| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
(Stroke. 1998;29:2254-2260.)
© 1998 American Heart Association, Inc.
Original Contributions |
From the Clinical and Magnetic Resonance Research System (A.S., W.L.S., W.M.B.) and the Departments of Internal Medicine (W.L.S., J.D.), Biostatistics Section of Family and Community Medicine (C.A.S.), and Neurosciences (W.M.B.), The University of New Mexico Health Sciences Center, Albuquerque, NM.
| Abstract |
|---|
|
|
|---|
MethodsForty-nine SLE patients (12 SLE-aPLS) and 23 control subjects were studied using magnetic resonance imaging and spectroscopy. N-Acetylaspartate/creatine (NAA/Cre) and choline/Cre (Cho/Cre) were measured in normal-appearing tissue. IgG and IgM antiphospholipid antibodies (aPL) were measured by enzyme-linked immunosorbent assay.
ResultsStroke, epilepsy, and elevated IgG-aPL were more common in SLE-aPLS patients than in SLE patients (P<0.001). NAA/Cre was lower (P<0.05) and Cho/Cre higher (P<0.001) in SLE-aPLS patients than in SLE patients without aPLS. Regression models showed NAA/Cre was most related to injury seen by imaging (P<0.01), disease duration (P<0.05), and prior neuropsychiatric SLE (NPSLE) (P=0.07). Reduced NAA/Cre was more closely related to IgG-aPL (P<0.01) than the presence of stroke or aPLS. When adjusted for all factors, Cho/Cre was most associated with the presence of aPLS (P=0.05).
ConclusionsSLE and SLE-aPLS are actually a clinical continuum describing brain injury in SLE, with SLE-aPLS being characterized by increased aPL, NPSLE, stroke, epilepsy, and disturbed neurochemistry. An elevated IgG-aPL level is a potent risk factor for brain injury as measured by NAA/Cre in SLE that is independent of stroke and aPLS. However, thrombotic phenomena and the presence of aPL (aPLS) are most closely associated with increased Cho/Cre in SLE. These results suggest that aPLs exacerbate SLE, resulting in increased thrombotic and nonthrombotic brain injuries. Spectroscopy detects brain injury in SLE and may permit better understanding of the neurological consequences of SLE and SLE-aPLS.
Key Words: antiphospholipid syndrome brain injuries lupus magnetic resonance neurochemistry spectroscopy
| Introduction |
|---|
|
|
|---|
| Subjects and Methods |
|---|
|
|
|---|
SLE-aPLS was defined as the presence of both SLE and aPL complicated by livedo reticularis, multiple miscarriages, deep venous thrombosis, arterial thromboembolism, pulmonary emboli, stroke, transverse myelitis, or multi-infarct dementia. Patients who had suffered strokes underwent diagnostic testing that included MR angiogram, hypercoagulability evaluation, echocardiogram, and noninvasive carotid studies to rule out large extracranial vessel, valvular lesion, intramural clot, medication, or a hypercoagulability syndrome other than SLE-aPLS as causing the stroke.
Immunologic Testing
All SLE patients were then tested for IgM and IgG-aPL, using a
standardized antigen in an enzyme-linked immunosorbent assay
(ELISA).17 18 Immulon I microtiter ELISA plates
(Dynatech) were coated with 30 µL (45 µg/mL) of a standardized
phospholipid antigen (Louisville aPL Diagnostics, Inc). The plates were
blocked for nonselective binding with a 1% gelatin solution. Serum
samples were diluted 1:100 in a 10% fetal bovine serum; 200 µL of
each sample was placed in duplicate wells and incubated for 2 hours.
The plates were then washed with a phosphate-buffered saline solution
with 0.05% Tween and 10% fetal bovine serum and developed with a
1:1000 dilution of goat anti-human IgG or IgM conjugated to horseradish
peroxidase and 2,2'-azino-di-(3-ethylbenzthiazoline) sulfonate. The
optical density at 405 nm was read on a micro-ELISA reader (Dynatech).
Previously established negative and positive controls were used as
standards to correct for plate-to-plate variability. Control samples
from 543 normal blood donors were used to standardize the assay. IgM-
and IgG-aPL were reported in MPL and GPL (standardized aPL units
based on reactivity to the standard antigen), respectively. Based on
this large control population, normal ranges (±2 SD) for IgG are 5 to
20 GPL and 0 to 10 MPL for IgM. aPL testing was performed once on each
individual according to the study design. Twelve patients with SLE-aPLS
(11 women) and 37 SLE patients (32 women) without aPLS were
identified.
Magnetic Resonance Examination
MR data were acquired with a 1.5-T clinical scanner (GE Medical
Systems). Critically ill or uncooperative SLE patients (n=35, 7 with
aPLS) and 10 control subjects were studied using single-column
short-echo spectroscopic imaging (SI) (TE=19 ms, TR=2000 ms; procedure
time, 24 minutes), which produced 8 (10 mm)3
voxels in deep occipitoparietal white matter
(WM).8 SLE patients who were medically stable and
cooperative (n=14, 5 aPLS) and control subjects (n=13) were studied
using multislice SI (TE=270 ms, TR=2300 ms; procedure time, 60
minutes), which produced a 32x32 spectroscopic grid of voxels across
the field of view.9 19 20 21 Sagittal T1-weighted
images (TE=16 ms, TR=600 ms) were used to select the location of the
spectroscopic data. Three slice locations aligned parallel to the
anterior-posterior commissure were chosen for the long TE acquisitions.
Oblique-axial T2-weighted MR images (TE=30/100 ms, TR=2800 ms; field of
view=200 mm, 15-mm slice thickness, 2.5-mm gap) coinciding with
the locations of the spectroscopic images were obtained.
Data Analysis
Images were scored (0=normal, 1=mild, 2=moderate, 3=severe) for
MRI abnormalities common to SLE (eg, cortical atrophy,
ventricular dilation, diffuse WM abnormalities,
periventricular WM abnormalities, infarct, small focal WM
lesions) as described previously.22 An injury
index was defined as the sum of the individual abnormality scores.
Infarcts were defined as focal irreversible high intensity lesions at
least 1 mL in volume (10 mm)3 on T2-weighted
images. Lesions smaller than 1 mL were categorized as small focal
lesions.
SI data sets were processed using cosine filtering in k-space, exponential apodization (3 Hz), zero filling to 1024 time-domain points, and Fourier transformation. Residual water signals were removed by high-pass time-domain convolution filtering. Initially, spectra were selected from normal-appearing occipitoparietal WM in all patients, avoiding voxels that were hyperintense on T2-weighted images. In SLE patients with stroke, further data were obtained from infarcts defined as focal regions of high intensity at least 1 mL in volume on T2-weighted images. A total of 21 lesions from 8 individuals with gross cerebral infarct were studied. Lesions that did not fill a spectroscopic voxel completely were not considered strokes and were not analyzed. Uninvolved areas in the contralateral hemisphere of the individuals with stroke were used to compare infarcts with normal-appearing tissues. Spectra were integrated to determine the area for NAA (1.9 to 2.1 ppm), creatine (Cre; 2.9 to 3.1 ppm), and Cho (3.1 to 3.3 ppm) and the ratios, NAA/Cre and Cho/Cre, were calculated. In normal-appearing tissues, the ratios from 5 adjacent voxels in each anatomic region were averaged to obtain values for each metabolite in each individual.
Data were analyzed individually for the TE=19 ms and TE=270 ms cohorts. The statistical observations among normal control subjects, SLE patients, and patients with SLE-aPLS in the 2 data sets were similar, although each cohort was composed of different individuals. To determine whether the trends in the independent cohorts represented true differences between patient subgroups, the 2 data sets were combined to provide greater statistical power, especially to determine unique characteristics of the SLE-aPLS subgroup (a total of 12 patients). Data from the long TE acquisitions were normalized to data at TE=19 ms using correction factors obtained from control data. Correction factors for individual metabolites were derived by dividing the mean metabolite ratio (ie, NAA/Cre or Cho/Cre) acquired at short TE by the mean value acquired at long TE. Thus, the corrections were made as follows: NAA/Cre(19 ms) =0.97xNAA/Cre(270 ms); Cho/Cre(19 ms)=0.70xCho/Cre(270 ms). These pooled data provided values from 23 control subjects and 49 SLE patients normalized to TE=19 ms.
Previous evaluation of SI reproducibility has shown the mean coefficient of variation to be 3.2% for NAA/Cre and 6.6% for Cre/Cho. The mean coefficient of variation for analysis reproducibility for NAA/Cre was 3.5% and for Cre/Cho was 4.4%.9
Statistical Evaluation
Summary statistics were obtained for all variables. Plots of
continuous variables were examined for distributional shape and for
outliers. Comparisons of continuous variables from 2 populations
were made using the 2-sample t test and the
nonparametric analogue, the Wilcoxon rank sum test.
Because results from both tests were similar, the Wilcoxon test
results are presented. When 3 populations were compared, ANOVA
was used with Fisher's least significant difference test to assess
differences between individual groups. Comparisons of categorical data
from 2 or more populations were made using Fisher's exact test.
Linear regression models were used to explore the relationships of the
predictor variables (aPLS, stroke, IgG-aPL, and IgM-aPL) to the
outcome variables (NAA/Cre and Cho/Cre). aPLS was coded so that the
coefficient for aPLS gives the change in mean NAA/Cre or Cho/Cre for
the group with aPLS relative to the group without aPLS. Similarly, the
coefficient for stroke gives the change in mean NAA/Cre or Cho/Cre for
the group with stroke relative to the group without. Because of
associations among the main predictor variables (aPLS, stroke,
IgG-aPL, and IgM-aPL), these variables were assessed individually
and with the other variables in the model. The association between
aPLS and the outcome variables NAA/Cre and Cho/Cho was the primary
focus of this study, so most models presented here include
aPLS. However, to determine the effect of aPLS that is independent of
stroke, IgG-aPL, and IgM-aPL, further models with subsets or all of
these predictor variables are presented. Finally, to adjust
for potential confounders (such as age, duration, SLEDAI, a history of
prior NPSLE episodes, and injury index) were added to the predictor
variables. SLEDAI was categorized as high (
10) versus low (<10).
Because of the association between injury index and IgG-aPL, the
independent effect of IgG-aPL was assessed by comparing the model with
all of the predictor variables to the model with all of the
predictor variables except injury index. To assess possible
nonlinear effects and to reduce the effects of outliers, the continuous
predictor variables IgG-aPL and IgM-aPL were also categorized.
Because the results were similar to those from models with continuous
variables, which are more informative, results for the models with
categorical variables are not presented.
Two subjects had high Cho/Cre values. Thus, Cho/Cre was categorized as
high (
0.9) and low (<0.9). Logistic regression models were developed
to determine variables that predict high Cho/Cre. The same
variables that were important in the linear regression
analysis were the important predictors in the logistic
regression models. However, because of the influence of outliers in the
estimates of the coefficients for the linear regression models, the
results for the logistic regression models are presented in
terms of odds ratios.
Although statistical significance was ascertained using P=0.05, some results that have probability values between 0.05 and 0.10 are discussed, because of an indication of an effect. Analyses were conducted using SAS software (SAS Institute).
| Results |
|---|
|
|
|---|
|
|
|
Table 3
shows the mean values for the
19-ms and 270-ms TE cohorts. Both data sets show decreased NAA/Cre in
SLE and SLE-aPLS patients and increased Cho/Cre in SLE-aPLS patients.
To confirm the trends noted in the individual data sets, the data were
combined as described above (Table 3
). NAA/Cre was significantly
decreased in SLE and SLE-aPLS patients relative to normal control
subjects (P<0.001; Figure 2
).
Moreover, in SLE-aPLS patients a significant (P<0.05)
reduction in NAA/Cre was demonstrated relative to SLE without aPLS.
|
|
To address potential correlations among the factors discussed above, we
used linear regression models to control for the effects of previous
NPSLE, injury index, age, IgG-aPL, IgM-aPL, SLE duration, and SLE
activity. For each metabolite ratio, a series of models was examined,
beginning with a univariate model of only aPLS and ending
with a model that adjusted for all of the potential predictor
variables. The univariate model for NAA/Cre indicates
that aPLS is associated with decreased NAA/Cre (Table 4
, model 1). However, the effect of aPLS
is not independent of IgG-aPL (model 3), a variable that is highly
associated with NAA/Cre regardless of whether aPLS is included (models
2 and 3). After further adjustment for stroke and IgM-aPL, IgG-aPL is
still an important predictor of NAA/Cre (model 4). Similarly, after
adjustment for previous NPSLE, age, SLE duration, and SLE activity, but
not injury index (because of its high correlation with IgG-aPL),
IgG-aPL is associated with NAA/Cre (model 5). Finally, to consider the
effect of IgG-aPL adjusted for MRI-visible abnormalities, injury index
was added, resulting in an erosion of the significance of IgG-aPL
(model 6). Thus, for NAA/Cre, the important predictors are injury index
or IgG-aPL, disease duration, and, possibly, prior NPSLE episodes.
|
Similar linear regression models were examined to assess the
association between aPLS and Cho/Cre. However, the estimates of the
linear regression coefficients differed significantly if 2 extreme
values were excluded. Thus, Cho was categorized into high Cho (
0.9)
and low Cho (<0.9), and logistic regression models were developed. We
present only the results from the logistic regression modeling,
although the same predictor variables were important in both
models. In the univariate models, aPLS and IgG-aPL are
associated with increased Cho/Cre (Table 5
, models 1 and 2). The effect of aPLS
remains significant when adjusted for either IgG-aPL or stroke, but the
estimate of the effect increases (models 3 and 4). Even after
adjustment for the potential confounders (age, duration, SLEDAI, prior
NPSLE, and injury index), the effect of aPLS remains significant
(models 5 and 6). One limitation of estimating the stroke and aPLS
effects is the high correlation between these 2 variables: most
aPLS patients had stroke (n=8), whereas only 4 aPLS patients did not,
and no patients had stroke without aPLS. Thus, it is difficult to
separate the aPLS effect from the stroke effect.
|
Paired comparison of lesions with comparable normal-appearing tissue in the same SLE-aPLS patients revealed an even greater reduction of NAA/Cre (P<0.002). Cho/Cre was similar in lesions and normal-appearing tissues in SLE-aPLS patients (P>0.6), but was elevated compared with normal control subjects and SLE patients without aPLS (P<0.001).
| Discussion |
|---|
|
|
|---|
The most prominent resonance in 1H-MRS of adult
brain is the neuronal marker NAA.29 A reduced
presence of NAA suggests neuronal injury or death and has been
associated with cognitive impairment, indicating an important
functional consequence of NAA depletion.9 30 31
The current study demonstrates abnormal brain metabolite ratios in SLE
and SLE-aPLS patients. The markedly reduced NAA/Cre of large focal
lesions in patients with SLE-aPLS is characteristic of infarct, whereas
reductions in normal-appearing tissue may indicate extensive
microlesions.5 6 8 28 These findings are
consistent with previous reports of disturbed neurometabolites
in NPSLE,7 8 9 10 although the metabolic
abnormalities observed here are more severe in patients with SLE-aPLS,
indicating a different or more extensive injury to brain. The presence
of aPLS alone is associated with reduced NAA/Cre (Table 3
). However,
when modeling included other clinical variables, including IgG-aPL
and injury index, the change in NAA/Cre in patients with SLE-aPLS
relative to those with SLE was not significant, indicating that the
majority of the observed change in NAA/Cre was associated with
MRI-visible brain injury, IgG-aPL, or, possibly, disease duration and
prior NPSLE (Table 4
).
Elevated Cho/Cre was observed in focal lesions and normal-appearing
tissues of patients with SLE-aPLS consistent with infarct, the
activation of cellular membranes, catabolism of myelin, or
inflammation.30 32 33 Cho/Cre was increased in
normal-appearing tissues even when other clinical factors were
included, suggesting exaggerated injury to normal-appearing tissue in
patients with SLE-aPLS consistent with widespread
microinfarction.5 6 28 34 The effect of aPLS
remained significant even after adjusting for all clinical factors
(Table 5
). The increase in Cho/Cre was associated with the presence of
aPLS but not stroke, IgG-aPL, or other clinical factors. However,
decreased NAA/Cre was not associated with aPLS after adjustment for
IgG-aPL, indicating a different or more complicated relationship with
IgG-aPL.35
The histological changes of NPSLE and aPLS can be similar. Bland vasculopathy with or without microthrombosis is common in NPSLE and aPLS, but inflammation is rarely seen in aPLS.28 36 Perivascular cuffing with inflammatory cells, microinfarcts, cortical atrophy, gross infarcts, hemorrhage, ischemic demyelination, and leukostasis has been observed in SLE.28 37 38 39 40 41 42 Thus, although NPSLE and primary aPLS have certain histological similarities, noninflammatory vasculopathy and thrombosis predominate in aPLS, whereas NPSLE has a more complex and variable pattern.
SLE and SLE-aPLS demonstrate similar patterns of neuronal injury by MRS. Reduced NAA/Cre and elevated Cho/Cre ratios characterize both disorders, consistent with neuronal injury, ischemic demyelination, and postischemic inflammation. These data also demonstrate that IgG-aPL may have a potent independent effect on brain injury, even after correcting for stroke and aPLS. Thus, the disorders may represent a continuum of SLE and IgG-aPLmediated disease. Future studies are required to determine whether the presence of IgG-aPL in a SLE patient should prompt therapy and whether the observed metabolite abnormalities in normal-appearing tissues are due to microscopic ischemic injury or to cytotoxic extension from adjacent gross infarct. Therapeutic options for the treatment of SLE-aPLS remain controversial, and efficacy for any intervention is extremely difficult to monitor even with widely accepted assays.4 40 41 43 Combined MRI/S may provide the means to detect brain injury and monitor therapy.
| Acknowledgments |
|---|
| Footnotes |
|---|
Received April 16, 1998; revision received July 30, 1998; accepted July 30, 1998.
| References |
|---|
|
|
|---|
This article has been cited by other articles:
![]() |
S. Appenzeller, L.M. Li, L.T.L. Costallat, and F. Cendes Neurometabolic changes in normal white matter may predict appearance of hyperintense lesions in systemic lupus erythematosus Lupus, December 1, 2007; 16(12): 963 - 971. [Abstract] [PDF] |
||||
![]() |
G Castellino, M Govoni, M Padovan, P Colamussi, M Borrelli, and F Trotta Proton magnetic resonance spectroscopy may predict future brain lesions in SLE patients: a functional multi-imaging approach and follow up Ann Rheum Dis, July 1, 2005; 64(7): 1022 - 1027. [Abstract] [Full Text] [PDF] |
||||
![]() |
M Govoni, G Castellino, M Padovan, M Borrelli, and F Trotta Recent advances and future perspective in neuroimaging in neuropsychiatric systemic lupus erythematosus Lupus, March 1, 2004; 13(3): 149 - 158. [Abstract] [PDF] |
||||
![]() |
P L Peterson, F A Howe, C A Clark, and J S Axford Quantitative magnetic resonance imaging in neuropsychiatric systemic lupus erythematosus Lupus, December 1, 2003; 12(12): 897 - 902. [Abstract] [PDF] |
||||
![]() |
A Katzav, J Chapman, and Y Shoenfeld CNS dysfunction in the antiphospholipid syndrome Lupus, December 1, 2003; 12(12): 903 - 907. [Abstract] [PDF] |
||||
![]() |
R L Brey, J Chapman, S R Levine, G Ruiz-Irastorza, R H. Derksen, M Khamashta, and Y Shoenfeld Stroke and the antiphospholipid syndrome: consensus meeting Taormina 2002 Lupus, July 1, 2003; 12(7): 508 - 513. [Abstract] [PDF] |
||||
![]() |
G. Sanna, M. L. Bertolaccini, M. J. Cuadrado, M. A. Khamashta, and G. R. V. Hughes Central nervous system involvement in the antiphospholipid (Hughes) syndrome Rheumatology, February 1, 2003; 42(2): 200 - 213. [Abstract] [Full Text] [PDF] |
||||
![]() |
J. CHAPMAN and Y. SHOENFELD Neurological and Neuroendocrine-Cytokine Inter-relationship in the Antiphospholipid Syndrome Ann. N.Y. Acad. Sci., June 1, 2002; 966(1): 415 - 424. [Abstract] [Full Text] [PDF] |
||||
![]() |
J S Axford, F A Howe, C Heron, and J R Griffiths Sensitivity of quantitative 1H magnetic resonance spectroscopy of the brain in detecting early neuronal damage in systemic lupus erythematosus Ann Rheum Dis, February 1, 2001; 60(2): 106 - 111. [Abstract] [Full Text] [PDF] |
||||
![]() |
J PALACE and B LANG Epilepsy: an autoimmune disease? J. Neurol. Neurosurg. Psychiatry, December 1, 2000; 69(6): 711 - 714. [Full Text] [PDF] |
||||
![]() |
J. A. Aarli Epilepsy and the Immune System Arch Neurol, December 1, 2000; 57(12): 1689 - 1692. [Abstract] [Full Text] [PDF] |
||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Stroke Home | Subscriptions | Archives | Feedback | Authors | Help | AHA Journals Home | Search Copyright © 1998 American Heart Association, Inc. All rights reserved. Unauthorized use prohibited. |