Clinical validation of the watershed sign as a marker for neuropsychiatric systemic lupus erythematosus

Authors


Abstract

Objective

To study the relationship between single-photon–emission computed tomography (SPECT) brain imaging and neuropsychiatric signs/symptoms in a cohort of patients with systemic lupus erythematosus (SLE), analyzed using a stereotactic surface projection (SSP) technique.

Methods

Thirty-seven SLE patients were referred for 99mTc–ethyl cysteinate dimer SPECT brain imaging because of neuropsychiatric signs/symptoms. Nineteen normal controls were studied with the identical protocol. Reconstructed images were computed and Z scores were calculated using the SSP technique with the 2-sample t-tests comparing normal controls with SLE patients, and patients with mild cognitive dysfunction with those with severe cognitive dysfunction. The clinical characteristics of SLE patients were collected by retrospective chart review and categorized according to American College of Rheumatology case definitions for neuropsychiatric SLE. Cognitive dysfunction was rated by the treating physician on a scale of 0–3.

Results

Thirty of 37 SLE patients had abnormal SPECT results. SLE patients had reduced perfusion in the watershed areas of the frontal lobes bilaterally compared with controls. Additionally, SLE patients with severe cognitive dysfunction had more severe perfusion deficits than those with mild cognitive dysfunction. In some patients with severe cognitive dysfunction, the watershed areas had Z scores ≥4 SDs below controls.

Conclusion

A convenience sample of patients with SLE and neuropsychiatric signs/symptoms demonstrated reduced perfusion in the watershed areas of the frontal lobes on SPECT scanning analyzed by the SSP technique. The severity of findings correlated with severity of cognitive dysfunction. The area of the brain affected is one that is susceptible to ischemia.

INTRODUCTION

Neuropsychiatric manifestations are commonly observed in patients with systemic lupus erythematosus (SLE) (1). The prevalence varies widely among studies, with such variation likely resulting from study method, differences in demographic characteristics, and the use of different classifications and case definitions of neuropsychiatric disease (2). In 1999 the American College of Rheumatology (ACR) Ad Hoc Committee on Neuropsychiatric Lupus Nomenclature developed case definitions for 19 syndromes in neuropsychiatric SLE (NPSLE) to standardize the nomenclature, including definition, diagnostic criteria, exclusions, and diagnostic testing (2, 3). However, the area remains one of confusion for many clinicians, specifically in terms of determining whether neuropsychiatric symptoms are attributable directly to active lupus as opposed to sequelae of prior disease or manifestations of another concomitant condition such as infection.

In particular, cognitive dysfunction is a common clinical concern in patients with SLE, and the prevalence of this finding in SLE patients has ranged from 11% (4) to 80% (5) using the ACR case definitions published in 1999 (3). Specifically, the clinical significance of cognitive dysfunction has been debated—some studies have found that cognitive dysfunction generally improves or does not worsen over time (6), and other studies have observed that patients who report cognitive dysfunction but do not have overt clinical findings often have objective deficits on clinical testing (7). Furthermore, it is unclear whether cognitive dysfunction is related to active SLE causing direct central nervous system (CNS) injury or is attributable to other causes such as medications, systemic illness, psychiatric disease, metabolic disease, pain, or sleep disturbance (8).

Many imaging modalities are used in the evaluation of NPSLE, particularly in features involving the CNS such as cerebrovascular disease, demyelinating syndrome, headache, and cognitive dysfunction. Imaging modalities include computed tomography (CT), magnetic resonance imaging (MRI), positron emission tomography (PET), single-photon–emission computed tomography (SPECT), and others. Several studies have demonstrated that SPECT can be very sensitive to CNS manifestations of NPSLE (9–14), and the most common finding has been small areas of decreased uptake at multiple sites, suggesting patchy hypoperfusion (15). In addition, our group recently reported preliminary findings of a significant reduction in the frontal watershed areas in patients with SLE with cognitive impairment (16, 17).

The existence of a large base of patients with SLE treated in rheumatology practices at an academic medical center has provided us the opportunity to study patients with SLE for whom a SPECT scan was ordered by their treating rheumatologist to evaluate neuropsychiatric signs and symptoms. We studied the results of brain SPECT scans, analyzed using a stereotactic surface projection (SSP) technique, and correlated the magnitude of the abnormalities with the patient clinical findings.

PATIENTS AND METHODS

Between August 1997 and March 2004, 37 patients were referred for SPECT scans by 3 rheumatologists in single specialty practices at Cedars-Sinai Medical Center. We were able to study these patients, retrospectively, after approval from the Cedars-Sinai Medical Center institutional review board was obtained. Patients carried a diagnosis of SLE, and their treating rheumatologist ordered a SPECT scan to evaluate cognitive dysfunction alone or in combination with other neuropsychiatric signs/symptoms suspected to be caused by SLE. The clinical characteristics of the patients, including the neuropsychiatric signs/symptoms at the time of the SPECT scan, were collected, subsequently, by chart review. The chart reviewer was not aware of the results of the SPECT scans. The signs/symptoms of the patients were classified, according to the ACR NPSLE nomenclature (3), by the chart reviewer, if the treating rheumatologist had not already done so. Additionally, the severity of cognitive dysfunction was rated by the treating rheumatologist as absent, mild, or moderate. This was a subjective rating performed after reviewing the 1999 ACR nomenclature defining cognitive impairment and without knowledge of the results of the SPECT. To assess the severity of cognitive dysfunction, the treating rheumatologist reviewed the progress notes in the chart around the time the SPECT scan was completed, and assigned a score based on his or her understanding of what was written in the record and his or her knowledge of the patient.

Patients underwent a 99mTc–ethyl cysteinate dimer (99mTc-ECD) SPECT brain scan according to guidelines established by the Brain Imaging Council of the Society of Nuclear Medicine (SNM) (18). A 3-detector SPECT camera (Prism-3000; Picker Corporation, Cleveland, OH) was used following 25 millicuries of 99mTc-ECD administered intravenously. Nineteen age-matched volunteer controls from the SNM established normal database who were scanned with identical detectors and scan protocols were used for comparison. Quantitative analysis was performed using an SSP technique (Neurostat; University of Washington, Seattle, WA). For confirmatory purposes, an additional comparison was made with the normals imbedded in the Neurostat program. One advantage of the SSP technique is visualization of the watershed areas of the brain, which are parts of the brain located at the terminal portion of the vascular distribution and are susceptible to ischemia and infarction.

Using the Neurostat program, pixel-wise 2-sample t-test statistics were computed between individual cases comparing each SLE patient with the normal database. The technique utilized a voxel-based analysis comparing patient findings with an age-matched normative database. The results were mapped to a computer atlas of the brain in which a continuous representation of perfusion information was expressed in a color scale related to standard deviation measurements above or below the mean voxel for a specific region. Findings were reviewed by a nuclear medicine physician blinded to clinical information. Reconstructed images were computed and Z scores were calculated comparing the normal subjects with the SLE patients.

RESULTS

Demographics.

All patients in the cohort were female. The mean age at the time of the study was 42 years (range 13–77 years). All patients fulfilled the 1997 ACR revised criteria for the classification of SLE (19).

To estimate the range of severity of SLE in the study group, we examined the frequency of ACR criteria as seen in the patient population. The numbers of patients in the cohort with each clinical manifestation of SLE fulfilling ACR criteria are shown in Table 1. All patients had a positive antinuclear antibody titer during the course of their disease, and the next most common signs were arthritis, oral ulcers, and photosensitivity. Only 2 (5.4%) of the 37 patients had a renal disorder.

Table 1. Number of patients in the cohort with each sign of systemic lupus erythematosus fulfilling the American College of Rheumatology (ACR) criteria*
ACR criteriaNo. (%) of patients
  • *

    All patients fulfilled at least 4 ACR criteria. All patients had a positive antinuclear antibody titer at some point in their disease, and the next most common signs were arthritis, oral ulcers, and photosensitivity. Only 2 patients had a renal disorder.

Antinuclear antibody positive37 (100)
Arthritis27 (73)
Oral ulcers22 (59)
Photosensitivity22 (59)
Immunologic disorder20 (54)
Serositis15 (41)
Malar rash12 (32)
Neurologic disorder (seizure or psychosis)10 (27)
Discoid rash7 (19)
Hematologic disorder6 (16)
Renal disorder2 (5)

Clinical findings.

The CNS neurologic signs and symptoms of the patient cohort at the time of the SPECT are displayed in Table 2. In addition, the numbers of patients with Raynaud's phenomenon and antiphospholipid antibodies were enumerated. All patients had cognitive dysfunction, 27 (73%) had headaches, 21 (56.8%) had mood disorder, 17 (45.9%) had Raynaud's phenomenon, 10 (27%) had antiphospholipid antibodies, 6 (16.2%) had cerebrovascular disease, 6 (16.2%) had anxiety, 5 (13.5%) had seizures, 1 (2.7%) had cranial neuropathy, 1 (2.7%) had demyelinating syndrome, and 1 (2.7%) had myelopathy. Thirty-six patients had multiple signs/symptoms and only 1 patient had a single sign/symptom. Of the patients with cognitive dysfunction, 15 (40.5%) had mild cognitive dysfunction, 12 (32.4%) had moderate cognitive dysfunction, and 10 (27%) had severe cognitive dysfunction, as rated by the treating rheumatologist.

Table 2. Neuropsychiatric signs and symptoms of the patients in the cohort*
Neuropsychiatric sign/symptomNo. (%) of patients
  • *

    These signs and symptoms were classified in accordance with the American College of Rheumatology case definitions for neuropsychiatric systemic lupus erythematosus (3). In addition, the prevalence of Raynaud's phenomenon and antiphospholipid antibodies was included.

Cognitive dysfunction37 (100)
Headache27 (73)
Mood disorder21 (56.8)
Raynaud's phenomenon17 (45.9)
Antiphospholipid antibodies10 (27)
Cerebrovascular disease6 (16.2)
Anxiety6 (16.2)
Seizures5 (13.5)
Cranial neuropathy1 (2.7)
Demyelinating syndrome1 (2.7)
Myelopathy1 (2.7)

SPECT results.

Of the 37 patients, 30 (81.1%) had abnormal scans. As shown in Figure 1, the scans of the SLE patients revealed markedly decreased perfusion in the watershed areas of the frontal lobes bilaterally when compared with the 19 SNM volunteer controls. The Z score shows SDs of reduced perfusion, and in some SLE patients the Z score reductions were ≥6 SDs below normal in the center of the watershed distribution. Similar findings were seen when the scans of the SLE patients were compared with the normal controls embedded in the Neurostat program (data not shown).

Figure 1.

The single-photon–emission computed tomography (SPECT) scans of the patients with systemic lupus erythematosus (SLE) in comparison with the historical normal controls. REF is the reference mapping image. The Z score indicates the SDs of hypoperfusion seen in the SLE patients as compared with the historical normal controls. Marked decreased perfusion was seen in the watershed areas of the frontal lobes bilaterally as compared with the controls. In some SLE patients the Z scores showed ≥6 SDs of reduced perfusion in the center of the watershed areas. GLB = global; RT.LAT = right lateral; LT.LAT = left lateral; RT.MED = right medial; LT.MED = left medial; SUP = superior; INF = inferior.

The comparison of SPECT scans of SLE patients with mild cognitive dysfunction and those with severe cognitive dysfunction is displayed in Figure 2. As shown by the Z scores, patients with severe cognitive dysfunction had more severe perfusion deficits than those with mild cognitive dysfunction. In some patients with severe cognitive dysfunction, the Z score reductions were ≥4 SDs in the center of the watershed areas.

Figure 2.

The single-photon–emission computed tomography (SPECT) scans of patients with mild cognitive dysfunction compared with patients with severe cognitive dysfunction. Patients with severe cognitive dysfunction had more severe hypoperfusion in the frontal watershed areas compared with patients with mild cognitive dysfunction. Some patients with severe cognitive dysfunction had Z scores ≥4 SDs in the watershed areas as compared with the patients with mild cognitive dysfunction. REF is the reference mapping image. RT.LAT = right lateral; LT.LAT = left lateral; SUP = superior; INF = inferior; ANT = anterior; POST = posterior; RT.MED = right medial; LT.MED = left medial.

Of the 7 patients with normal SPECT scans, 2 had mild cognitive dysfunction, 2 had moderate cognitive dysfunction, and 3 had severe cognitive dysfunction. Similar analysis was performed to explore differences between patients with headaches or mood disorder in comparison with all patients with SLE, and differences in perfusion were not found with these other clinical features.

DISCUSSION

This retrospective study of a convenience sample found that patients with SLE and neuropsychiatric signs and symptoms had decreased perfusion in the watershed areas of the brain on SPECT scanning as compared with historical normal controls. It should be pointed out that this study sample is unique and not representative of the entire spectrum of CNS lupus. These patients' signs and symptoms were mild and not well defined compared with patients with more obvious objective findings (stroke, meningitis, psychosis, etc.) who were otherwise treated without undergoing SPECT scans.

Of the 37 patients in the study, 30 had an abnormal SPECT scan. The abnormalities were most prominent in the watershed areas of the brain, which are susceptible to ischemia and infarction. In addition, we found that patients with severe cognitive dysfunction had more severe hypoperfusion than patients with mild cognitive dysfunction. Signs and symptoms other than cognitive dysfunction did not correlate with abnormal SPECT, possibly due to the small number of patients in the study and the fact that most patients had multiple neuropsychiatric signs and symptoms. These results are the first that we are aware of that demonstrate a link between the severity of clinical cognitive dysfunction and the severity of abnormalities on SPECT scanning in NPSLE.

Multiple imaging modalities have been used in NPSLE. CT has limited utility in NPSLE but may be helpful to diagnose large infarcts, hemorrhage, and to exclude confounding diagnoses such as large masses and severe brain edema, or when MRI is not available (20). MRI is the imaging modality of choice in NPSLE because it appears to be very sensitive for infarcts, hemorrhage, masses, brain abscesses, and myelitis, among other abnormalities. MRI findings show that small white matter lesions are the most common abnormality, followed by cerebral atrophy, periventricular white matter changes, ventricular dilatation, diffuse white matter changes, and gross infarct (20). MRI abnormalities are seen in 15–79% of patients with active NPSLE (20). MRI appears not as useful for diffuse manifestations of CNS SLE such as cognitive dysfunction. However, functional imaging, including PET and SPECT, are increasingly being viewed as useful for the evaluation of the diffuse manifestations of CNS SLE, in which CT and MRI can be normal. PET is a radionuclide imaging technique that uses short-lived positron-emitting radiotracers to detect changes in regional cerebral blood flow, glucose metabolism, and brain oxygen consumption, among many functional parameters (15). PET has the advantage of high sensitivity and good correlation with disease outcome but at times lacks specificity and has limited availability (15).

SPECT brain perfusion imaging uses a lipophilic radiopharmaceutical (in this case 99mTc-ECD), which emits single gamma rays and provides information about blood flow and the distribution of radioactive material in the body. SPECT findings have been abnormal in patients with SLE, showing regional cerebral blood flow abnormalities (20). In some studies, SPECT has been observed to be very sensitive in CNS manifestations of NPSLE (9–14), and the most common finding has been small areas of decreased uptake at multiple sites, suggesting patchy hypoperfusion (15). SPECT has also been shown to be more sensitive than MRI for diffuse manifestations of CNS SLE and may be used to monitor disease severity and guide therapy (14). The finding of decreased perfusion in the watershed areas of the brain, an area susceptible to ischemia, is consistent with vasculopathy playing a role in the pathophysiology of NPSLE. Furthermore, SPECT can detect increased hypoperfusion in the brain after a cold stimulus in patients with SLE and Raynaud's phenomenon (21), which leads to the hypothesis that this hypoperfusion in the watershed area may have a vascular mechanism similar to Raynaud's phenomenon in the peripheral circulation.

This study was limited in several important ways because the design was cross-sectional and because the study patients were prone to selection bias. As stated above, patients with more obvious and definable CNS findings were excluded from SPECT scanning by physician preference. Also, although we attempted to describe the neuropsychiatric syndrome according to the ACR case definitions, the suggested diagnostic tests were not completed in all the patients. In addition, the cognitive dysfunction scoring (i.e., absent, mild, moderate, or severe) was completed by the treating rheumatologist retrospectively, and most patients did not have formal neuropsychological testing or consultation from a neuropsychologist. Further limitations include the lack of SPECT in a lupus control group without any neuropsychiatric symptoms; alternatively, historical controls were used. Finally, we are unable to correlate other measures of disease activity (such as a Systemic Lupus Erythematosus Disease Activity Index or British Isles Lupus Assessment Group score) with our findings as an alternate explanation for the results because these data were not collected.

Despite these limitations, the important finding is that there appears to be an association between the severity of hypoperfusion in the watershed region of the brain on SPECT and the severity of cognitive dysfunction in NPSLE. The fact that this has not been previously described may indicate that the surface-rendered images used by the investigators offer an alternative analytic method regarding the gray matter, which may not be readily visualized using conventional cross-sectional evaluation of SPECT images. This is an exciting finding given the complexity of determining whether a CNS event in a patient with SLE is a manifestation of active SLE in the CNS as opposed to being caused by other organ dysfunction or side effects of medication. Although there are significant limitations resulting from the design of our study, this finding of more severe hypoperfusion in patients with more severe cognitive dysfunction provides a testable hypothesis on which future prospective studies can be based.

AUTHOR CONTRIBUTIONS

Dr. Weisman had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Driver, Wallace, Pourrabbani, Waxman, Weisman.

Acquisition of data. Driver, Wallace, Lee, Forbess, Pourrabbani, Waxman, Weisman.

Analysis and interpretation of data. Driver, Wallace, Lee, Pourrabbani, Minoshima, Waxman, Weisman.

Manuscript preparation. Driver, Wallace, Pourrabbani, Minoshima, Waxman, Weisman.

Statistical analysis. Lee, Minoshima, Waxman.

Ancillary