Abstract
Objective. Studies that have examined abnormalities in cerebral blood flow (CBF) in patients with systemic lupus erythematosus (SLE) reported CBF relative to a region assumed to be normal in the brain. We examined the absolute differences in both regional CBF and cerebral blood volume (CBV) between patients with SLE and healthy controls.
Methods. CBF and CBV were measured with dynamic susceptibility contrast (DSC) magnetic resonance imaging (MRI), a technique that provides an alternative to radionuclide perfusion studies and permits quantitative anatomic, CBF, and CBV imaging in a single scanning session. CBF and CBV were measured in lesions and in normal-appearing tissue in the major cerebral and subcortical brain regions. Unlike most perfusion studies in SLE, CBF and CBV values were not normalized to a region of the brain assumed to be healthy.
Results. CBF and CBV within MRI-visible lesions were markedly reduced relative to surrounding normal-appearing white matter. CBF and CBV in normal-appearing tissue were both higher in SLE patient groups, with or without lesions, relative to the control group.
Conclusion. DSC MRI, without normalization to a region presumed to be healthy, revealed that CBF and CBV in normal-appearing tissue in patients with SLE was higher than CBF and CBV in controls. Since this finding was made in subgroups of patients with and without lesions, the higher CBF and CBV appear to precede lesion pathology.
Systemic lupus erythematosus (SLE) can produce a wide variety of neurologic and psychiatric syndromes, including headache, chorea, cognitive dysfunction, and psychosis1. Histopathologic studies have shown evidence of widespread parenchymal and cerebrovascular injury, including thromboembolic vasculopathy, infarction, gliosis, and diffuse neuronal and axonal loss with varying degrees of inflammation2. Evidence of brain injury and altered brain physiology in SLE also comes from studies on neurometabolism measured by 18F fluorodeoxyglucose (18FDG) positron emission tomography (PET) or magnetic resonance spectroscopy3,4,5,6,7,8, and radionuclide cerebral blood flow (CBF) studies3,8,9,10,11,12,13,14,15,16,17,18,19,20,21, the majority based on single photon emission computed tomography (SPECT). Diffuse patchy areas of regionally reduced perfusion have been reported in diffuse neuropsychiatric SLE (NPSLE) syndromes, while focal areas of regionally reduced perfusion may occur in focal NPSLE13,16,17,18. These abnormalities may be conspicuous in presumed areas of neurologic impairment and improve with corticosteroid therapy or clinical resolution15,19. However, at least 1 study examining SLE subjects with and without neuropyschiatric symptoms reported no relationship between NPSLE, neurocognitive function, and perfusion abnormalities, suggesting that reduced perfusion may not be a hallmark of NPSLE11.
In contrast to the many SPECT studies on SLE, far fewer studies have examined CBF using perfusion methods based on PET or magnetic resonance imaging (MRI). PET, while costly and exposing the patient to high-energy ionizing radiation, is considered one of the most accurate methods for CBF measurement22. A less costly alternative approach is dynamic susceptibility contrast (DSC) MRI23, which can be conveniently performed in the same examination as anatomic MRI, is inherently coregistered to the anatomical images, and most important, does not expose the patient or examiner to ionizing radiation or radionuclides. While 2 early studies using PET found evidence of transient reduced21 or lower CBF in a single subject20, a more recent PET study found higher CBF relative to control subjects in anemic, hypertensive patients with chronic renal failure, including 1 patient with SLE24. Only 1 study on SLE using DSC MRI has been reported25. In that study, both SPECT and DSC MRI were applied and revealed areas of relative hypoperfusion in SLE subjects, with SPECT detecting more hypoperfused areas than DSC MRI. However, CBF differences in these studies were evaluated qualitatively or normalized relative to CBF in a region of the brain presumed to be unaffected by disease. To our knowledge, absolute regional differences in CBF between SLE and control subjects have not been reported. Further, quantitative cerebral blood volume (CBV) abnormalities in SLE, not measureable by conventional SPECT, have yet to be reported.
The purpose of our study was to use DSC MRI to quantitatively measure both cerebral perfusion and blood volume in patients with SLE and healthy controls using non-normalized measures of CBF and CBV. Segmentation of the anatomical images acquired in the same scanning session as the DSC MRI images allowed calculation of mean gray and white matter CBV and CBF values in cerebral (occipital, parietal, temporal, and frontal) and subcortical (caudate, putamen, thalamus, and globus pallidus) regions and within brain lesions for both patients and controls.
MATERIALS AND METHODS
Subjects
Forty-two SLE subjects and 19 age- and sex-matched healthy controls were studied. The diagnosis of SLE was established in each subject using the American College of Rheumatology (ACR) 1997 revised criteria26. SLE disease activity was determined with the SLE Disease Activity Index (SLEDAI), SLE-associated injury was measured with the Systemic Lupus International Collaborating Clinics (SLICC)/ACRDI (damage index), and NPSLE activity was determined with the neurologic subsets of both neuro-SLEDAI and neuro-SLICC/ACRDI as reported27. Past and present neuropsychiatric symptoms and findings were classified using the 1999 ACR Case Definitions for NPSLE28. Important demographic features of the SLE group are shown in Table 1.
The study was approved by the University of New Mexico Institutional Review Board and conformed to the Declaration of Helsinki. All subjects provided informed consent and signed formal permission for all procedures.
MRI
Anatomical MRI, DSC MRI, and MR angiography were performed on all subjects using a 1.5-Tesla Siemens Sonata scanner with an 8-channel head coil. The total time of the MRI protocol was about 1 hour. Scout images were used to prescribe a series of whole-head T1-weighted, T2-weighted, and fluid attenuated inversion recovery (FLAIR) images aligned with the interhemispheric midline and parallel to the anterior commissure-posterior commissure (AC-PC) line. T1-weighted images [3D fast low angle shot (FLASH) sequence, TR/TE 12/4.76 ms, flip angle 20°, field of view (FOV) 256 × 256 mm, resolution 1 mm × 1 mm, 128 slices, slice thickness 1.5 mm] and T2-weighted images (turbo spin-echo sequence, TR/TE 9040/64 ms, turbo factor 5, FOV 220 × 220 mm, resolution 1.1 × 1.1 mm, 128 slices, slice thickness 1.5 mm) were acquired for anatomical segmentation of normal-appearing gray matter (GM), white matter (WM), and cerebrospinal fluid (CSF). The FLAIR image (variable flip, TR/TE/TI 6000/358/2100 ms, averages 2, slice thickness 1.5 mm, FOV 220 mm, matrix size 192 × 192) was used to manually segment lesions. The segmented GM, WM, CSF, and lesion maps were used to identify these regions in the processed perfusion maps. The DSC MRI images were acquired with a perfusion-weighted echo planar imaging (EPI) sequence (TR 1430 ms/TE 46 ms/flip 90°, 20 slices, time course of 50 sequential acquisitions, FOV 200 mm2, matrix 128 × 128). Gadopentetate dimeglumine contrast agent (Magnevist®; Bayer HealthCare Pharmaceuticals, www.bayerus.com) was injected into an antecubital vein at the standard dose (0.1 ml/kg body weight) using a power injector at 5 ml/s, starting 15 s after the start of scan acquisition, followed by 20 ml saline at the same rate. The central volume principle allows perfusion characteristics (blood flow and volume) to be estimated from the signal profile generated by the contrast bolus23. Raw DSC MRI data were analyzed using the Penguin software package (NordicImagingLab, Bergen, Norway). The software converts pixel intensity changes due to contrast (gadopentetate) passage over the time series of images into pixel by pixel tissue contrast concentration curves, C(t). With user supervision, it selects candidate arterial pixels to construct an arterial input function, AIF(t), based on the higher intensities and narrow time courses of the concentration changes. An attempt was made to locate and use pixels associated with the middle cerebral artery consistently for the construction of the AIF(t). A gamma variate function is fit to the raw C(t) and AIF(t) curves to identify just the first pass of the contrast bolus and is used to represent C(t) and AIF(t) in further analyses. The CBV in any pixel is calculated as the ratio of areas under the tissue and artery concentration curves K × ∫C(t) dt / ∫AIF(t) dt. The constant K was set to 1 rather than to a value based on assumed capillary and artery hematocrits and tissue density29. The calculation of CBF is more complex, and involves an equation relating C(t) to CBF and the convolution of AIF(t) and the residue function R(t), the fraction of contrast agent in the vasculature at any time: C(t) = K × CBF × [AIF(t) ⊗ R(t)], where ⊗ denotes the convolution operator. Various strategies have been proposed for deconvolving AIF(t) from R(t) in order to solve this equation for CBF23. We applied the circular singular value decomposition method30 available as an option in Penguin that has been shown to provide robust CBF measures in the presence of variable delays between bolus arrival in the arterial pixels defining the AIF(t) and the tissue pixels defining C(t). The CBV and CBF maps constructed with Penguin were coregistered to the T1-weighted images to identify distinct brain regions within gray and white matter. This procedure involved automatically segmenting the T1-weighted image into GM, WM, and CSF and coregistering the lesion map that was manually traced from the FLAIR image to the T1-weighted image. The T1-weighted image was warped into a standard brain template for identifying the major cortical and subcortical brain regions. Finally, the resulting regionally segmented map was transformed back into the native space of the CBV and CBF maps. These steps were accomplished with the BRAINS2 software31. GM and WM values for 4 cerebral regions (occipital, parietal, temporal, and frontal) and 4 subcortical regions (caudate, putamen, thalamus, and globus pallidus) for each hemisphere are reported. These regions of interest are shown in a representative T1-weighted image from one subject in Figure 1. Mean CBV and CBF values in lesions are also reported. In order to resolve CBV and CBF in normal-appearing tissue from CBV and CBF in lesions, the mean regional cerebral or subcortical values of CBV and CBF reported were calculated without the lesion pixels included. This allowed us to observe similarities in normal-appearing tissue between lupus groups (those with and those without lesions) without the confound of the lower lesion CBV and CBF values reducing the overall mean values for the region.
Statistical analysis
Group differences in regional CBV and CBF among SLE subgroups (with and without lesions) and the controls were analyzed by repeated measures (RM) ANOVA with the 3 groups as the grouping factor and the different brain regions as a repeated factor. This approach was motivated by the fact that CBF and CBV are known to vary from region to region in the brain and hence, statistical comparisons based on CBF or CBV averaged over the whole brain will fail to identify differences in the regional variation of these measurements. RM ANOVA was performed in each of 3 general tissue types: cerebral gray matter, cerebral white matter, and subcortical gray matter. Posthoc pairwise comparisons were made using Wald’s t-test. The RM model in each general region was constructed with brain hemisphere and specific region (e.g., occipital, parietal, temporal, and frontal) as repeated factors (e.g., see Table 2). Correlations between CBF and antiphospholipid antibody (aPL) tests were examined with linear regression and Spearman’s rank correlation coefficient. All statistical analyses were performed with SAS® (http://www.sas.com).
RESULTS
Forty-two SLE subjects and 19 controls were studied. The SLE and control groups were similar in age (SLE 38.1 ± 12.4 yrs, controls 34.8 ± 11.9 yrs; p = 0.33), sex (SLE 95% women, controls 89%; p = 0.58), weight (SLE 71.4 ± 14.0 kg, controls 69.7 ± 11.9 kg; p = 0.63), and body mass index (SLE 26.65 ± 4.7 kg, controls 25.8 ± 5.0 kg; p = 0.56). Twenty-three (54.8%) SLE subjects had focal lesions that were detectable on FLAIR images. As expected, the total lesion volume in the group of patients with SLE varied, with a median of 0.70 cm3 and interquartile range 0.21 to 5.99 cm3. However, regression analysis revealed no significant correlation between lesion volume and CBF or CBV in non-lesion tissue in these subjects. Among patients without hypertension (systolic blood pressure < 140 mm Hg, diastolic < 90 mm Hg), 48% (16/33) had lesions and among those with hypertension, 88% (7/8) had lesions. Using Fisher’s exact test, the correlation between the presence of lesion and hypertension was found not to be significant. Subjects with lesions tended to be older (p = 0.051) and to have lower levels of anti-dsDNA antibodies (p = 0.02) than those without lesions. However, subjects with and without lesions were very similar in all other clinical and laboratory measures, including anticardiolipin antibodies and lupus-like inhibitor, SLE disease activity (SLEDAI), SLE damage (SLICC-ACRDI), NPSLE activity (Neuro-SLEDAI), and damage caused by NPSLE (Neuro-SLICC; Table 1). The lesions were located almost exclusively in white matter regions and predominantly in the frontal and parietal lobes. Anatomical and perfusion images from one subject are shown in Figure 1, demonstrating a pronounced reduction in perfusion at the site of a lesion evident in the FLAIR image. A significant group difference in aPL levels between SLE groups with and without lesions was not observed in this sample.
A regional pattern of posterior to anterior increase in CBF was observed in all groups (Figure 2), in agreement with perfusion studies using other methods32,33 and reflected by a similar pattern of CBV across the brain (Figure 2). Thus, occipital and parietal lobes were consistently less perfused than the frontal lobes and, within subcortical regions, the globus pallidus was less perfused than the caudate nucleus in all subject groups. Also, as expected, both CBF and CBV in MRI-visible lesions were significantly reduced (p < 0.001) relative to normal-appearing white matter (Tables 2 and 3), indicating a disruption of vascular-parenchymal integrity in MR-visible lesions. Finally, both CBF and CBV were higher in the SLE groups (with or without lesions) relative to the control group (Tables 2 and 3). RM ANOVA and posthoc pairwise t-tests, used to examine group differences across either all cerebral gray matter, all cerebral white matter, or all subcortical gray matter regions, revealed that both CBV and CBF were significantly higher in SLE subjects than in controls in all regions except possibly in subcortical gray matter for CBF, where the p value was 0.051. Neither prednisone use nor multiple measures of disease activity (i.e., SLEDAI, Neuro-SLEDAI, SLICC, Neuro-SLICC) correlated significantly with CBF or CBV. A trend to higher CBV and CBF in SLE subjects without MRI-detectable lesions relative to those with lesions was also observed, indicating that the underlying cause of elevated blood volume and perfusion in SLE subjects may be independent of MR-visible lesions and may precede MRI-detectable lesion formation.
DISCUSSION
Studies have found evidence of abnormal cerebral perfusion in patients with SLE3,8,9,10,11,12,13,14,15,16,17,18,19,20,21. The majority of these were based on SPECT methods and reported regional deficits in CBF relative to brain regions presumed to be normal within the individual subject. Using quantitative DSC MRI without normalization, we found that CBF was significantly reduced within MRI-visible brain lesions in patients with SLE, indicating a disruption of vascular-parenchymal integrity consistent with prior SPECT studies. However, our results indicate that both CBV and CBF were consistently higher in normal-appearing tissue in patients with SLE than in corresponding tissues in age- and sex-matched healthy controls (Figure 3, Tables 2 and 3). Additionally, no significant group differences in CBV or CBF between SLE subjects with lesions and those without lesions were observed, indicating that the causal factor or factors contributing to elevated cerebral perfusion were not directly related to and may not be caused by lesion load.
The CBF group differences demonstrating increased CBF in SLE that we observed are not necessarily in disagreement with previous findings of relative regional reductions in CBF in SLE patients, including at least 1 study comparing SPECT findings to DSC MRI findings25. The group differences in CBF in normal-appearing tissue observed in our study, based on non-normalized measures of CBV and CBF, would have been obscured by normalization to any particular region, since CBV and CBF in all brain regions in the SLE group were found to be elevated relative to those in the control group. As well, the precise coregistration of perfusion and lesion maps inherent in integrated anatomic-perfusion MRI allowed us to exclude lesions from the analysis of normal-appearing tissue regions. Hence, the use of analysis methods in prior studies that normalized CBF by values in other brain regions and did not exclude obvious anatomic abnormalities could account for the absence of findings similar to those reported here. Finally, to our knowledge, this is the first report of direct comparisons of quantitative CBV between SLE and control groups. While issues concerning the accuracy of either MRI and SPECT-based perfusion measurements relative to PET are well known22,34, the measurement of CBV by DSC MRI is more straightforward than the calculation of CBF, and derived simply from the ratio of the time courses of the arterial and tissue bolus concentration curves. The CBV group differences we observed closely paralleled the CBF group differences.
Cerebral hyperperfusion in SLE has been reported previously, but only in the basal ganglia of patients with SLE who have Parkinsonian syndrome, using relative measures of perfusion with SPECT35,36. However, in a study that included at least 1 patient with SLE, quantitative PET was used to observe higher CBF throughout the brain of patients with renal disease relative to healthy controls24. Hyperperfusion has also been reported in other neurological disorders, including stroke37, occlusive sickle-cell disease38, transient ischemic attack39, seizures and epilepsy40, traumatic brain injury41, metabolic brain disease42, and migraine43. Increased CBV can also be observed in the symptom-free interval of a transient ischemic attack, the acute state of a completed infarction, traumatic brain injury, nonenhancing lesions of multiple sclerosis, and normal-appearing tissue in multiple sclerosis44. The increased CBV reported in stroke may indicate intact vascular beds with potentially salvageable areas in the penumbra of the lesion45. Further, increases in CBV can be a sign of vasomotor instability, a compensatory mechanism indicating resolving injury, or a reduction of compensatory capacity in response to further ischemia or injury46.
MRI is presently the preferred anatomic neuroimaging modality to evaluate NPSLE47. While H215O PET remains the “gold standard” for cerebral perfusion studies, it is costly, involves the use of a catheter to measure the arterial H215O concentration, involves high-energy ionizing radiation, and requires immediate proximity to a cyclotron that can generate the short-lived 15O radioisotope. SPECT offers a practical alternative to PET, but may have limitations in accuracy relative to PET, and therefore is most commonly used to assess relative CBF differences rather than absolute quantitative measures of CBF34. While studies have demonstrated that SPECT and DSC MRI are comparable to each other in accuracy48, DSC MRI has the advantages of not using radioisotopes, having higher spatial resolution, and allowing the measurement of CBV as well as CBF. The risks with the use of gadolinium-based contrast agents in DSC MRI are small and identical to those associated with the use of these agents for routine contrast MRI scans, but include an association with nephrogenic systemic fibrosis, possibly due to free gadolinium toxicity, in subjects with endstage renal disease49. Therefore, caution should be exercised in administering gadolinium-based contrast agents to subjects with highly compromised renal function.
Compared to non-MRI methods, DSC MRI allows easy and accurate registration of the CBF and CBV maps to MRI anatomical images acquired in the same scanning session. This latter advantage, for example, permitted accurate region-of-interest analysis of the CBF and CBV maps of this study in which mean CBF and CBV values were specifically assigned to gray matter, white matter, lesions, and the major anatomical regions after coregistration to high resolution T1-weighted and FLAIR images. This allowed us to observe not only the expected pattern of regional CBF and CBV differences within the brain, but also lower CBF and CBV in regions of small lesions. By not normalizing CBF and CBV values to values in a region presumed to be “normal,” as has been done in most studies on SLE, we were able to observe differences between control and SLE groups. However, it is important to note that uncertainty in the arterial input function measurement (see Methods) has been shown to limit the accuracy of CBF and CBV values obtained with DSC MRI50, hence the values reported here cannot be considered quantitative in the absolute sense. However, under the assumption that there was no systematic bias in the measurement of the AIF in either group, these values allow relative differences in both CBF and CBV between groups to be observed. Our findings, therefore, suggest that a similar approach of examining non-normalized CBF values, irrespective of the measurement method, be taken in future perfusion studies in SLE.
Our results demonstrate lower CBF and CBV in lesions but higher CBF and CBV throughout cortical and white matter regions of the brain in patients with SLE relative to controls. The higher global CBF and CBV values do not appear to be directly related to the presence of MRI-detectable lesions and may be due to other reactive physiologic or pathogenic factors underlying SLE, such as vasomotor instability, compensatory mechanisms for resolving injury, low-grade excitotoxicity due to antibody impairment of N-methyl D-aspartate (NMDA) receptors, or inflammatory factors51. Integrated DSC MRI and MRI is a robust method to measure CBF and CBV in specific anatomically defined tissues and may provide new insights into the neuropathology of SLE.
Footnotes
-
Supported by grants RO1 HLO77422-01-A3 and M01 RR00997 from the National Institutes of Health.
- Accepted for publication April 1, 2010.