| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Neuroradiology |
1 From the Department of Radiology (H. Kado, I.K.), and Department of Neuropsychiatry and Neurology (K.N.), Akita Research Institute for Brain and Blood Vessels, 6-10 Kubotamachi, Sensyu, Akita 010-0874, Japan; and Departments of Radiology (H. Kimura) and Neuropsychiatry (T.M.), Fukui University, Fukui, Japan. Received August 11, 2004; revision requested October 21; revision received January 13, 2005; accepted February 16; final version accepted March 11. Address correspondence to H. Kado.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Materials and Methods: All subjects gave written informed consent to an institutional committee for clinical research–approved study protocol. The clinical course after medication in 52 patients with depressive psychosis (age range, 52–78 years; 21 men, 31 women) was investigated. In all patients, MR spectroscopy was performed with a 1.5-T MR imaging unit before the initiation of medical therapy. Cerebrovascular lesions (CVLs), which appear as high-signal-intensity areas on T2-weighted MR images, were evaluated by using the Fazekas rating scale. Patients were classified into two groups on the basis of the ratio of N-acetylaspartate (NAA) to creatine and phosphocreatine (Cr): Patients in group A had an NAA/Cr ratio greater than 1.91, and patients in group B had an NAA/Cr ratio of 1.91 or less. To assess the response of the patients to medication, standard psychiatric tests—the Verbal Associative Fluency Test (VAFT), the Digit Symbol Test (DST), the Mini-Mental State Examination (MMSE), and the Hamilton Depression Rating Scale (HAM-D)—were administered before and after medical therapy was initiated. Mean test scores before and after medication were compared with paired t testing. P < .05 was considered to indicate a significant difference.
Results: There were 25 patients in group A and 27 in group B. In group A, the mean VAFT and DST scores increased and the mean HAM-D score decreased after medication. There was no significant difference in mean MMSE scores before and after medication (P = .945 for group A and P = .934 for group B). In group B, there were no significant differences in any of the psychiatric test scores before and after medication. The high-signal-intensity area score in group B was significantly higher than that in group A (P = .004).
Conclusion: MR spectroscopy data obtained before the initiation of medical therapy were useful in predicting prognosis in patients with depressive psychosis; this suggests that the combined burden of all CVLs may affect the response to antidepressant medication.
© RSNA, 2006
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
In our previous study (4), the findings of marked periventricular hyperintensities and deep white matter hyperintensities on T2-weighted MR images were significantly more frequent in a group of patients with late-onset geriatric depression (onset after 50 years of age) than in a group with early onset geriatric depression (onset before 50 years of age). We also validated the hypothesis that patients with severe periventricular hyperintensities and deep white matter hyperintensities on T2-weighted MR images had more pronounced cognitive impairment and more severe depression than those with mild or no periventricular hyperintensities and deep white matter hyperintensities. Moreover, we observed an important relationship between deep white matter lesions and proton MR spectroscopy data in elderly depressive patients: The mean ratio of N-acetylaspartate (NAA) to the sum of creatine and phosphocreatine (Cr) in the late-onset depression group was lower than that in the early onset depression group.
Although MR imaging is widely used to investigate T2-hyperintensity lesions and atrophy in patients with geriatric depression, MR spectroscopy is still not commonly used for clinical applications. We believe that MR spectroscopic techniques will prove useful for improving our understanding of the pathologic mechanisms of depression in elderly patients. We hypothesized that an understanding of the neural conditions and pathologic mechanisms involved in depression before therapy could be helpful for anticipating the effectiveness of different therapeutic agents and selecting the treatment modality. Thus, the purpose of our study was to prospectively assess the usefulness of MR spectroscopy data obtained before medical therapy in predicting prognosis in patients with depressive psychosis.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Healthy Volunteers
In addition to 52 patients with depressive psychosis, 15 healthy volunteers were recruited for the sake of examining normal metabolite ratios at cerebral MR spectroscopy and examining high-signal-intensity areas on T2-weighted MR images and the degree of brain atrophy in healthy subjects. The volunteers included nine men and six women who ranged in age from 54 to 74 years. These volunteers met no DSM-IV criteria. They had no history of cortical infarction, cerebral hemorrhage, or other major CVLs with neurologic symptoms at conventional MR imaging and no history of alcoholism, brain degenerative disease, or dementia. All healthy volunteers underwent brain MR angiography, at which no definite steno-occlusive lesions in the main arterial vessels were confirmed. Therefore, these patients were considered to be healthy for the purposes of the present study, and they gave their written informed consent to a protocol approved by the committee for clinical research at our institution.
MR Imaging and Interpretation
A 1.5-T MR imaging unit (Signa; GE Medical Systems, Milwaukee, Wis) was used to perform conventional MR imaging. Transverse spin-echo T1-weighted (repetition time msec/echo time msec, 333/15), T2-weighted (3000/80), and intermediate-weighted (3000/30) sequences were performed parallel to the orbitomeatal line. The section thickness was 5 mm with a 2-mm gap, the field of view was 22 cm, and the acquisition matrix was 512 x 512.
Conventional MR imaging was performed in both the 52 patients with depressive psychosis before medical therapy and the 15 healthy volunteers for assessment of high-signal-intensity areas on T2-weighted MR images and the degree of brain atrophy. High-signal-intensity areas are considered to reflect microvascular lesions, ischemic disease, degenerative changes, and dilated perivascular spaces. Two board-certified neuroradiologists (H. Kado and H. Kimura) who had 12 and 17 years of experience with brain MR imaging and who were blinded to the clinical information for all subjects (the 52 patients and the 15 healthy volunteers) independently evaluated the high-signal-intensity areas for both the patients with depressive psychosis and the healthy volunteers by using the Fazekas rating scale (5).
Fazekas classifications are defined by the frequency, forms, and regions of high-signal-intensity areas. The high-signal-intensity areas were assigned one of four grades, as follows: Grade 0 indicated the absence of such areas; grade 1, the presence of punctate foci of high signal intensity; grade 2, the presence of beginning confluence of foci of high signal intensity; and grade 3, the presence of large confluent areas of high signal intensity. The regions examined in high-signal-intensity area estimations are the periventricular white matter, the deep white matter, and the deep gray matter (the basal ganglia and thalamus). We quantitatively evaluated the high-signal-intensity areas by using the combined score for the above three regions. We also estimated the degree of brain atrophy in the patients with depressive psychosis and the healthy volunteers by using the following four grades: Grade 0 indicated no definite atrophy; grade 1, mild atrophy; grade 2, moderate atrophy; and grade 3, severe atrophy.
Proton MR Spectroscopic Technique
By using the same MR imaging unit, proton MR spectroscopy was performed in all patients with depressive psychosis before the initiation of medical therapy and in the healthy volunteers. MR spectra were acquired by using point-resolved echo spectroscopy with 2000/136 and 64 signals acquired. Water suppression was achieved with three chemical shift–selective pulses followed by dephasing gradients in the preparation periods. Spectral data were acquired in 2K data points with a 2500-Hz window. Line broadening with a Gaussian function of 1.25 Hz was performed. Zero filling was performed once to obtain 4K data, followed by Fourier transformation and automatic phase correction. All peaks were fitted by using the Marquardt algorithm after automated calculation of peak area ratios. All spectral data processing was automatically performed with the Probe sequence on the MR imaging unit console. The probe acquisition protocols (probe-P; GE Medical Systems) in this study included radiofrequency transmitter receiver gain adjustment, water suppression, and an autoshim process, followed by data acquisition (6).
Symmetrical volumes of interest of 8 cm3 were placed in the deep white matter of the frontal lobe bilaterally for all patients and healthy volunteers. The placement of volumes of interest was performed independently by two trained neuroradiologists (H. Kado and H. Kimura) who had 12 and 17 years of experience with brain MR spectroscopy and who were blinded to the subjects' clinical information. The areas of three major peaks were measured: those for choline (Cho) (at 3.2 ppm), Cr (at 3.0 ppm), and NAA (at 2.0 ppm). The area ratios of NAA and Cho relative to Cr were used for relative evaluation. We used the area of Cr as a reference because it has been reported to remain relatively stable, even in the presence of rapid fluctuations in energy metabolism (7–10). In this context, the NAA/Cr and Cho/Cr ratios were judged to be clinically effective for use in the assay of cerebral metabolites.
Group Classification
The 52 patients were separated into two groups on the basis of the proton MR spectroscopy data. Because the mean NAA/Cr ratio in the 15 healthy volunteers (mean age, 63.9 years ± 6.5 [standard deviation]) was 2.29 ± 0.38, patients with mean NAA/Cr values greater than 1.91 were included in group A, and patients with mean NAA/Cr values of 1.91 or less were included in group B. The threshold value, which was defined as 1 standard deviation lower than the mean NAA/Cr ratio in healthy volunteers, was 1.91. The mean values of NAA/Cr calculated by averaging the values in the left and right frontal lobe volumes of interest were used for classification of groups in this study. Typical examples of MR spectroscopy data in groups A and B are given in the Results section.
Medical Therapy
In this study, a tricyclic antidepressant (imipramine) and a tetracyclic antidepressant (setiptiline) were used as the first and second choices for treatment, respectively. The dosage, the kind of drug used, and duration of medical therapy were individualized according to the requirements of each patient, with the control of a senior neurologist (K.N.) and a senior psychiatrist (T.M.). The former had 25 years of experience; the latter, 17. Each physician treated a portion of the total number of patients. The initial dosage of imipramine was 10 mg three times daily; the dosage was then increased gradually as required to up to 150 mg per day. If imipramine could not be used owing to adverse effects such as dry mouth, blurred vision, disturbance of accommodation, dysuria, and constipation, setiptiline was used at a dosage of 1 or 2 mg three times daily as required. The medication was continued for approximately 3–6 months according to the needs of the patient. During the period of medication, if the neurologist or psychiatrist recognized clinical improvement, neuropsychological tests were performed. If little response was observed in the patients, the medication was nonetheless continued for 6 months before the neuropsychologic tests were performed.
Neuropsychologic Examination
Four neuropsychologic tests—the Verbal Associative Fluency Test (VAFT), the Mini-Mental State Examination (MMSE), the Digit Symbol Test (DST), and the Hamilton Depression Rating Scale (HAM-D)—were administered before and after medical therapy to each patient. The VAFT is considered to be the optimal test for assessing verbal fluency. In Japan, the modified VAFT for Japanese is usually used for patients with depression (11). The patients are asked to think of words that begin with a particular Japanese letter and are assessed on the basis of the number of words that they produce in 1 minute. This trial was performed a total of three times with three different letters; hence, it took a total of 3 minutes for each patient to complete the test. The MMSE is a widely used method for assessing global cognitive function (12). It can be used both to screen for cognitive impairment and to estimate the severity of cognitive impairment. Although its results are not diagnostic, it is a sensitive and specific test for dementia and delirium. The DST is helpful for evaluating attention, psychomotor speed, and perceptual organization (13). DST scores reflect the number of correct answers within a timed trial, with higher numbers indicating better performance. The HAM-D is an instrument used for the subjective and objective evaluation of depression (14). It is a screening instrument that is designed to measure the severity of illness in adults who have already received a diagnosis of depression. The HAM-D offers high validity and reliability in measuring response to treatment.
With the VAFT, MMSE, and DST, higher scores indicate better performance, while with the HAM-D, lower scores are desirable. These neuropsychological tests were administered independently by the same senior neurologist (K.N.) and senior psychiatrist (T.M.) who supervised the medical therapy. They were blinded to the findings at MR imaging and proton MR spectroscopy. Each physician examined a portion of the total number of patients. Patients with depressive psychosis were considered to be cured if they showed improved scores on at least two of three tests—the VAFT, DST, and HAM-D—after medical therapy. MMSE results were excluded from the evaluation of cure because this test is not suitable for the assessment of delicate improvements in depression.
Statistical Analysis
To investigate differences in mean age among group A, group B, and the healthy volunteers, a one-way analysis of variance, followed by a post hoc Scheffé test to determine group differences, was performed. To test whether there were significant differences in mean age at onset of depression between group A and group B, an unpaired t test was performed.
To examine whether there were significant differences in mean scores on neuropsychologic tests before and after medical therapy in group A and group B, a paired t test was used. The test determined whether or not the average difference between patients' scores on each neuropsychologic test (VAFT, MMSE, DST, and HAM-D) before and after medical therapy was essentially zero.
A one-way analysis of variance, followed by a post hoc Scheffé test to determine group differences, was used to test whether there were significant differences in Fazekas scores among group A, group B, and the healthy volunteers. The same statistical analysis was performed for the atrophy grading scale scores.
With a paired t test, we examined whether there was a significant difference in mean NAA/Cr ratio between the left and right frontal lobes in groups A and B. To investigate differences in mean Cho/Cr ratio among group A, group B, and the healthy volunteers, a one-way analysis of variance, followed by a post hoc Scheffé test to determine group differences, was performed.
In these analyses, a P value of less than .05 was considered to indicate a statistically significant difference. We performed these analyses with statistical software (SPSS II, based on SPSS 11.0J for Windows; SPSS, Chicago, Ill).
To assess the accuracy of the NAA/Cr ratio, Fazekas rating scale, and atrophy grading scale in predicting prognosis in the patients with depressive psychosis, a receiver operating characteristic curve analysis was performed. For this analysis, prognosis was defined as both good and poor response to medication in the patients with depressive psychosis. We used statistical software (ROCKIT, version 9.1; provided by the Kurt Rossmann Laboratories for Radiologic Image Research, Chicago, Ill) for this analysis. The estimated area under the receiver operating characteristic curve (Az) values reflect the accuracy with which use of each method predicted the prognosis. In addition, sensitivity, specificity, positive predictive value, negative predictive value, and accuracy were calculated for the NAA/Cr ratios. When a patient who had received a diagnosis of depressive psychosis showed a good response to the medication, we considered this to be evidence of a positive diagnosis. When a patient showed a poor response to the medication, we considered this to be evidence of a negative diagnosis. A good response to the medication was considered to have occurred when the patient was cured; all other responses were considered to be poor responses.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
|
Neuropsychologic Test Scores before and after Medical Therapy
Figure 2a shows the pre- and postmedication scores on each neuropsychologic test (VAFT, MMSE, DST, and HAM-D) for patients in group A; Figure 2b shows the same scores for patients in group B. The mean pre- and postmedication VAFT scores, respectively, were 19.6 ± 5.3 and 24.4 ± 6.5 (P = .006) in group A and 16.6 ± 5.7 and 17.2 ± 5.8 (P = .593) in group B. The mean pre- and postmedication MMSE scores, respectively, were 27.7 ± 2.2 and 27.8 ± 1.9 (P = .945) in group A and 26.7 ± 3.1 and 26.7 ± 3.5 (P = .934) in group B. The mean pre- and postmedication DST scores, respectively, were 32.6 ± 9.1 and 41.5 ± 11.4 (P = .004) in group A; in contrast, these scores were 28.6 ± 9.2 and 29.2 ± 8.9 (P = .796) in group B. The mean pre- and postmedication HAM-D scores, respectively, were 9.9 ± 3.7 and 6.0 ± 3.1 (P < .001) in group A and 11.8 ± 3.3 and 11.3 ± 3.8 (P = .590) in group B.
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
These findings indicate that disorder or dysfunction in the frontal lobes may be associated with cognitive impairment in patients with depressive psychosis. In particular, the corticosubcortical loops in the frontal lobe white matter, which connect with the limbic system, are considered to be one of the important targets in elucidating the mechanism of onset of the disease. Bench et al (15) also insisted on the importance of this anatomic network in patients with depressive psychosis. In this context, we believe that measurements of cerebral metabolites at MR spectroscopy could become suitable indexes for evaluating abnormalities in this loop's network, because MR spectroscopic measurements can assess the metabolism of only the axons in the white matter, while fluorine 18 fluorodeoxyglucose uptake and regional cerebral blood flow measurements reflect the combined effect of both axons and glial cells in the white matter. From the point of view of predicting the response to antidepressant medication, estimating the damage in the axon loops is considered useful in predicting the prognosis of patients with depressive psychosis.
NAA is a specific amino acid that exists exclusively in neurons and axons, and its concentration as a neural marker has been well established (17,18). Although the exact mechanism of NAA concentration remains unknown, the NAA concentration is considered to reflect neural density and functioning (18–20). Hence, the NAA concentration measured in the cerebral white matter reflects the density and functioning of the axons alone. In mature brain tissue, neurons cannot proliferate through mitosis, so the reduction in NAA/Cr ratios observed in group B in our study suggests that loss, dysfunction, and/or degeneration of neurons and axons in the frontal lobes has a close relationship with the onset of depression that is resistant to antidepressant medication (2,4,15–16) and occasionally may lead to the structural alterations in brain tissue that many researchers have reported (2,21,22). On the other hand, the total Cho value is thought to reflect predominantly phosphorylcholine and glycerophosphorylcholine (23). Phosphorylcholine and glycerophosphorylcholine are generated by membrane synthesis and the degeneration pathway in myelin and are affected by the clinical condition of the patient and the kind of brain white matter disease (24). Hence, it is noteworthy that no significant differences in frontal lobe Cho/Cr ratios were found among group A, group B, and the healthy volunteers. This indicates that a myelin disorder such as demyelination has little effect on the onset of depression that is resistant to antidepressant medication. With MR spectroscopy, it is easy to assess neurons, axons, and myelin separately. These characteristic MR spectroscopic measurements are also considered suitable for ascertaining the total burden of abnormalities in the brain tissue.
In geriatric depression, the main factor that decreases the frontal lobe NAA/Cr ratio is thought to be CVLs. In group B in our study, the high-signal-intensity area scores on T2-weighted MR images as assessed with the Fazekas rating scale were significantly higher than those in group A, whereas the high-signal-intensity area scores in group A were similar to those in healthy volunteers. Such high-signal-intensity areas in geriatric patients are primarily attributed to leukoaraiosis, which includes CVLs such as microinfarctions, incomplete infarctions, and ischemic changes (25). Some researchers have also pointed out that CVLs have a profound effect on the onset of geriatric depression (2,26,27). Leuchter et al (2) suggested the possibility that CVLs may alter the brain structure in geriatric depression. The results of these studies indicate that the total burden of CVLs has a close relationship with the onset of depression that is resistant to antidepressant medication and that the acquisition of this kind of depression could be assessed by measuring NAA/Cr ratios.
It is interesting to note that, in the present study, there was no significant difference in mean NAA/Cr ratio between the left and right frontal lobes in both group A and group B. These results are in contrast to those of previous studies of fluorine 18 fluorodeoxyglucose uptake and regional cerebral blood flow measurement in which abnormal distribution in the left frontal lobe was observed. This is partly because high-signal-intensity areas are distributed symmetrically. But we would like to focus attention on the threshold effect. Fluorine 18 fluorodeoxyglucose uptake and regional cerebral blood flow are affected not only by the distribution of CVLs but also by focal brain activities, whereas NAA concentration in brain tissue with CVLs may change on the basis of a threshold value that determines whether they are reversible or irreversible. Taking into consideration the reserve function in brain blood vessels, it seems reasonable to suppose that the threshold effect has a close relationship with a reduction in NAA/Cr ratios. It has been proved that regional cerebral oxygen metabolism, which may be associated with NAA concentration, is maintained until regional cerebral blood flow decreases into irreversible ranges (28–30). The threshold theory holds that the total disease burden must exceed a certain fixed level before symptoms begin to appear. In the burden range that is below the threshold, tissues attempt to maintain a stable condition through their reserve function. Many brain disorders, such as cerebral infarction, dementia, Parkinson disease, degenerative diseases, and DNA disorders, have been reported to have a threshold effect (28,31–33). For depression that is resistant to antidepressant medication, some researchers indicate the participation of a threshold effect, providing the concept of vascular depression (34). This effect would support the hypothesis that NAA concentration reflects the threshold value in brain tissue with CVLs. It seems appropriate to apply frontal lobe NAA/Cr ratios as an index in predicting the prognosis after medical therapy in patients with depressive psychosis.
In the brain tissue of patients with depressive psychosis, the total production and supply volumes of serotonin and norepinephrine are thought to decrease. Therefore, in Japan, tricyclic and tetracyclic antidepressants are used as the major drugs for inhibiting the reuptake of these substances. In the raphe of the medulla oblongata, many neurons with neurotransmitters such as serotonin and norepinephrine project axons connecting to the limbic system, basal ganglia, thalamus, and frontal cortex. Thus, if there are brain tissue disorders in the frontal lobe, brain stem, and projected regions, it will be difficult to alleviate depression by administering antidepressant medication alone. We believe that these findings may support the target effect theory that even minimal destruction in important tissues with many functions and neurotransmitters can alter clinical manifestations of brain function in patients. In fact, in the present study, even some patients with high NAA/Cr ratios were revealed to have depression that was resistant to antidepressant medication. Probably, both the threshold effect and the target effect participate in the disease.
Results of the receiver operating characteristic analysis indicated that the NAA/Cr ratios were more useful than the high-signal-intensity area scores and atrophy scaling in predicting prognosis. The main reasons for this finding are as follows: First, an exact estimation of high-signal-intensity area volume and atrophy are very difficult. Second, high-signal-intensity areas and atrophy are affected by both CVLs and non-CVLs. Third, CVLs, which do not appear as high-signal-intensity areas, may participate in psychotic depression, as does, for example, lacunar hemorrhage (35). At this point, MR spectroscopy data are considered superior to high-signal-intensity area scores and atrophy scaling for evaluating CVLs.
Our study had limitations. We do not know exactly how much of an effect lacunar hemorrhage has on the acquisition of depression that is resistant to antidepressant medication, but we suspect that lacunar hemorrhage may have a close relationship with the acquisition of this kind of depression as well as lacunar infarction. It is important to investigate the relationship between lacunar hemorrhage and the frequency of onset of depression with resistance to antidepressant medication. Further studies are needed to clarify the mechanism of the acquisition of depression that is resistant to antidepressant medication.
In conclusion, a group of patients with depressive psychosis who had decreased NAA/Cr ratios showed resistance to antidepressant medication; the reduction in the NAA concentration suggested a disorder of neurons and axons caused by CVLs in the mood regulatory circuits. Clinical application of this MR spectroscopic technique is useful in predicting prognosis in geriatric depression because MR spectroscopy data reflect the total burden of CVLs.
|
FOOTNOTES |
|---|
Abbreviations: Az = area under receiver operating characteristic curve • Cho = choline • Cr = creatine and phosphocreatine • CVL = cerebrovascular lesion • DST = Digit Symbol Test • HAM-D = Hamilton Depression Rating Scale • MMSE = Mini-Mental State Examination • NAA = N-acetylaspartate • VAFT = Verbal Associative Fluency Test
Authors stated no financial relationship to disclose.
Author contributions: Guarantors of integrity of entire study, H. Kado, H. Kimura; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; literature research, H. Kado, H. Kimura, T.M., K.N.; clinical studies, H. Kado, H. Kimura, T.M., K.N.; statistical analysis, H. Kado, I.K.; and manuscript editing, H. Kado, H. Kimura, I.K.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
