Eosinophilia-myalgia syndrome (EMS) is a multisystem disorder characterized by severe myalgia, along with neuromuscular, cutaneous, and pulmonary features and an increased peripheral blood eosinophil count.1 Patients with EMS experience painful cramps, severe myalgia, fatigue, proximal muscle weakness, peripheral neuropathy, scleroderma-like skin changes, alopecia, and neurocognitive dysfunction.2—4 These symptoms develop at disease onset and persist throughout the illness in most patients (mean follow-up interval, 22.5 months).3 The goal of this study was to examine the relationship between neurocognitive dysfunction and other EMS symptoms.
The development of EMS has been associated with the ingestion of contaminated l-tryptophan—containing products; the contaminants appear to have been introduced during bulk production.5 These contaminants were recently isolated by liquid chromatography as several peaks including "Peak E" molecules, which may be involved in the overactivation of eosinophils.6 Laboratory findings early in the disease include leukocytosis, elevated liver function test results, and elevated immunoglobulin E levels.1 Biopsy specimens of affected tissue show mononuclear inflammatory infiltrate.2 Cerebrospinal fluid findings have been inconsistent, but white matter lesions have been noted on magnetic resonance imaging.7 The pathogenesis of the white matter lesions and cognitive dysfunction in this disorder is not yet understood.
Problems in verbal learning, verbal and visual memory, and motor speed have been documented in neuropsychological studies of EMS.8—11 However, the relationship between cognitive loss and EMS is controversial. Many investigators have speculated that the cognitive problems in EMS are due to nonspecific symptoms such as pain, depression, or impaired sleep. The present study was designed to determine the extent to which the cognitive deficits associated with EMS are related to pain or fatigue. In addition, we sought to replicate prior studies that examined the relationships among sleep disturbance, depression, and cognitive symptoms in EMS.
Twenty-four individuals satisfying the Centers for Disease Control and Prevention criteria for EMS were studied.12 Most patients reported one or more physical symptoms, including myalgia, paresthesia, neuropathy, arthralgia, arthritis, skin rash, and fever. These patients were part of a larger group from a prior study (N=43) that showed deficits in verbal and nonverbal memory relative to healthy control subjects.10 In the present study, all EMS patients completed self-report measures that were correlated with their neuropsychological performance. These questionnaires were added to the testing protocol after the first wave of patients had been treated and given neuropsychological testing.10 Detailed clinical and neuropsychological data on these subjects have been previously reported. Patients with a history of head trauma severe enough to require overnight hospitalization, learning disability, current alcohol or substance abuse, or other concurrent medical or neurologic disorders that could interfere with cognitive functioning were excluded.
The number of subjects using any one class of drug was too small for meaningful statistical analysis, especially since medications were frequently taken in combination. Eleven of the 24 patients were using medications with potential CNS side effects at the time of testing, and 18 patients were using more than one drug. Among patients taking multiple medications (n=18), 8 patients were taking more than two drugs concurrently. The most commonly used medications included antihypertensives, tricyclic antidepressants, narcotic analgesics, nonsteroidal anti-inflammatory agents, steroids, antihistamines, and benzodiazepines.
Patients rated their pain intensity by using a visual analogue scale (VAS) taken from the Short-Form McGill pain questionnaire.13 All study patients also completed the Fatigue Severity Scale (FSS), a nine-item Likert scale scored from 1 (no fatigue) to 7 (severe fatigue) for several measures of fatigue.14 The FSS also asks questions concerning family, social, or work-related difficulties caused by the fatigue, as well as effects of fatigue on physical functioning. Patients' subjective reports of sleep disturbance during the 24 hours prior to the neuropsychological testing session were assessed with the St. Mary's Hospital (SMH) Sleep Questionnaire.15 Responses to the total set of questions were computed for all subjects by using a numerical rating system from 1 (least sleep disturbance) to 6 (maximum sleep disturbance). EMS patients also completed the Center for Epidemiologic Studies Depression Scale (CES-D), a self-report screening measure for depression scored on a scale from 0 to 60.16
The results of neuropsychological testing on this sample of EMS patients have been reported for a previous study.10 The test battery had been chosen to evaluate a range of cognitive functions, with specific emphasis on attention and concentration, verbal and nonverbal memory, and motor speed. Specific tests included the information and vocabulary subtests from the Wechsler Adult Intelligence Scale—Revised (WAIS-R),17 the reading subtest of the Wide Range Achievement test,18 the Raven Progressive Matrices,19 Object Assembly and Block Design subtests of the WAIS-R,17 Trail Making Test,20 Symbol Digit Modalities Test,21, the six-trial version of the Selective Reminding Test,22 the California Verbal Learning Test (CVLT),23 Paired Associates and Logical Memory measures of the Wechsler Memory Scale—Revised (WMS),24 the Benton Visual Retention Test (BVRT),25 and the Finger Tapping Test.26
Because the number of EMS patients who completed the self-report measures for the present study (n=24) was a smaller subset of those completing the neuropsychological test battery in the previous study,10 we repeated the statistical analyses, comparing our smaller sample of EMS patients with healthy control subjects (n=43) from the previous study on these neuropsychological tests. The control subjects were matched in aggregate to the EMS patients within 2 years of age and 1 year of education (T1). To reduce the number of separate statistical analyses in the new data set, several of the dependent variables associated with specific cognitive domains were combined by using a multivariate analysis of variance (MANOVA) strategy. A "Premorbid IQ" MANOVA tested group differences for the scores on the Information and Vocabulary WAIS-R subtests, the WRAT-R Reading subtest, and the Raven Progressive Matrices. An "Attention and Short-Term Memory" MANOVA tested for group differences on the WMS Paired Associate Memory, WMS Logical Memory, total number of words recalled for trials 1—5 on the CVLT and trials 1—6 on the SRT, and the Digit Span WAIS-R subtest. A "Long-term Memory—Verbal" MANOVA tested for group differences on the 20-minute delayed free recall and delayed recognition number correct on the CVLT. A "Concentration and Vigilance" MANOVA tested for group differences on the Digit Symbol and Trail Making (A and B) tests. A "Motor Control" MANOVA tested for group differences on the Finger Tapping Test (dominant hand) and the Finger Tapping Test (nondominant hand). Group differences on the single measure of nonverbal memory (BVRT—total number correct) were tested by using a univariate ANOVA, as were the group comparisons for age and education. Specific neuropsychological tests in which EMS patients performed significantly worse than control subjects were correlated (EMS patients only) with total scores on the SMH sleep questionnaire, FSS, Visual Analog of Pain, and CES-D scales, using Spearman rank-order correlations. In order to reduce the likelihood of statistical (experimentwise) error, the 0.01 level of significance was used when evaluating the results of the Spearman correlations.
We compared our subgroup of EMS patients with healthy control subjects and found that the patients performed significantly worse than the control subjects on tests of long-term memory for verbal material (F=3.25, df=2,63, P=0.04). Follow-up tests using the Tukey highly significant difference correction for experimentwise error showed that this difference was primarily due to impaired long-delay recognition memory on the CVLT (P=0.01). There was also a trend for impaired performance on long-delay free recall on the CVLT (P=0.08). EMS patients also performed significantly worse than healthy control subjects on the test of nonverbal memory (BVRT total number correct; F=5.08, df=1,65, P=0.03). No significant group differences in Premorbid IQ, Age and Education, Attention and Short-term Memory, or motor speed were found.
Correlations: Neuropsychological Deficits and Self-Report Measures
The mean of EMS patients' scores on the FSS was 5.00 (patient range=1.70—7.00; maximum possible FSS score=7.00). The mean of their scores on the CES-D was 14.91 (patient range 0—39; maximum possible CES-D score=60). The mean of their scores on the SMH sleep questionnaire was 19.42 (patient range 0—38; maximum possible SMH score=42). The mean of their scores on the VAS Pain questionnaire was 3.43 (patient range 0—5.80; maximum possible VAS Pain score=7.00).
Spearman rank-order correlations analyzed the relationship between the neuropsychological impairments in our sample of EMS patients and the levels of self-reported depression, fatigue, sleep disturbance, and pain (T2). No significant correlations were found between total score on the FSS, CES-D, or VAS Pain questionnaires and either the CVLT long-delay recognition, CVLT long-delay free recall, or BVRT number correct. Significant negative correlations were observed, however, between total score on the SMH sleep questionnaire and the CVLT long-delay free recall (r=—0.53, P=0.01).
Intercorrelations: Pain, Fatigue, Sleep, Depression
A significant positive correlation was observed between patients' scores on the FSS and the SMH sleep questionnaire (r=0.67, P<0.005). No significant correlations were found between scores on the FSS and VAS Pain questionnaires, the CES-D and SMH sleep questionnaires, or the CES-D and VAS Pain questionnaires.
Because CNS-active medications can potentially impair test performance, patients' medication profiles were also examined. Mean scores on the CVLT long-delay recognition, CVLT long-delay free recall, and BVRT number correct were all better for patients taking CNS-active medications (n=12) than for the patients not taking these types of medications (n=12). It is therefore unlikely that these medications were responsible for the impaired neuropsychological performance. Correlations between patients' SMH sleep questionnaire scores and CNS-active medication profiles were not significant at the P<0.05 level. Spearman correlations between the number of concurrent medications (median=2; range 0—10) and CVLT long-delay recognition, CVLT long-delay free recall, and BVRT number correct were also not significant at the 0.05 level.
To our knowledge, this is the first study that directly addresses the relationship between cognition and symptoms common to most rheumatologic disorders (pain and fatigue). These symptoms were assessed in EMS, a disorder in which cognitive deficits have been documented. This study shows that deficits in verbal and nonverbal memory could not be attributed to the pain or fatigue experienced by these patients. Although pain is a primary complaint and can interfere with neuropsychological test performance in chronic medical illnesses such as cancer,27 pain in EMS does not significantly correlate with any of the neuropsychological impairments.
Sleep disturbance may have a greater impact on neuropsychological test performance than do pain or fatigue. We found that verbal memory scores and sleep disturbance were significantly correlated in our sample of EMS patients. However, nonverbal memory scores were independent of self-reported sleep difficulty. A significant positive correlation was also observed between the patients' scores on the FSS and the SMH sleep questionnaire, suggesting that the previous night's sleep quality is related to daytime fatigue and to cognitive dysfunction. This study's correlational nature and lack of a control group make it difficult for us to draw firm conclusions about this finding. Future studies that experimentally manipulate the degree of fatigue experienced by EMS patients (perhaps by having subjects perform a fatiguing task) might show more precisely the interrelationships of sleep quality, fatigue, and cognition.
As half of EMS patients originally took l-tryptophan because of insomnia,3 sleep complaints are common in this patient group. Sleep loss does have an adverse effect on the recognition and recall of verbal information in healthy populations.28,29 To evaluate the potential role of insomnia as a contributing factor to the cognitive loss in EMS, Murray and Ruff11 compared the neuropsychological function of EMS patients and a group of insomnia sufferers. They found deficits in visuospatial memory in EMS, but not in the insomnia group. Our findings support the conclusion of the Murray and Ruff study11 that verbal memory deficits seen in the EMS group are in part related to poor sleep, but that deficits in visuospatial memory in the EMS group are not the result of insomnia or self-reported sleep disturbance.
Despite a high frequency of pain, fatigue, disrupted sleep, and depression in EMS, the cognitive problems in this patient group cannot be attributed solely to patients' subjective complaints. Specifically, EMS patients' visual memory deficits are independent of self-reported sleep, fatigue, depression, or pain.