| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Nuclear Medicine |
1 From the Departments of Radiology (D.H.L., M.M.G.) and Medicine (D.B., S.A.), University of Washington, Seattle; the Department of Epidemiology and Biostatistics, University of Illinois, Chicago (J.G., M.E.F.); and the Department of Psychiatry and the Rotman Research Institute, University of Toronto, Ontario, Canada (H.S.M.). Received June 29, 2000; revision requested August 10; final revision received November 30; accepted December 4. D.B. supported by National Institutes of Health grant U19 AI38429. Address correspondence to D.B., Harborview Medical Center, 325 Ninth Ave, Box 359780, Seattle, WA 98104.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONSTATISTICAL CONSULTANT...REFERENCES |
|---|
MATERIALS AND METHODS: The authors conducted a co-twin control study of 22 monozygotic twins in which one twin met criteria for CFS and the other was healthy. Twins underwent a structured psychiatric interview and resting technetium 99m–hexamethyl-propyleneamine oxime single photon emission computed tomography of the brain. They also rated their mental status before the procedure. Scans were interpreted independently by two physicians blinded to illness status and then at a blinded consensus reading. Imaging fusion software with automated three-dimensional matching of rCBF images was used to coregister and quantify results. Outcomes were the number and distribution of abnormalities at both reader consensus and automated quantification. Mean rCBF levels were compared by using random effects regression models to account for the effects of twin matching and potential confounding factors.
RESULTS: The twins with and those without CFS were similar in mean number of visually detected abnormalities and in mean differences quantified by using image registration software. These results were unaltered with adjustments for fitness level, depression, and mood before imaging.
CONCLUSION: The study results did not provide evidence of a distinctive pattern of resting rCBF abnormalities associated with CFS. The described method highlights the importance of selecting well-matched control subjects.
Index terms: Brain, abnormalities, 13.899 • Brain, perfusion • Brain, SPECT, 13.12162 • Nervous system, abnormalities, 13.899 • Twins
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONSTATISTICAL CONSULTANT...REFERENCES |
|---|
Single photon emission computed tomography (SPECT) of regional cerebral blood flow (rCBF) is widely available and clinically useful in the assessment of dementia, cerebrovascular disease, and epilepsy (15). SPECT has been used also to assess for CFS (8–14). The results of some (9–12) but not all (13,14) of these studies have demonstrated substantially lower rCBF in patients with CFS compared with that in healthy control subjects. However, definitive statements regarding the association of CFS with changes in rCBF are tempered by the lack of consistent brain region involvement and by various methodologic shortcomings related to image acquisition, data analysis, and subject selection (16).
Regarding subject selection, previous investigations (8) have lacked healthy control subjects; failed to account for coexistent psychiatric morbidities (9,10–12,14), age, and sex (9–13); and not made adjustments for the subjects’ emotional state or medication status at the time of SPECT radiopharmaceutical injection. Some authors (16,17) have emphasized the need to better characterize the sociodemographic background, psychiatric status, and physical fitness of patients with CFS and control subjects in SPECT studies.
In this investigation, we applied a co-twin control method, which is a matched-pair analysis that includes substantial adjustments for many genetic and environmental factors that generally are not considered in typical case-control studies (18). This research design offers a powerful alternative to conventional approaches in which patients with CFS are compared with healthy, depressed, or sedentary control subjects. Our objectives were to investigate the relationship between CFS and cerebral perfusion in monozygotic twins discordant for CFS. Our primary questions of interest were as follows: (a) Do probands with CFS demonstrate a lower resting rCBF than do their unaffected twin counterparts? (b) If so, what is the location and extent of such differences? (c) To what extent are abnormalities the result of psychiatric illness or the mental status just prior to radiopharmaceutical injection?
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONSTATISTICAL CONSULTANT...REFERENCES |
|---|
Psychiatric Disorders
To determine lifetime psychiatric diagnoses, a trained research assistant administered the diagnostic interview schedule, version III-A (20), to the registry participants by means of a telephone interview. The diagnostic interview schedule, which assigns diagnoses on the basis of criteria from the Diagnostic and Statistical Manual of Mental Disorders, version III revised (21), included modules on major depression, dysthymia, generalized anxiety, panic, agoraphobia, posttraumatic stress disorder, mania, bipolar affective disorders, schizophrenia, eating disorders, somatization, and substance abuse and/or dependence. The psychotic symptoms in the schizophrenia section were exclusionary, and melancholia was scored without symptoms attributable to CFS.
Chronic Fatigue Syndrome
CFS criteria were determined according to subject responses to the CFS symptom checklist and the diagnoses generated by using the diagnostic interview schedule. The same inclusion and exclusion criteria (eg, body mass index and certain psychiatric disorders) and review processes were applied to both the twin with CFS and the healthy twin. Thus, a healthy twin with a history of bipolar depression would preclude the study of the twin pair, even if the other twin met the other study eligibility criteria. A physician (D.B.) knowledgeable about CFS reviewed the subjects’ medical records for the past 5 years for potentially exclusionary medical conditions, including head trauma that was recurrent or accompanied by a loss of consciousness that lasted over 5 minutes. A psychologist and an infectious disease specialist also independently reviewed each subject’s medical chart to verify the illness and health status and to approve the twins for participation. Thyroid function test results were obtained for twins who were receiving replacement therapy if the recent results were not available. Before the scheduled visit, we confirmed that the twin with CFS still met the CFS criteria and that the control twin was healthy.
Zygosity
Monozygosity was initially determined by using previously validated self-report methods (22,23) and then confirmed with analysis of restriction fragment length polymorphisms. DNA samples were extracted and digested with the restriction endonuclease HaeIII. The restriction fragments were separated according to molecular size in agarose gel, southern blotted onto a nylon membrane, and hybridized with a variable number of tandem probes. After six probes, the probability of monozygosity can be ascertained with 99.9% certainty (24).
Participant Selection
Monozygotic twins discordant for CFS were chosen for a 7-day in-person evaluation on the basis of registry information and additional telephone screening results. The twin sets were required to (a) be at least 18 years of age; (b) be reared together; (c) be discordant for CFS—that is, one twin met the Centers for Disease Control and Prevention CFS criteria at the time of evaluation, and the other twin was healthy; (d) discontinue consumption of alcohol, caffeine, and all medications known to affect cognition or sleep at least 2 weeks before the evaluation; and (e) travel to the study site together.
Of 193 twin pairs, 119 (62%) were discordant for at least 6 months of fatigue; 67 (56%) of these 119 twin pairs were monozygotic and had complete data available. Of these 67 monozygotic chronic fatigue discordant pairs, 14 were excluded for psychiatric illness; four, for medical disorders; and one, for a body mass index greater than 45. In an additional nine pairs, the fatigued twin did not meet the CFS symptoms criteria, four pairs were excluded because the healthy twin had an exclusionary condition, and six pairs were excluded for a variety of other reasons (eg, recent death of the co-twin, inadequate English, pregnancy). This process left 29 eligible twin pairs in which one twin met strict criteria for CFS and the other was healthy and denied having chronic fatigue. Of these, 22 (76%) twin pairs completed the study, one (3%) pair refused, two (7%) pairs could not be scheduled, and four (14%) pairs were unable to stop taking potentially interfering medications.
SPECT Studies
For orientation purposes, the twins were taken to the nuclear medicine clinic at least 1 day before the procedure. On the day of SPECT, 30 mCi (1,110 MBq) of technetium 99m (99mTc)-hexamethyl-propyleneamine oxime (HMPAO) (Ceretec; Nycomed-Amersham, Princeton, NJ), a radioactive tracer that localizes in the brain in accordance with rCBF, was administered intravenously to both twins simultaneously. SPECT was performed 45–75 minutes after injection of the radiotracer with a triple-headed gamma camera (Prism 3000; Marconi, Cleveland, Ohio) equipped with ultra-high-resolution fan-beam collimators (25). Resting injections were performed with the subject’s eyes open and fixated on a specific point, the ears unplugged, and a low ambient level of noise. Transverse images were oriented by using an axis defined by the frontal and occipital poles for each scan (approximating the bicommissural line). Attenuation-corrected images were produced by using the Chang boundary method with one ellipse per section and a coefficient of 0.11 cm-1 (26). In addition, coronal and sagittal axes were generated. Contiguous images were 6.7 mm thick with a SPECT spatial resolution of approximately 6.3 mm of the system (full width at half maximum, center of field of view in air). The images were produced on black and white film and in warm color (Picker Color; Marconi) on photographic paper for visual reporting.
On the basis of SPECT data, the dependent variables for visual interpretation were defined as rCBF divided into eight areas yielding 16 variables, which were paired for right versus left hemisphere. These were the inferior frontal (orbitofrontal to polar), superior frontal (prefrontal to motor strip), lateral temporal, mesial temporal, basal ganglial, thalamic, parietal, and occipital regions. Two nuclear medicine physicians (D.H.L., M.M.G.) read each scan independently by using a standardized reporting form; this was followed by a consensus reading. The independent and consensus readings both were performed without knowledge of which member of the pair had CFS. The readings consisted of localizing, enumerating, quantifying (ie, small, moderate, or large), and qualifying (mild, moderate, or severe) cerebral perfusion defects.
Each defect was localized by identifying the defect in at least two of the three orthogonal planes and then placing it in a specific region of the brain on the basis of clinical experience and knowledge of anatomy. The two readers (D.H.L., M.M.G.) had 10 and 20 years experience, respectively, in interpreting brain SPECT findings. The consensus reading was considered the final visual interpretation and used for data analysis. (Separate statistical analyses of each reader’s findings did not alter our conclusions, and these results are not presented.)
For quantitative image analysis, an rCBF ratio (ie, brain region to whole brain or brain region to cerebellum) was defined as the mean count per voxel in regional volume divided by the whole brain or the cerebellar count per voxel. The regions comprised seven brain areas: frontal, temporal, parietal, occipital, basal ganglial, thalamic, and cerebellar. To calculate the volume of interest mean count per voxel (a three-dimensional volumetric unit) and the volume of image difference at a threshold of 10% (ie, twin A region/whole brain is greater than twin B region/whole brain by 10% and vice versa), the images were automatically coregistered by using a computer (Hermes; Nuclear Diagnostics, Stockholm, Sweden), overlaid, normalized to the global mean and mean cerebellar value, and then subtracted to calculate the percentage difference.
The imaging fusion software (Brain Registration and Analysis of SPECT Studies [BRASS]; Nuclear Diagnostics) used three-dimensional matching of functional images to coregister and automatically quantify the images obtained in the discordant twin pairs (27,28). BRASS is more precise and less sensitive to noise than surface-matching techniques because it uses entire volumes for registration by means of the simplex minimization algorithm (29). Image preprocessing and registration time is less than 5 minutes and includes thresholding, filtering, shifting, dilation, and rotation. This preprocessing includes segmenting the regions of interest and generating the volumes to be matched, and anisotropic scaling is used to correct for differences in shape. The threshold of a 10% difference was based on the results of test-retest 99mTc-HMPAO SPECT studies at rest in healthy control subjects in whom regional count ratio variability ranged up to 7.8% (30). For each region, we determined the percentage of cerebral and brain stem brain volume that differed by 10% perfusion, normalized to the cerebellum and whole brain.
Additional Measures
As part of our diagnostic interview schedule assessment, we also obtained a diagnosis of major depressive episode. To assess the current mood, participants were given the positive and negative affect schedule (31) within 10 minutes before radiotracer injection. With this instrument, subjects are asked to rate the intensity with which they have experienced 20 emotions or feelings within the last 10 minutes on a scale from 1, which means not at all or very slightly, to 5, which means extremely. The responses yield two scales: a negative and a positive effect. Finally, the physical fitness of all twins was assessed during the intensive clinical examination by using maximum oxygen consumption during a full exhaustive exercise challenge.
Statistical Analyses
At initial descriptive analysis of rCBF, we assessed the percentage of defects determined at consensus reading on a region-by-region and overall basis; the mean numbers of defects in the CFS and healthy twin groups also were estimated. Differences in percentage of defects were assessed by using the McNemar test; mean differences were determined by using the matched-pair Student t test. Raw mean defect counts per voxel and counts adjusted to the cerebellum were calculated separately for the left and right hemispheres in seven brain regions in the CFS and healthy twin groups. These data were analyzed by using a multivariate random effects regression model (32,33) to assess the association between CFS status and rCBF in all regions of interest. This model is especially well suited for analysis of twin data because it accounts for both the paired (twin with CFS and healthy twin) and repeated measurement structure of the data (multiple regions and two hemispheres). The random effects model avoids the problems inherent in piecemeal hypothesis testing for the effects of CFS status in each region and hemisphere.
The models we used included random effects for the pair and the regions and hemispheres within an individual. Indicator variables were entered for CFS status (CFS vs healthy), hemisphere (right vs left), and region (seven regions for the whole-brain normalized data and six regions for the cerebellum normalized data). Two- and three-way interactions among the three class variables were evaluated. Adjustments were made in the random effects model for lifetime depression, positive and negative affect schedule depressed and anxious mood, and maximum oxygen consumption during the exhausting exercise challenge. Intraclass correlations were calculated by using the variance estimates from the best-fitting random effects regression model. We used BRASS software, which automatically coregistered the SPECT images to each other, and a 10% difference threshold (normalized to the cerebellum and whole brain) to predict the twin with CFS on the basis of the hypothesis of less rCBF in the affected twin. Analyses were conducted by using a software program (SPSS for Windows, version 6.2; SPSS, Chicago, Ill) and the MIXREG (SAS Institute, Cary, NC) program (32).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONSTATISTICAL CONSULTANT...REFERENCES |
|---|
Visual Reporting of SPECT Findings
As shown in Table 1, there were no significant differences in rCBF between the CFS and healthy twin groups at visual inspection; specifically, the rCBF was not lower in the affected proband. The largest difference was found in the mesial temporal lobe, but even there, the hypoperfusion observed in the affected twin was not significantly different from the rCBF in the healthy twin (P = .11). It is worth noting that other investigators (10,11) have observed perfusion defects in the mesial temporal lobe to be the most common visual abnormalities in the healthy control subjects.
|
|
|
When data were normalized to the whole brain, the quantitative model predicted the twin with CFS in 11 (50%) of 22 pairs. In addition, we found no differences in the mean percentage of brain volume that differed by the 10% perfusion threshold for both whole-brain and cerebellar normalizations. When normalized to the whole brain, the mean percentage of brain volume that differed by at least 10% when the twin with CFS had less perfusion was similar to that when the healthy twin had less perfusion (14.6% vs 13.1%, P = .74). Likewise, when normalized to the cerebellum, the mean percentage of brain volume that differed by 10% or more when the twin with CFS had less perfusion was similar to that when the healthy twin had less perfusion (16.9% vs 19.0%, P = .75).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONSTATISTICAL CONSULTANT...REFERENCES |
|---|
The findings of several imaging techniques have provided evidence for structural or functional cerebral abnormalities in patients with CFS. In an early report of an outbreak of a chronically fatiguing illness, magnetic resonance (MR) imaging depicted punctate lesions in the high convexities of 70% of patients compared with in 20% of healthy individuals (35). In one investigation (36), a substantially larger number of small, punctate, subcortical white matter hyperintensities (predominantly in the frontal lobes) were detected in patients with CFS—compared with the number of these findings in the healthy control subjects—but only among the subgroup without comorbid psychiatric disease. However, functional brain SPECT appears to depict more abnormalities in CFS than does anatomic MR imaging (10) and 2-[fluorine-18]fluoro-2-deoxy-D-glucose (FDG) positron emission tomography (PET) (8).
Prior investigations (9–14) of rCBF SPECT in CFS have focused on perfusion in the cortex and brain stem. In one study, SPECT imaging findings were abnormal in 81% of subjects but in only 21% of healthy adults; substantially more cortical defects were noted in the CFS group (7.3 vs 0.4). The majority of defects—defined as a greater than 1 cm area spanning the full thickness of the cortex, with less than 60% of maximal uptake—were in the lateral frontal and lateral and mesial temporal lobes (10). In a companion article (11), SPECT findings were compared in patients with CFS, healthy individuals, and adults with acquired immunodeficiency syndrome dementia and unipolar depression. The patients with CFS had a mean of 6.5 defects per subject, whereas the healthy control subjects had a mean of 1.7 defects. However, the distribution of defects was similar in all disease comparison groups, with the findings located primarily in the frontal and temporal cortices. Although the investigators could distinguish CFS from acquired immunodeficiency syndrome encephalopathy by comparing regional perfusion deficits, differentiation between CFS and major depression was not possible. It is notable that the patients with CFS were not examined for psychiatric disease (11).
Similarly, other reports (14) on SPECT have not been controlled for high levels of depression among patients with CFS. Thus, in these studies, major depression in patients with CFS could have accounted for the observed results. Other investigators (12) have compared brain stem perfusion measurements by using SPECT and found reduced rCBF in the CFS group (0.71 brain stem counts to cerebellar counts) compared with that in the healthy control (0.80) and depressed control (0.77) groups. Results of other studies (8,9) have shown similar areas of cortical defects; however, no pattern predictive for CFS was demonstrated. One report (13) showed no difference in rCBF between healthy control and CFS groups.
There are several methodologic issues of concern regarding these prior reports, including subject selection, procedure conditions, and medication use. In one study (11), almost a quarter of the patients with CFS had experienced an acute neurologic event not typical of CFS, such as seizure, transient visual impairment, or temporary weakness. In addition, descriptions of the injection conditions and medication status of the subjects at the time of the study are infrequent (9–12). These are important considerations because mental state (eg, sadness) (37) and medication use (38,39) may affect cerebral perfusion. In some studies (9–12), images were obtained 10–30 minutes after radiopharmaceutical injection, a time when 99mTc-HMPAO is clearing from the background and cerebral blood volume (40), instead of at least 45–60 minutes after injection when there is less disequilibrium of cranial activity (25). In addition, application of the Chang method for attenuation correction (26) (eg, one ellipse per section vs one ellipse per study) often is not described. Furthermore, the use of one operator or observer for visual inspection of the data and the quantitative region of interest may lead to unknown biases (12). Finally, the finite spatial resolution of the SPECT system and partial volume effects at the brain stem region also are potential limitations of this method (12).
Because SPECT scans can be influenced by many factors, the selection of both patients and control subjects is particularly critical. In prior works, control subjects were described as well-adjusted healthy volunteers, without adequate reference to screening procedures to detect medical conditions (9,12–14), assess the neuropsychiatric and/or psychologic status (9–14), or determine medication use. In one study (11), healthy control subjects were selected from spouses of subjects with disease (11). In our study, the healthy twin was identical in age, sex, and family history to the affected twin. Furthermore, all twins had been free of medications and substance use for at least 2 weeks before imaging, and the pairs had undergone scanning within 1 hour of each other.
Such considerations are crucial, because lighting, stimulation, emotional state, drugs, and genetic factors all affect rCBF, and this may explain why our findings differ from those previously reported. Considering the low rate of abnormal SPECT findings in the twins with CFS in our study, it may be that the requirement for travel and our extensive testing yielded a less affected population. On the other hand, the age, sex, and mean fatigue duration in our CFS group suggested a cohort that was typical of previously described CFS sample populations.
One aspect of this study that deserves mention is our use of illness discordant twins. Neuroimaging techniques such as MR imaging have previously demonstrated cerebral manifestations of systemic lupus erythematosus (41) and multiple sclerosis (42) in both members of disease discordant twin sets. In the latter report, 75% of unaffected monozygotic twins but none of unaffected dizygotic twins had lesions. Because findings were more commonly demonstrated in the affected twin, this type of result may be associated with disease vulnerability. Likewise, among monozygotic twins discordant for schizophrenia, the affected twin almost invariably has a smaller hippocampus at MR imaging and lower regional metabolism at PET of the prefrontal cortex than does the healthy twin (43,44).
Although both twin groups in our study had few abnormalities at SPECT, a notable finding was the striking similarity of cerebral perfusion between members of a pair, regardless of health status. Similarly, in an FDG PET study, the cerebral glucose metabolism in healthy monozygotic twin pairs was similar in several regions, including the prefrontal cortex, which is involved in attention, vigilance, and integrative functions (45). Because it has been previously observed that the brain volume in monozygotic twins is highly correlated (46,47), digital techniques that enable matching of SPECT image volumes should perform optimally in monozygotic twins (27,28). We believe that our co-twin control design provided excellent control for background variation in rCBF and thus maximized our ability to detect potential abnormalities associated with CFS.
This co-twin control study had several limitations. First, the method used to recruit the sample population was not ideal. Solicitation by advertisement resulted in a volunteer sample of twin pairs with the potential for ascertainment problems. However, to our knowledge, the more desirable strategy of systematically identifying twins from a well-defined, population-based twin registry cannot be readily accomplished in the United States. Thus, how representative the twins in this study were of either twins in general or persons with CFS is not known. Because the twins in this study were adults—primarily women—recruited from community practices and the affected twins met strict criteria for CFS, our results may not be generalizable to other sample populations and more specialized settings. Nonetheless, the demographic and clinical characteristics of our sample population were similar to those previously described.
A final caveat is that passive resting examinations of rCBF may be insensitive to disorders like CFS, in which abnormalities may manifest only when active, challenging, or stressful cognitive or physical tasks are performed. For example, results of cognitive challenge tests with the paced auditory serial addition task have shown poorer performance in subjects with CFS, as compared with healthy control subjects (7). Likewise, in a study of 99mTc-HMPAO SPECT performed in 20 patients with diabetes mellitus, decreased activation of the right anterior and left posterior cingulates has been demonstrated during performance of the paced auditory serial addition task compared with that at resting rCBF (48). In a small study (49), physical exertion, as compared with the resting state, increased perfusion abnormalities at SPECT in 60% of individuals with CFS and in 10% of healthy subjects. These stress-induced perfusion abnormalities were of small to medium size at visual impression and preferentially located in the left temporoparietal regions. The results of these studies considered together suggest that challenge or stress tests combined with SPECT may be more sensitive than standard resting SPECT techniques.
In summary, the results of this co-twin control study with 22 monozygotic twins discordant for CFS failed to enable differentiation between the affected and nonaffected twins and thus do not support a major role of resting rCBF abnormalities in CFS. Because twin studies are especially well suited for investigations of illnesses for which the appropriate comparison groups are not clearly defined (18), they offer an alternative approach to examining cerebral perfusion abnormalities that have been associated with CFS in previous studies. Future research should focus on constructing "real life" situations, such as stressful paradigms, and involve the use of either SPECT or functional MR imaging to elucidate the cognitive complaints reported by individuals with CFS. Future investigations will need to also involve larger groups of patients and control subjects and the use of automated and operator-independent analytic programs. In conclusion, the present study results do not provide evidence of a distinctive pattern of resting rCBF abnormalities associated with CFS. Resting rCBF SPECT does not appear to be a useful examination for the diagnosis or confirmation of CFS.
|
STATISTICAL CONSULTANT COMMENTARY |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONSTATISTICAL CONSULTANT...REFERENCES |
|---|
Furthermore, although the left and right hemispheres are fixed in all individuals, the regions are not. That is, measurements in all the individuals were not performed in the same region. In other words, the regions chosen within an individual are a random sample from many possible regions. Therefore, the regions chosen are specific (or nested) in an individual. This also must be carefully considered in the analysis.
The authors’ choice of the multivariate random effects regression model can account for both the correlations and the individual specific regions. In addition, the authors used this model to include many other covariables and interactions. This model, which enables one to test all hypotheses simultaneously while controlling for an overall significance level for the study, was the best choice.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Author contributions: Guarantor of integrity of entire study, D.B.; study concepts and design, D.H.L., H.S.M., J.G., D.B.; literature research, D.H.L., H.S.M., D.B.; clinical studies, D.H.L., D.B.; data acquisition, D.H.L., S.A., D.B.; data analysis/interpretation, D.H.L., H.S.M., M.E.F., J.G., M.M.G.; statistical analysis, M.E.F., J.G.; manuscript preparation, D.H.L., H.S.M., M.E.F., J.G., D.B.; manuscript definition of intellectual content, D.H.L., H.S.M., J.G., D.B.; manuscript editing, D.H.L., H.S.M., J.G., M.M.G., D.B.; manuscript revision/review, D.H.L., M.E.F., J.G., D.B.; manuscript final version approval, D.H.L., H.S.M., M.M.G., D.B., J.G., M.E.F.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONSTATISTICAL CONSULTANT...REFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  S. A. Spence All in the mind? The neural correlates of unexplained physical symptoms Advan. Psychiatr. Treat., September 1, 2006; 12(5): 349 - 358. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. Afari and D. Buchwald Chronic Fatigue Syndrome: A Review Am J Psychiatry, February 1, 2003; 160(2): 221 - 236. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. B. Schmaling, D. H. Lewis, J. I. Fiedelak, R. Mahurin, and D. S. Buchwald Single-Photon Emission Computerized Tomography and Neurocognitive Function in Patients With Chronic Fatigue Syndrome Psychosom Med, January 1, 2003; 65(1): 129 - 136. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
