SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. REFERENCES

Objective

To investigate changes in brain activation patterns detected by functional magnetic resonance imaging (FMRI), and the relationship between FMRI activation patterns and results of formal neuropsychological testing, in patients with childhood-onset systemic lupus erythematosus (SLE).

Methods

Ten patients with childhood-onset SLE underwent formal neuropsychological testing and FMRI using 3 paradigms: a continuous performance task (CPT) to evaluate attention, an N-Back task to assess working memory, and verb generation to evaluate language processing. Composite Z maps were generated to summarize the brain activation patterns for each FMRI paradigm in patients with childhood-onset SLE and to compare these patterns with those observed in healthy controls. Between-group comparison Z maps showing differences in activation between childhood-onset SLE patients and controls were generated, using a significance level of P < 0.05 in a general linear model.

Results

Compared with the control group, the childhood-onset SLE group showed statistically significant increased activation of brain areas involved in the CPT, N-Back, and verb generation tasks. In contrast, in the absence of active stimulus, e.g., during times of the paradigm control tasks, childhood-onset SLE patients consistently undersuppressed activity in the expected brain areas. Activation in selected cortical areas was found to correlate negatively with results of a subset of individual neuropsychological test scores.

Conclusion

FMRI abnormalities are present in childhood-onset SLE, manifesting as an imbalance between active and inhibitory responses to an array of stimuli. Differences in brain activation patterns compared with those observed in controls suggest that childhood-onset SLE may be associated with abnormalities in white matter connectivity resulting in neuronal network dysfunction, rather than injury of specific gray matter areas.

Childhood-onset systemic lupus erythematosus (SLE) is among the most severe pediatric rheumatic diseases (“pediatric” defined as onset prior to age 16). Often diagnosed in US females of minority populations, SLE is associated with significant morbidity and at least 10 times higher mortality than that of the age-matched general population (1). The reported prevalence of neuropsychiatric involvement in SLE (NPSLE) varies between 15% and 95% (2, 3), and children with SLE tend to exhibit a more severe phenotype than adults (4).

Neuropsychiatric involvement is arguably the least understood manifestation of SLE and is associated with a complicated range of clinical presentations (5). NPSLE syndromes range from diffuse central nervous system (CNS) disorders (e.g., acute confusional state, psychosis, anxiety, and depressive disorders) including cognitive dysfunction to focal CNS syndromes (e.g., seizures, cerebrovascular disease, chorea, myelopathy, transverse myelitis, demyelinating syndrome, aseptic meningitis, headaches) and peripheral nervous system disorders (e.g., neuropathies, acute inflammatory demyelination) (6).

Cognitive dysfunction is likely second only to headache as the most commonly encountered symptom of NPSLE. Cognition impairment in NPSLE can affect any or all of the following functions: simple or complex attention, reasoning, executive skills (e.g., planning, organizing, sequencing), memory (e.g., learning, recall), visual-spatial processing, language (e.g., verbal fluency), and psychomotor speed (7–9). The severity of cognitive dysfunction in NPSLE ranges from mild impairment to severe dementia. Currently, formal neurocognitive testing is considered the criterion standard for diagnosing cognitive dysfunction in SLE (6), but the etiology of this complication remains elusive and continues to be an active research question.

Neuroimaging is a potential means to noninvasively assess brain pathology in NPSLE. Anatomic magnetic resonance imaging (MRI) has commonly revealed lesions in the brain tissue of individuals with NPSLE, but these findings have not been effective markers of active NPSLE, nor have they explained the specific cognitive changes encountered (3). While evidence of cortical and subcortical neuropathology has been demonstrated in MRI studies using magnetization transfer imaging and fluid-attenuated inversion recovery (10, 11), no clear association of these findings with clinical symptoms that could be considered when making or evaluating treatment decisions has emerged. Likewise, limited insight into patterns of cognitive deficit in NPSLE has been provided with other imaging methods, such as magnetic resonance spectrometry (10, 12, 13), positron emission tomography (PET) (14, 15), and single-photon–emission computed tomography (SPECT) (16, 17).

Functional MRI (FMRI) is a promising method for recording brain activation patterns associated with specific cognitive tasks, including patterns that could be related to the neuropathologic basis of NPSLE. Standard FMRI studies involve acquisition of serial images while the subject alternates between performing active and control tasks (FMRI paradigms). The image intensity is weighted by the relative oxygenation level of blood hemoglobin (blood oxygenation level–dependent [BOLD]). Contrast between images obtained during active and control task periods of a paradigm reflect changes in regional brain activity. Although FMRI has been helpful in delineating changes in brain activity in various chronic diseases involving cognition (18–20), its usefulness for studying brain pathology in SLE, particularly of childhood onset, has not been well explored.

In this pilot study, activation patterns on FMRI were investigated in patients with childhood-onset SLE. In addition, the relationship between FMRI activation patterns and results of formal neuropsychological testing was explored.

PATIENTS AND METHODS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. REFERENCES

Patients.

From a larger cohort study assessing the prevalence of cognitive dysfunction in pediatric SLE, 10 patients with childhood-onset SLE were selected for FMRI. Results of formal neuropsychological testing suggested the presence of cognitive dysfunction in 6 of the 10 patients and normal cognition in the remaining 4. The designation of cognitive dysfunction was made based on a score below the normative mean (Z score of −2 or less) on at least 2 tests in the battery (see below). The patients with childhood-onset SLE were all female and had a mean ± SD age of 17.3 ± 3.4 years (range 12–21 years), with disease onset at age <16 years. All patients with normal cognition, but only 3 of the 6 patients with cognitive dysfunction, had tested positive for antiphospholipid antibodies in the past. No information was available regarding the presence or absence of antineuronal and antiribosomal P antibodies. None of the childhood-onset SLE patients were assessed as having seizures, psychosis, headache, or vasculitis, as defined according to the SLE Disease Activity Index (SLEDAI) (21,22). In addition, all patients scored 0 in the neuropsychological category of the Systemic Lupus International Collaborating Clinics/American College of Rheumatology Damage Index (23). There was a trend toward a shorter median disease duration among childhood-onset SLE patients with neurocognitive dysfunction versus those with normal cognition (2.5 years [range 1–6 years] versus 3.4 years [range 3–10 years]), but the difference was not statistically significant. Sociodemographic and clinical characteristics of the patients are shown in Table 1.

Table 1. Characteristics of the patients with childhood-onset SLE*
PatientAge, yearsMedian family income, US dollarsSLEDAI scoreSDI scoreaPLNeuropsychiatric drugsDaily prednisone dosage, mgCognitive dysfunction found on formal neuropsychological testing
  • *

    SLE = systemic lupus erythematasis; SLEDAI = SLE Disease Activity Index; SDI = Systemic Lupus International Collaborating Clinics/American College of Rheumatology Damage Index; aPL = antiphospholipid antibodies.

11256,87500No0Yes
22127,29600No15Yes
32160,19100No5Yes
41641,98820+No5No
51846,11740+No25No
61561,64280+No5Yes
71935,02060+Fluoxetine HCl5No
81954,91641+Fluoxetine HCl15Yes
92050,70340+No5No
101265,89320+No10Yes

The study protocol was approved by the Institutional Review Board (IRB) of Cincinnati Children's Hospital Medical Center. Written consent was obtained from all adult patients and from the parent/legal guardian of all minor patients, prior to participation in the study. In addition, written assent was obtained from minor patients participating in the study.

Formal neuropsychological testing.

All childhood-onset SLE patients were administered a battery of standardized, validated, age-appropriate neuropsychological tests by a trained psychometrician, to assess various domains of cognition including visuoconstructional processing, memory, psychomotor speed, and attention/executive functioning. Visuoconstructional processing was tested using the Copy Trial of the Rey-Osterreith Complex Figure (24). Memory was assessed via the Stories Subtest of the Children's Memory Scale (25) or the Logical Memory Subtest of the Wechsler Memory Scale III (26), depending on age (Immediate and Delayed Recall Trials), as well as the Rey-Osterreith Complex Figure Immediate Recall and Delayed Recall Trials (27). The Trail Making Test Part A (28) and the Color/Word Naming and Reading Subtests of the Delis-Kaplan Executive Function System (29) measured psychomotor speed performance, while the Trail Making Test Part B (28) and the Color/Word Interference and Substitution Subtests of the Delis-Kaplan Executive Functioning System evaluated attention/executive functioning. The neuropsychological test battery is summarized in Table 2. Z scores indicating the deviation of each test score from the normative mean for age were subsequently correlated with FMRI activations on paradigms designed to exercise similar cognitive domains.

Table 2. Neuropsychological test battery administered to the patients with childhood-onset systemic lupus erythematosus
Principal cognitive domain tested, neuropsychological test(s)Applicable age
Visuoconstructional processing 
 Rey-Osterreith Complex Figure Copy TrialAll ages
Memory 
 Children's Memory Scale Stories Subtest, Immediate and Delayed Recall Trials12–16 years
 Wechsler Memory Scale III Logical Memory Subtest, Immediate and Delayed Recall Trials17–21 years
 Rey-Osterreith Complex Figure Immediate and Delayed Recall TrialsAll ages
Attention/executive functioning 
 Delis-Kaplan Executive Function System Color/Word Interference and Color/Word Interference/Substitution SubtestsAll ages
 Trail Making Test, Part BAll ages
Psychomotor speed 
 Delis-Kaplan Executive Function System Color/Word Naming and Color/Word Reading SubtestsAll ages

FMRI paradigms.

All childhood-onset SLE patients performed 3 FMRI paradigms, and findings were compared with those obtained in previously imaged healthy controls. The 3 FMRI paradigms selected for this preliminary study probe 3 distinct cognitive domains previously found to be impaired in NPSLE (7). All 3 FMRI paradigms were of a block periodic design, with 30-second “active” task intervals interleaved with 30-second “control” task intervals. The control task served to compensate for “noise” and baseline brain activity.

Prior to the FMRI scan, each patient was trained on the FMRI tasks outside of the magnet. For each task, this involved a short training session using a Macintosh desktop computer running the MacStim paradigm presentation software (version 3.2.2; WhiteAnt Occasional Publishing, Melbourne, Victoria, Australia). The tasks were sequentially explained and then demonstrated during this prescan training session. Audio and video stimuli were presented via loudspeakers and/or a computer screen, and button responses were practiced. The training session lasted ∼15 minutes for all 3 tasks. Following satisfactory performance during the training sessions, the patients were given instructions specific to performance of the tasks in the MRI scanner. Patients were then encouraged to visit the rest room prior to entering the MRI scanner. Once in the scanner, they were given further instructions using an MRI-compatible intercom and audio system. They were instructed to remain still and also were reminded about the requirements for each task immediately before each one was administered. The procedures used for prescan preparation and desensitization in children have been further described by Byars et al (30). Details of each of the FMRI paradigms are presented below.

Language paradigm.

The verb generation task is based on a task initially developed for PET imaging to assess language function (31) and adapted by investigators at our institution for use with FMRI in children (32–34). The active task is administered by auditory presentation of a series of concrete nouns. The patient hears a noun every 5 seconds during 30-second active blocks and is required to covertly generate as many verbs associated with each noun as possible during each corresponding 5-second interval. For example, if the noun “ball” were presented, the subject might generate the verbs “throw,” “kick,” and “hit.” The subject was instructed to think of the verbs without articulating them, in order to minimize motion artifacts associated with speech.

Bilateral finger tapping is used as the control task for the verb generation paradigm. During the 30-second control blocks, the subject is asked to tap his or her fingers each time a target tone is heard. Presentation of the target tone occurred every 5 seconds to parallel the initiation of response to nouns during the active task. This control task accomplished 3 objectives: 1) to control for the auditory stimulation present in the verb generation block, 2) to prevent the subject from continuing to generate verbs during the control block, 3) and to provide a reference area of activation within the motor strip as an independent means of validating subject compliance.

Attention paradigm.

Attention may be broadly defined as the ability to concentrate on a specific stimulus over a period of time, to the exclusion of extraneous stimuli (35, 36). A continuous performance task (CPT) was utilized to measure attention in this study. At its simplest, a CPT may consist solely of the subject responding to a particular item or pattern of items in a series of data. The identical pairs version of the CPT (CPT-IP), consisting of identifying the repetition of any item in a sequence, is somewhat more complicated (37, 38), thus demanding a higher processing load. The specific version of the CPT-IP used in this study (39) consisted of an active attentional task which proceeded by showing a random single digit between 0 and 9 at the center of a screen every 0.75 seconds. Subjects were instructed to press a button whenever consecutive numbers were identical.

During the control task interval, the number 1 was shown repeatedly with the same 0.75-second period. The subject was asked to simply press the response button 5 times at the beginning of this interval, with no further response required. Contrast between the attention and control tasks in this paradigm minimizes motor response from pressing buttons as well as visual stimulation associated with watching the numbers. The primary contrast between the phases of these tasks therefore lies in the attention needed to detect sequential pairs of numbers as they appear on the screen.

Working memory paradigm.

Working memory may be defined as the portion of short-term memory in which information is maintained and made available for use for a brief time period (40). During a paradigm referred to as N-Back, subjects performed an attention “0-back” control task and a “2-back” working memory task. Delayed response tasks of this kind have been widely used to study working memory (41, 42). In both tasks, the integers from 1 to 4 were presented singly, on a screen with each number appearing consistently in a corresponding quadrant of a diamond. Subjects responded by pressing buttons on a response device on which buttons had been arranged in a diamond to correspond spatially with the numbers appearing on the screen. Each response, including time stamp, was recorded via computer interface. Seventeen numbers were presented at a steady rate during each 30-second block. The 0-back control task required the subject to merely press the button corresponding to the number currently showing on the screen. During the 2-back working memory block, subjects were instructed to press the button corresponding to the number that preceded the current number by 2 intervals. Since both tasks have the same sensorimotor and attentional elements, the contrast between them isolates the working memory component exclusive to the 2-back task.

Imaging.

FMRI scanning was performed on a Bruker Biospec 30/60 MRI scanner with a 60-cm, 3.0T magnet (Bruker BioSpin, Ettlingen, Germany). A T2-weighted gradient-echo echo-planar imaging sequence was used for all FMRI scans, with the following parameters: repetition time (TR) 3,000 msec, echo time (TE) 38 msec, field of view (FOV) 25.6 × 25.6 cm, matrix 64 × 64 pixels, 24 axial slices, slice thickness 5 mm. Throughout alternating 30-second control and active blocks for each task, a 24-slice image covering the brain volume was acquired for every TR.

In addition, an inversion-prepared 3-dimensional (3-D) modified driven equilibrium Fourier transform (43, 44) whole brain scan was acquired for each subject. Parameters for this scan were as follows: TR 15 msec, TE 4.5 msec, inversion recovery time 550 msec, FOV 25.6 × 19.2 × 19.2 cm, matrix 256 × 96 × 96 pixels (total time 5 minutes, 16 seconds). These MRI scans served as T1-weighted 3-D anatomic reference images for coregistration and overlay of functional data.

Each imaging session for the childhood-onset SLE patients began with a pilot image to position the patient, followed by acquisition of a reference scan for artifact removal in postprocessing. The verb generation paradigm was then performed, followed by the anatomic imaging. The session continued by completion of the CPT-IP and N-Back paradigms. This task order was fixed in the childhood-onset SLE patients recruited for the pilot imaging study. However, the data on control subjects were borrowed from several other studies that used the same FMRI paradigms incorporated in this study (32, 41, 45); consequently, there was variation in the order of tasks between the childhood-onset SLE and control groups. Previously we have found that task order does not influence motion or performance on FMRI tasks in pediatric subjects. Therefore, we do not believe the difference in the task order between groups would introduce significant confounding in this study.

A series of 110 images (total time 5 minutes, 30 seconds) was acquired for the verb generation task paradigm, while the CPT-IP and N-Back tasks each required 121 images (total time 6 minutes, 3 seconds). Data from the initial 10 time points were discarded from the imaging series of each paradigm to allow for attainment of T1 relaxation equilibrium.

FMRI data on healthy controls.

Data on healthy controls were obtained from investigations of brain function utilizing identical FMRI paradigms. All control data were acquired using the same 3T Bruker scanner, with imaging methods that corresponded exactly to those used for the childhood-onset SLE patients. For each study generating control data, IRB approval and prior written consent of all subjects or their parent/legal guardian were obtained.

Controls for the verb generation task were 10 healthy female subjects with a mean ± SD age of 16.3 ± 2.5 years (range 12–18 years). Controls for the CPT-IP task were 10 healthy female subjects with a mean ± SD age of 14.7 ± 1.7 years (range 12–17 years). The ages of the N-Back controls (5 male, 5 female) ranged from 14 to 22 years, with a mean ± SD of 18.3 ± 3.4 years.

Differences in the mean values of available measures of socioeconomic status between each group of control subjects and the childhood-onset SLE patients were not significant (P > 0.05 by Mann-Whitney U test). The mean ± SD of the median family income (zip code data, based on the year 2000 US Census) in the childhood-onset SLE group, verb generation control group, and N-Back control group was $50,064 ± $12,396, $54,702 ± $9,201, and $51,249 ± $11,692, respectively; income data were available on only 4 of the N-Back control subjects. CPT-IP controls reported their family income on questionnaires (mean ± SD midrange $63,250 ± $24,609).

Statistical analysis and image processing.

Processing of 3-D anatomic and FMRI imaging data was done using Cincinnati Children's Hospital Image Processing Software (CCHIPS), an IDL-based program (ITT Visual Information Solutions, Boulder, CO) (46). Prior to statistical parameter mapping of activation patterns during application of each of the FMRI paradigms, several preprocessing steps were completed. Imaging artifacts (ghosting, geometric distortion, truncation, motion) were removed using appropriate, well-established techniques (47, 48). Compensation for subject motion was achieved by aligning all images in an FMRI series to the first image of the series (49). CCHIPS was also used to transform the image data into the Talairach reference frame (50). This is achieved by first identifying fiduciary brain structures, such as the anterior and posterior commissures and points defining the midline plane, in each subject's anatomic image. The relative positions of these points are used to generate a nonlinear transformation of the subject's anatomic image to a standard brain coordinate system. Application of the same transformation to the corresponding statistical parametric maps for each subject permitted their overlay onto the anatomic reference and allowed the voxel-by-voxel combination of data from several subjects into a single composite or comparison activation map for each FMRI task.

Z score maps were calculated for each subject in the Talairach reference frame, under a general linear model (51), corresponding to the correlation between the functional response for each voxel and the block periodic time series for the task under study. Composite Z maps were generated for each task for both the childhood-onset SLE group and the control group, using a random effects analysis (52), to identify regions of significant group activation. The differences in activation patterns between the childhood-onset SLE and control groups were then assessed for each task separately and described statistically using difference Z maps. Each voxel of the difference map represents the Z score of the difference in mean activation between groups. Significance of differences in activation patterns for each parametric map was ascertained using a clustering method (53). A nominal Z threshold in the range of 1.96–2.10, combined with a cluster size threshold of 25, resulted in a corrected P value of <0.05 per reported cluster, as determined via Monte Carlo simulation.

In exploratory analysis, Pearson correlation coefficients between the activation in corresponding voxels of each childhood-onset SLE patient's Z map for each task and selected scores on the formal neuropsychological testing battery outlined above were calculated. This resulted in computed brain maps showing areas of significant correlation between findings on each FMRI task and those on each standardized neuropsychological test.

RESULTS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. REFERENCES

Activation patterns for the FMRI paradigms.

Figures 1–3 show composite general linear model Z score maps of activation patterns in the childhood-onset SLE and control groups, for each of the 3 FMRI tasks. These images demonstrate the average activation in all subjects performing the task. For each task, the composites were constructed using a general linear model with random effects, with significance threshold of P < 0.05 ascertained with a nominal Z value of 1.96–2.10 and a cluster size threshold of 25 voxels, as described above. Consistent thresholds across subjects and tasks enabled visual comparison of the associated activation patterns.

thumbnail image

Figure 1. Composite Z score activation for the verb generation task in the childhood-onset systemic lupus erythematosus (cSLE) and control groups. Areas of activation above the statistical threshold of P < 0.05 in a random effects general linear model are indicated in color overlaid on the mean anatomic reference image. Activation is indicated by red-yellow, while deactivation is indicated by blue-green. Statistical significance was ascertained at a nominal Z value of 1.96–2.10 and a cluster size threshold of 25 voxels. Images are in the radiologic convention; thus, the right side of the image corresponds to the left side of the brain and vice versa.

Download figure to PowerPoint

thumbnail image

Figure 2. Composite Z score activation for the continuous performance task–identical pairs task in the childhood-onset systemic lupus erythematosus (cSLE) and control groups. Areas of activation above the statistical threshold of P < 0.05 in a random effects general linear model are indicated in color overlaid on the mean anatomic reference image. Activation is indicated by red-yellow, while deactivation is indicated by blue-green. Statistical significance was ascertained at a nominal Z value of 1.96–2.10 and a cluster size threshold of 25 voxels. Images are in the radiologic convention; thus, the right side of the image corresponds to the left side of the brain and vice versa.

Download figure to PowerPoint

thumbnail image

Figure 3. Composite Z score activation for the working memory–N-Back task in the childhood-onset systemic lupus erythematosus (cSLE) and control groups. Areas of activation above the statistical threshold of P < 0.05 in a random effects general linear model are indicated in color overlaid on the mean anatomic reference image. Activation is indicated by red-yellow, while deactivation is indicated by blue-green. Statistical significance was ascertained at a nominal Z value of 1.96–2.10 and a cluster size threshold of 25 voxels. Images are in the radiologic convention; thus, the right side of the image corresponds to the left side of the brain and vice versa.

Download figure to PowerPoint

The verb generation task revealed stronger negative activation in the control group compared with the childhood-onset SLE group, as reflected in the areas highlighted in blue in Figure 1. This most likely arose from increased activation during the control intervals of bilateral finger tapping during application of this paradigm. In language areas, the childhood-onset SLE subjects demonstrated greater activation volume in the Broca area (yellow), while the controls exhibited a more balanced typical pattern of left hemisphere activity in the Broca and Wernicke areas. At the selected threshold of P < 0.05, the Wernicke area was not shown to be activated in the childhood-onset SLE group composite map. The control group also had more pronounced negative activation (blue) in the visual associative region as well as in the cingulate gyrus, particularly its anterior portion.

Differences between the childhood-onset SLE and control groups were evident in the comparison of CPT-IP maps shown in Figure 2. Positive activation in visual association areas of the fusiform gyrus in the occipital lobe appeared to be greater in the childhood-onset SLE population. In the control group, a large area of negative activation appeared in the posterior cingulate gyrus, with much greater intensity than in the childhood-onset SLE group. Negative activation may represent positive correlation of the BOLD signal with the CPT-IP control task. Also notable in the composite maps for this task was the appearance of a positive area of activation in the left dorsolateral prefrontal cortex in the control group. This region is reflected in the right hemisphere in the childhood-onset SLE group. Normally this region is associated with short-term working memory, and normal subjects performing a CPT task would likely demonstrate some level of activation in the right hemisphere. This contrast likely represents an important difference in neural circuitry of attention in presymptomatic childhood-onset SLE.

General linear model composite maps for the N-Back task (Figure 3) also suggested increased brain activity in the childhood-onset SLE group, in the visual association regions (fusiform gyrus bilaterally) as well as the somatosensory association regions (frontal–parietal junction). In addition, images from the control population exhibited strong negative activation in the frontal eye fields and the anterior portions of the dorsolateral prefrontal cortex, suggesting increased activation of these areas during the control task in the healthy control group as compared with the childhood-onset SLE group. Elements of the dorsolateral prefrontal cortex that lie more posterior, however, appeared to be positively activated to a similar level in the 2 groups.

Differential activation maps for the FMRI tasks.

Differential activation maps between the childhood-onset SLE and control groups (Figure 4) depicted statistically significant differences between the childhood-onset SLE group and the control group (P < 0.05) in the activation patterns for each of the FMRI tasks shown in Figures 1–3. Positive differences in activation levels between groups are indicated by red-yellow colors, while negative differences are highlighted in blue-green. Thus, voxels with greater positive activation in the childhood-onset SLE group, or more negative activation in the control group, are depicted as a positive difference on the map and are highlighted in red-yellow. Conversely, negative differences, which can result from the childhood-onset SLE group having either less positive activation or greater negative activation than controls, are shown in blue-green. These difference maps serve to quantify differences between the childhood-onset SLE and control groups described qualitatively by the visual comparisons seen in Figures 1–3.

thumbnail image

Figure 4. Significant differential activation between the childhood-onset systemic lupus erythematosus (SLE) and control groups on 3 functional magnetic resonance imaging tasks (verb generation [Verb Gen.], continuous performance task–identical pairs [CPT-IP], and working memory–N-Back). Suprathreshold regions (P < 0.05, nominal Z = 1.96–2.10, cluster threshold 25 voxels) are indicated in color overlaid on the mean anatomic reference image. Regions in which activation was greater in the childhood-onset SLE group than in the control group are shown in red-yellow; regions in which activation was greater in the control group than in the childhood-onset SLE group are shown in blue-green. Images are in the radiologic convention; thus, the right side of the image corresponds to the left side of the brain and vice versa.

Download figure to PowerPoint

The differential activation map of the verb generation task suggested positive differences in small regions of the visual and somatosensory association cortex as well as in the Wernicke area. For CPT-IP, the differential activation map showed significant positive differences in the visual association regions and the posterior cingulate gyrus, as suggested by visual inspection of the corresponding composite maps in Figure 2. Significant N-Back activation differences between groups were observed in the anterior dorsolateral prefrontal cortex and small portions of the visual association cortex. A more detailed summary of the Brodmann areas (54) in which significant differences in activation between the childhood-onset SLE and control groups were found is presented in Table 3.

Table 3. Brain regions with significant differences in activation between the childhood-onset SLE and control groups, for each FMRI task assessed*
Brain region/Brodmann area(s)Hemisphere(s) with significant difference
FMRI CPT-IP taskFMRI N-Back taskFMRI verb generation task
  • *

    SLE = systemic lupus erythematosus; FMRI = functional magnetic resonance imaging; CPT-IP = continuous performance task–identical pairs; L = left; R = right.

  • See ref.54.

Visual association region/18, 19L, RL, RL, R
Fusiform gyrus/37L, R
Posterior cingulate gyrus/23, 31L, R
Thalamus/–L, R
Motor cortex/4, 6LRL
Auditory cortex/41, 42L, RL
Dorsolateral prefrontal cortex/9, 46L, R
Inferior prefrontal gyrus/47R
Frontal eye fields/8L
Superior temporal gyrus/22RL
Frontopolar region/10L
Somatosensory association cortex/7L, R

Correlation analysis of FMRI findings and results of formal neuropsychological testing.

For the group of subjects with childhood-onset SLE, Pearson correlation coefficients were calculated between brain activation for each voxel corresponding to each FMRI task and the Z scores from the neuropsychological tests shown in Table 2. This resulted in the identification of focal brain regions for which activation was significantly correlated (P < 0.05) with the performance on specific neuropsychological tests. In general, these correlated brain regions included small cortical volumes in the dorsolateral prefrontal cortex and the posterior cingulate gyrus. Most commonly, correlations between neuropsychological test scores and FMRI activation were negative, suggesting that lower cognitive scores result in stronger task-related brain activation. Areas of significant correlation between FMRI data and neuropsychological test scores are shown in Table 4.

Table 4. Brain regions showing significant correlation between FMRI results and scores on neuropsychological tests, in the childhood-onset SLE patients*
Principal cognitive domain tested, neuropsychological testRegion for which activation during FMRI task correlated with neuropsychologic test result
FMRI CPT-IP taskFMRI N-Back taskFMRI verb generation task
  • *

    DLPC = dorsolateral prefrontal cortex (see Table 3 for other definitions).

Visuoconstructional processing, Rey-Osterreith Complex Figure Copy TrialL: posterior cingulate (negative correlation); R: DLPC (negative correlation)L, R: posterior cingulate (negative correlation); L: visual association (negative correlation)L: motor cortex (positive correlation)
Memory, Rey-Osterreith Complex Figure Immediate Recall TrialL, R: retrosplenial (negative correlation); L, R: posterior cingulate (negative correlation)L, R: anterior cingulate (negative correlation); L, R: motor cortex (negative correlation)L: insula (positive correlation)
Attention/executive functioning, Delis-Kaplan Executive Function System Color/Word Interference SubtestR: DLPC (negative correlation)R: DLPC (positive correlation)None
Psychomotor speed, Delis-Kaplan Executive Function System Color/Word Reading SubtestR: inferior prefrontal gyrus (negative correlation); R: supramarginal gyrus (negative correlation); L, R: sensory association (negative correlation)L, R: anterior cingulate (negative correlation); L, R: ventral posterior cingulate (negative correlation); L: DLPC (positive correlation)None

DISCUSSION

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. REFERENCES

This preliminary study introduces the application of FMRI techniques for exploring a range of neurocognitive domains in patients with childhood-onset SLE in comparison with healthy control groups. Functional imaging paradigms were chosen to exercise those domains of cognitive function that have been implicated, by formal testing, to be subject to early decline in NPSLE. Activation patterns for each of the FMRI paradigms comprised areas previously found to be activated in normal control populations in both groups included in this study. More importantly, significant and diffusely distributed differences in activation between patients with childhood-onset SLE and healthy controls were found, particularly for tasks probing working memory and attention. It is notable that significant differences in the childhood-onset SLE patients were detectable despite the small sample size and lack of clinically overt cognitive deficits.

Deficits in language ability are among the earliest markers of the presence of a developmental or acquired neurologic disorder. It has been suggested that even individuals with SLE without overt NPSLE, such as the patients in this study, have difficulties with verbal memory (55, 56). Difficulty with verbal memory (based on the results of formal neuropsychological testing) was present in some of the subjects with childhood-onset SLE in this study. Thus, tests of language processing appear to be particularly suited for studying neural substrates that are involved in NPSLE. Of particular note is that areas in which there were differences in activation patterns between childhood-onset SLE patients and controls during the verb generation task included the Wernicke area (Brodmann area 22), which is involved in language association. Whereas controls showed typical left hemisphere dominance of activation in both the Broca area (Brodmann area 44) and the Wernicke areas (32, 33), the childhood-onset SLE group did not exhibit significant activation in the Wernicke area (corrected P < 0.05). Conversely, the Broca area was covered more broadly in the group of childhood-onset SLE patients, with activation extending more inferiorly. Based on these FMRI findings, it appears that a deficit in the neural circuitry for word fluency emerges with childhood-onset SLE, suggesting that FMRI may be sensitive in detecting early influences of childhood-onset SLE on the neural substrates of semantic association and other language functions.

A variety of tests have been developed to measure attention (57, 58). The CPT-IP task used in this study was designed to measure attention by requiring focus on a continuous stream of data while responding to specific stimuli, in accordance with a set of a priori instructions (59–61). The group of childhood-onset SLE patients exhibited more extensive activation than control subjects in large tracts of the fusiform gyrus and visual associative cortex, regions associated, in previous research, with abnormal attention (62, 63). These results suggest that functional changes in these brain areas underlie attentional changes experienced in childhood-onset SLE.

Preclinical studies and biopsy findings suggest that the frontal cortex and the hippocampus mediate working memory (40, 64–66). Similarly, functional neuroimaging (SPECT, PET, FMRI) has consistently identified prefrontal and temporal brain activation during working memory performance (41, 42, 67–72). This study probed working memory via the N-Back paradigm adapted for FMRI and demonstrated that the same cortical structures were involved. Earlier investigations have revealed evidence of impairment of working memory in SLE (73). The childhood-onset SLE group in the present study exhibited stronger and more extensive activation in regions associated with working memory, suggesting that SLE has an effect in this domain. A possible explanation for this observation of increased activation in visual and attention areas during performance of this task by childhood-onset SLE patients is that they must exert greater effort to perform the task. There is evidence to suggest that increased effort will result in increased activation with the same task (42, 74, 75).

Differential activation mapping, irrespective of the FMRI task performed, tended to show more pronounced activation of the targeted cortical regions in childhood-onset SLE patients compared with controls. The cingulate gyrus, an inhibitory brain region (76), was consistently activated during control task periods in all 3 paradigms in this investigation. A possible interpretation of the decreased negative activation in the childhood-onset SLE patients would be that there is less suppression of brain activity during the control tasks of the paradigms. Extending this hypothesis, these observations could suggest that patients with childhood-onset SLE manifest an imbalance of cognitive effort, with more effort required to do the task at hand, while less is applied to inhibiting task action during control periods. Association of greater task activation with cognitive deficiency is further supported by the preponderance of significant negative correlations between task activation across paradigms and neuropsychological test scores.

The present FMRI findings reveal widespread differences and imbalances of brain activation in childhood-onset SLE patients compared with healthy controls. While the systemic nature of the disease may play a role in explaining these diffuse changes, it is also reasonable to consider damage or malfunction of connectivity of the underlying neural networks. Diffuse changes in brain network distribution are a plausible outcome of damage to critical connections between network elements, and their occurrence has been suggested previously (13). This hypothesis can be supported by previous observations of deep-seated white matter involvement in SLE (77).

This pilot study was subject to a number of limitations that should be considered in interpreting these initial results. First, minor discrepancies in the age and sex distribution and possibly socioeconomic status between the childhood-onset SLE and control groups included for each paradigm arose from the opportunistic selection of controls for this pilot study from other projects ongoing or completed in our laboratory. Although suboptimal in terms of matching controls to childhood-onset SLE subjects, without this efficiency it would not have been possible to complete the pilot study within the limited budget available. Furthermore, investigators at our institution have previously shown that sex differences in activation patterns associated with the verb generation task are very small and are not evident in studies with small sample sizes (78). We would expect this observation to hold true for the attention and working memory tasks included in this pilot study as well, and previously reported studies of these tasks and control subjects have not accounted for sex differences. Age distribution may have been a confounder in the findings on the attention task, since differences between younger and older adults performing this task have been reported previously (45).

Another possible limitation of the present study is the fact that the medications used for the treatment of SLE can affect cognition (79). Since nearly all of the patients with childhood-onset SLE in this study were treated with similar corticosteroid regimens (Table 1), the impact of medications on the results cannot be assessed conclusively. Followup work with untreated patients will be needed in order to explicitly address the role of corticosteroids in these FMRI findings. Although none of the patients with childhood-onset SLE in this study exhibited CNS or vascular involvement as assessed with standard activity and damage indices, a role of specific pathologies in explaining the observed activation differences in this limited work cannot be ruled out.

A final point worth noting is that SLE may act on the brain indirectly, through a pathologic influence on neurovascular physiology. Given that the BOLD effect used to create FMRI activation maps is based on subtle changes in blood flow to locally activated cortical areas, the etiopathogenesis of SLE, as well as possible effects of medications on vascular reactivity, could influence this effect. We attempted to control for this possible confounding factor by assessing the presence of vasculitis as defined in the SLEDAI (21, 22). None of the childhood-onset SLE patients were found to have evidence of vascular disease by this measure; however, it is still possible that there could have been an unspecified difference in neurovascular physiology between the childhood-onset SLE patients and the control subjects. Given that the differences found for each of the FMRI tasks were discrete, it seems unlikely that there is a global difference in vascular physiology between the 2 groups. However, this point may remain obscure until research reveals more about the underlying mechanisms of NPSLE.

Despite promising results in animal models, to date there are no laboratory markers or conventional imaging methods that allow for the reliable early diagnosis of cognitive dysfunction in SLE (80–82). At present, the degree of sensitivity of FMRI does not permit its use as a diagnostic tool for cognitive function in individual SLE patients. Nevertheless, based on the present findings, FMRI may be a promising approach to elucidate the areas and mechanisms involved in the development of cognitive dysfunction in SLE. Benefits from such understanding could include improved treatment strategies. Our detection of changes in brain function in childhood-onset SLE prior to overt manifestation of cognitive dysfunction encourages the establishment of longitudinal studies to measure changes in FMRI activation patterns during the early stages of disease. Use of this approach may identify categories of patients at risk for subsequent development of overt cognitive dysfunction. In future studies with larger sample sizes, connectivity analysis should be performed to further analyze the impact of childhood-onset SLE on brain network integrity.

AUTHOR CONTRIBUTIONS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. REFERENCES

Dr. Brunner 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. DiFrancesco, Brunner.

Acquisition of data. DiFrancesco, Holland, Ris, Adler, Nelson, DelBello, Brunner.

Analysis and interpretation of data. DiFrancesco, Holland, Ris, Adler, Brunner.

Manuscript preparation. DiFrancesco, Holland, Ris, Adler, Nelson, DelBello, Altaye, Brunner.

Statistical analysis. DiFrancesco, Adler, Altaye, Brunner.

REFERENCES

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. REFERENCES