0)
12.7 (2.3) 13.6 (2.91
9.3 (2.31 9.3 (2.6)
9.3 (1.8) 9.6 (2.3)
9.9 (2.2) 10.2 (1.7)
9.1 (2.0) 10.2 (2.5)
10.5 (1.3) 10.2 12.1)
9.0 (1.8) 10.2 (1.5)
10.7 (1.8) 10.9 12.1)
9.0 (3.1) 8.9 11.8)
10.6 (2.7) 10.5 12.2)
9.2 (1.3) 10.7 13.0)
33.9 (3.0) 34.5 (5.1)
34.1 (2.8) 34.3 (3.3)
34.6 (3.5) 34.5 (2.7)
35.8* (3.6) 33.6 (2.8)
34.2 (4.1) 34.3 (4.11
32.3** (2.1) 34.7 (2.81
36.4 (3.7) 35.7 (2.6)
35.3** (3.7) 32.4 (2.6)
35.1 (4.6) 35.7 (3.5)
31.8 (2.3) 34.1 (5.3)
Control %Pi
Patient Control
%PDE
Patient Control
%PCr
Patient Control
%l-ATP
Patient Control
%Y,-ATP
Patient Control
%P-ATP
Patient Control
%yaP-ATP
Patient Control
* P 0 . 5 , ** P=O.Ol, U-test, (SD)
456
8.8*
Chronic schizophrenia poral regions plus brain stem (voxels 8, 9) and the left temporal regions (voxels 6, lo). The 31P NMR spectra thus obtained were subjected to phase correction first and then to baseline correction. After baseline correction, the curve-fitting parameters (peak height, peak position and peak width) were determined. With these parameters, it was possible to calculate the peak areas for the following 7 peaks: PME, Pi, PDE, PCr, y-ATP, a-ATP, and 0-ATP. The above 7 peaks were then added to obtain the total phosphorus peak area, and the peak ratio ( % ) of each individual peak to the total peak area was determined. These are given as %PME, %Pi, %PDE, "/,PCr, % y A T P , %a-ATP and XP-ATP, respectively. The total ATP peak ratio (obtained by summation of the a-ATP, P-ATP, and y-ATP peaks) has been expressed here as the %y a P ATP. The PCr/Pi ratio was calculated and the pH determined from the chemical shift difference between Pi and PCr. Results
Table 1 shows the results of a comparison between the individual peak/total phosphorus peak ratios of the 31PNMR spectra for the chronic schizophrenic patients and the healthy controls. There was no significant difference in %Pi, %y-ATP, %a-ATP and % P-ATP. % PME, however, was significantly higher in voxel 5 for the patients. % P D E was significantly lower in voxels 4 and 5 (Fig. 2). However, in voxel 6, % P D E was significantly higher for the patients
(Fig. 3). %PCr was significantly lower in voxel 2. % y 'Y P ATP was significantly lower in voxel 6, but conversely higher in voxels 4 and 8. There were no significant differences in the PCr/Pi ratio (Table 2) or the intracellular pHi (Table 3 ) . Discussion
With the 31P CSI method used in this study, the primary problem concerns the reliability of the in vivo 3'P NMR spectra. The 3 ' P NMR spectra obtained with a 2 T MRS unit are known to offer a superior signal/noise (S/N) ratio and an increased spectral dispersion as compared with a 1.5 T MRS system (9). The 31PNMR spectra achieved with the present 31PCSI method had a non-uniform signal intensity due to inhomogeneity in the magnetic field associated with patient head size and shape. It was therefore difficult to compare the spectra in terms of their signal intensities. When the individual peak/total phosphorus peak ratio was considered, however, it was found to be stable. For the 20 healthy controls, the individual peak/total phosphorus peak ratio remained within a constant standard deviation. When the 20 healthy controls were divided into two groups of 10, it was seen that the ratios did not significantly change between the two groups. The human 31P NMR spectra obtained by Stanley et al. ( 5 ) with a 2 T MRS presented practically the same trend for the total peak ratios, so that their reliability can be considered very satisfactory.
A1 0
0 0
I
e o 0
*
O
55
PCr
I
0 0 0 0
0 0
-0
Fig. 2. 31P-NMR spectra of the central region (voxel 5 ) (mainly basal ganglia). A. Chronic schizophrenic patient (male 46 years old) %PDE is smaller than for the control. B. Control subject (male, 35 years old).
457
Fujimoto et al.
m
PCr
i\
111
'Cr
A
r-ATP
I
PDE
;1
r-ATP
, * is 0
PME
xi
Pi
Fig. 3. 31P-NMR spectra of the left temporal region (voxel 6). A. Chronic schizophrenic patient (male, 39 years old) %PDE is higher than for the control. B. Control subject (male, 37 years old).
The %Pi for the chronic schizophrenic patients showed no significant difference from those of the normal healthy controls. Nor did the results of Stanley et al. ( 5 ) show a significant difference in the % Pi for their chronic schizophrenic patients. The in vivo % PDE levels for the chronic schizophrenic patients of this study were significantly lower in the central regions (voxels 4 and 5, mainly basalganglia) (Fig. 2). Conversely, the %PDE level in the left temporal region (voxel6) was significantly higher (Fig. 3), thus indicating that an opposite phenomenon from that in the basal ganglia region had occurred. In in vitro 31P NMR, there are mainly two components in the PDE regions. One is glycerol 3-
%PME was significantly higher in voxel5 (mainly basal ganglia) in the chronic schizophrenic patients. The main components of the PME peaks are phosphocholine, phosphoethanolamine and a-glycerophosphate (4, 10, 11). PME reflects the anabolic activity of phospholipids. The higher level of % PME possibly indicates a higher anabolic activity of membrane phospholipids in this region. Stanley et al. (5), however, reported a lower level of %PME. This difference in terms of %PME is associated partly with the fact that the signals were obtained from different regions of the brain and also that the patients selected for this study have different characteristics from those chosen by Stanley et al.
Table 2. PCr/Pi ratio in voxels. Comparison between chronic schizophrenic patients and normal controls Voxel
1
2
3
4
5
6
7
8
9
10
Patients
2.00 (0.73)
1.57 (0.48)
1.60 (0.37)
2.01 (0.40)
2.14 (1.01)
2.66 (0.98)
2.45 (1.65)
2.64 (0.96)
2.97 (1.73)
2.17 (0.60)
Controls
1.93 (0.84)
1.78 (0.55)
1.91 (0.50)
2.09 (0.74)
2.10 (0.88)
2.20 (0.85)
2.44 (0.88)
2.76 (1.18)
2.23 (0.59)
2.64 (1.18)
Table 3. lntracellular pH in voxels. Comparison between chronic schizophrenic patients and normal controls Voxel
1
2
3
4
5
6
7
8
9
10
Patients
7.03 (0.03)
7.04 (0.04)
7.04 (0.02)
7.01 (0.03)
7.04 (0.05)
7.03 (0.03)
7.06 (0.06)
7.04 (0.05)
7.03 (0.05)
7.05 (0.05)
Controls
7.04 (0.04)
7.05 (0.04)
7.04 (0.04)
7.01 (0.03)
7.03 (0.03)
7.04 (0.03)
7.06 (0.07)
7.03 (0.05)
7.04 (0.06)
7.05 (0.05)
458
Chronic schizophrenia
phosphoethanolamine and the other is glycerol 3-phosphocholine. These two components occur only in the catabolic pathways of phospholipids (10, 1I). It has been reported that the in vivo and in vitro PDE peaks are markedly different. According to Cerdan (12)’ the large peak in the PDE region of the dog brain in vivo accounted for more than 35 % of the total phosphorus peaks using 31PNMR with a 2.1 T (35.8 MHz for phosphorus). However, it was only 8.32% in vitro measured by the methano1:HCl extraction method using high resolution 31PNMR at 143.73 MHz for phosphorus. The reason for this significant difference is that the PDE resonance in the in vivo 31P NMR spectra is due to substantial contributions from a fraction of mobile brain phospholipids. However, the in vivo PDE peaks depend on the magnetic field. Murphy et al. (13) proposed the view that the PDE component of the in vivo 31P NMR spectra was primarily from the phospholipid bilayer with a small contribution from a motionally averaged macromolecule (s). This is different from Cerdan’s interpretation. Our findings can be interpreted as reflecting regional dysfunctions of the membrane phospholipids themselves, probably of the phospholipid bilayer. It has been reported that a chronic ethanol-induced increase in PDE in the rat liver may show an alteration in the fluidity, structure or amount of the membrane (14). This finding may make it possible to interpret the changes in the % PDE in this study. The physical state of phospholipid bilayer is normally liquid crystalline and is affected by temperature (15). Individual phospholipid molecules have high mobility within the bilayer plane. The protein constituents of membranes are embedded in and through the phospholipid bilayer (the fluid mosaic model: Singer & Nicolson). Fluidity of the phospholipid bilayer permits specific interactions among different integral proteins. Translational motion of membrane proteins are limited only by the phospholipid bilayer viscosity. Membrane phospholipids have a close relation to the mechanisms of detection, integration and transmissions of signals. Membrane transport mechanisms regulate concentrations of ions and molecules within the cell and extracellular environment (15). The dysfunction of the phospholipid bilayer in the brains of chronic schizophrenic patients may alter fluidities, enzyme activities, receptor enzyme couplings, neurotransmitters’ actions, ion transport ( N a + , K’, Ca’ +,ATP) and multiple regulatory mechanisms of signals (16, 17). Turnovers of catecholamines and the complex activity of multiple neurotransmitter interactions may also be affected. Inositol-containing phospholipids that give rise to a dual second-messenger system may be altered. What sorts of factors could relate to the dysfunc-
tion of the membrane phospholipid bilayer in the chronic schizophrenic patients? First, the dysfunction may be specific to chronic schizophrenia. The drug-naive schizophrenic patients showed a higher level of % PDE (3-5); the higher level of
