Changes in Levels of Phosphorus Metabolites in Temporal Lobes of Drug-Naive Schizophrenic Patients

In vivo ³¹P magnetic resonance spectroscopy (MRS) allows direct measurement of membrane phospholipids and high-energy phosphate metabolism in the brain (1). With ³¹P MRS, increased levels of phosphodiesters and decreased levels of phosphomonoesters have been observed in the frontal lobes of drug-naive schizophrenic patients (2, 3). Increased β-phosphates of ATP (β-ATP) and decreased inorganic orthophosphate also have been reported (2). When ³¹P MRS studies have included chronically medicated schizophrenic patients, findings have been inconsistent: while most investigators have found decreased phosphomonoesters or increased phosphodiesters or both (4–6), a recent study demonstrated a decrease in phosphodiester level in schizophrenic patients (7). These inconsistencies may be caused by differences in patient characteristics such as chronicity of illness, diagnostic subtype, and medication status, as well as the MRS method employed. ³¹P MRS specifically investigating the temporal lobes also has yielded contradictory results in medicated schizophrenic patients (5, 8, 9). Disturbed phospholipid metabolism has been proposed as a neurodevelopmental pathogenesis of schizophrenia (10). Under this hypothesis, MRS in the drug-naive state is likely to give more useful information for investigating pathophysiologic mechanisms of schizophrenia than studies of medicated patients. However, no report of MRS has described metabolite changes in the temporal lobes of drug-naive schizophrenic patients. In this study, we report metabolite changes observed in the temporal lobes of schizophrenic patients during initial psychotic episodes while they were still drug naive.

METHOD

Seventeen first-episode, drug-naive Japanese patients (10 men and seven women; mean age=23.1 years) who met DSM-III-R criteria for schizophrenia and were right-handed according to the Edinburgh Handedness Inventory were recruited from the outpatient clinics of Fujimoto Hospital and Kagoshima University Hospital from 1991 to 1996. One neuropsychiatrist evaluated the patients with the Oxford version of the Brief Psychiatric Rating Scale (BPRS). All the patients were treated in the two institutions, and their diagnosis was reconfirmed 1 year after the first scan. Patients had been ill for 6.6 months (SD=6.2). Seventeen age- and gender-matched healthy subjects (mean age=22.5 years) served as a comparison group. All subjects gave written informed consent for participation in the study. None had a recent history of alcohol or drug abuse.

The method of MRS data acquisition and processing has been described in our previous report (4). Spectroscopy was performed with a Siemens-Asahi Meditec MR system with a magnetic field strength of 2.0 T. A circular polarizing head coil was tuned to 84.5 MHz for proton imaging and to 34.2 MHz for in vivo multivoxel ³¹P MRS (two-dimensional chemical shift imaging). The field of view was 24 cm with an 8×8 data matrix and a 4-cm section thickness. Spectra were obtained from the two volumes of interest, each consisting of 72 ml (Figure 1). The TR was 2 seconds, and the TE was 1.72 msec. Twelve measurements were obtained for each spectrum. Data were processed with Fourier transformation and exponential multiplication (16 Hz) and then phase-corrected. Spectral peaks were obtained for phosphomonoesters, inorganic orthophosphate, phosphodiesters, phosphocreatine, and γ-, α-, and β-ATP. Spectra were quantified according to peak-area measurements. An automated baseline correction technique removed the distortion in the baseline of the spectra (11). Peak measures such as height, position, and width were obtained by a Lorentzian curve-fitting procedure (Figure 2). For each spectrum, the integrated areas of the seven metabolites were measured, and mole percentages of total phosphorus signal were calculated.

Repeated measures analysis of variance (ANOVA), with a between-subject factor of diagnosis and a within-subject factor of side, was applied to the mole percentage of the seven metabolites.

RESULTS

Significant effects of diagnosis were seen for phospho­monoesters and phosphodiesters, with increased levels of phosphodiesters and decreased levels of phospho­monoesters in both temporal lobes (Table 1). A significant diagnosis-by-side interaction was observed for phosphocreatine, with higher values on the left than on the right side in schizophrenic patients relative to healthy subjects. MR imaging revealed no obvious abnormalities in patients or healthy subjects. The mean BPRS score was 35.6 (SD=12.7). Kendall’s rank correlation coefficient revealed no significant correlation between the total BPRS score and percentages of phosphomonoesters (N=17; left: τ=0.17, p=0.34; right: τ=0.10, p=0.59), phosphodiesters (N=17; left: τ=0.04, p=0.84; right: τ=0.04, p=0.84), or phosphocreatine (N=17; left: τ=–0.07 p=0.68; right: τ=– 0.19, p=0.30).

DISCUSSION

In this study, ³¹P MRS detected an elevation of phospho­diesters and reduction of phosphomonoesters in the temporal lobes of drug-naive schizophrenic patients compared with healthy subjects. These results are consistent with previous observations reported in the prefrontal cortex of schizophrenic patients (2, 3). Pettegrew et al. (2) have speculated that decreased phosphomonoesters and increased phosphodiesters in the frontal lobes of schizophrenic patients may reflect decreased synthesis and increased breakdown of membrane phospholipids. However, the interpretation of the findings is not easy, as they previously suggested. Decreased phosphomonoesters resonance in this study may imply a reduction in freely mobile phospho­monoesters (phosphocholine, phosphoethanolamine) or less mobile molecules (including phosphorylated proteins) or both (1). Reduced synthesis of membrane phospholipids is one of these possibilities. The phosphodiester resonance in ³¹P MRS in vivo is believed to be derived from mobile phosphodiester moieties (small membrane phospholipid structures such as micelles and vesicles) and breakdown products (1, 12). Phosphodiesters are more concentrated in white than in the gray matter (13). Therefore, the increase in phosphodiesters could have resulted from increased mobile phosphodiester moieties including glycerophosphocholine and glycerophosphoethanolamine, as well as small membrane phospholipid structures, or it could have been a reflection of a decreased ratio of gray-to-white matter volume in the volume of interest (14). Which phosphodiester components contribute to the elevation of phosphodiester resonance could be distinguished by using ¹H-decoupled ³¹P MRS. A preliminary study has shown that membrane or mobile phospholipids are increased in the frontal lobes of chronically medicated schizophrenic patients (6). On the other hand, elevations of glycerophosphocholine and glycerophosphoethanolamine concentrations have been demonstrated in the parietal lobes of young medicated schizophrenic patients compared with elderly schizophrenic patients and healthy subjects (15). Although definitive determination of the origins of decreased phosphomonoesters and increased phosphodiesters is difficult, the disturbed membrane phospholipid metabolism may not be restricted to the frontal lobe in the manner of the gray matter volume reduction observed in schizophrenic patients (16).

The level of phosphocreatine was increased in the left temporal lobe in schizophrenic patients. Phosphocreatine is known to be rapidly transformed to ATP when ATP is consumed by neuronal activity (17). An increased percentage of phosphocreatine may imply reduced ATP utilization in the left temporal lobe of drug-naive schizophrenic patients. This asymmetric abnormality in energy metabolism agrees with left-sided functional impairments observed in the temporal lobes of schizophrenic patients with single photon emission computed tomography (18).

Several methodologic limitations need to be addressed. The phosphorus metabolites were analyzed without correction for multiple comparison because of the small group size and the exploratory nature of this study, in spite of the increased risk of a type I error. Our curve-fitting method may have a drawback in that all seven metabolites were modeled as single spectral peaks, which could influence the results. Other methodologic limitations inherent in MRS procedures have been outlined in previous reports (8, 19). Further studies, in a larger group and with more sophisticated in vivo MRS techniques, will be needed to confirm our preliminary findings.

Received July 13, 1998; revision received Feb. 5, 1999; accepted Feb. 9, 1999. From the Department of Neuropsychiatry, Faculty of Medicine, Kagoshima University; and the South Japan Health Science Center, Miyazaki. Address reprint requests to Dr. Hiroshi Fukuzako, Department of Neuropsychiatry, Faculty of Medicine, Kagoshima University, 8-35-1 Sakuragaoka, Kagoshima 890-8520, Japan; [email protected] (e-mail) Supported by grants from the Ministry of Science, Culture, and Education (05770735 and 06770764) and the National Center of Psychiatry and Neurology of the Ministry of Health and Welfare (3A-5) of Japan (Dr. H. Fukuzako).

TABLE 1

Enlarge table