ARTHRITIS & RHEUMATISM
Vol. 50, No. 9, September 2004, pp 2877–2881
DOI 10.1002/art.20654
© 2004, American College of Rheumatology
Selective Gray Matter Damage in Neuropsychiatric Lupus
A Magnetization Transfer Imaging Study
S. C. A. Steens, F. Admiraal-Behloul, G. P. Th. Bosma, G. M. Steup-Beekman, H. Olofsen,
S. le Cessie, T. W. J. Huizinga, and M. A. van Buchem
Objective. Damage of brain parenchyma in pa-
tients with primary diffuse neuropsychiatric systemic
lupus erythematosus (NPSLE) has been indicated by
magnetization transfer imaging (MTI). However, the
location of MTI abnormalities is unknown. This study
was undertaken to assess the distribution of MTI ab-
normalities over gray matter (GM) and white matter
(WM) in SLE patients with a history of NP symptoms
without explanatory magnetic resonance imaging (MRI)
evidence of focal disease.
Methods. MTI was performed in 24 female SLE
patients with a history of diffuse NP symptoms and 24
healthy female controls. Magnetization transfer ratio
(MTR) maps were calculated for GM and WM sepa-
rately, and GM and WM MTR histograms were gener-
ated. Univariate and multivariate analyses with age as
an additional covariate were performed on the histo-
gram parameters peak location (PL), peak height (PH),
and mean MTR.
Results. Compared with controls, significantly
reduced PH (mean ⴞ SD 136 ⴞ 22 arbitrary units
versus 151 ⴞ 13 arbitrary units) and mean MTR
(33.3 ⴞ 1.0 percent units versus 33.6 ⴞ 0.5 percent
units) were found in the GM of NPSLE patients (P ⴝ
0.002 and P ⴝ 0.033, respectively, in multivariate ana-
lyses). No significant differences were observed for WM
MTR parameters.
S. C. A. Steens, MD, F. Admiraal-Behloul, PhD, G. P. Th.
Bosma, MD, MA, PhD, G. M. Steup-Beekman, MD, H. Olofsen, MSc,
S. le Cessie, PhD, T. W. J. Huizinga, MD, PhD, M. A. van Buchem,
MD, PhD: Leiden University Medical Center, Leiden, The Nether-
lands.
Address correspondence and reprint requests to S. C. A.
Steens, MD, Department of Radiology, C2-S, Leiden University
Medical Center, PO Box 9600, 2300 RC, Leiden, The Netherlands.
E-mail: s.c.a.steens@lumc.nl.
Submitted for publication November 7, 2003; accepted in
revised form May 12, 2004.
2877
Conclusion. This is the first study to demonstrate,
using MTI, that in SLE patients with a history of NP
symptoms and without explanatory focal abnormalities
on MRI, the GM is particularly affected. These findings
support the hypothesis that neuronal injury may under-
lie central nervous system manifestations in NPSLE.
Up to 75% of patients with systemic lupus ery-
thematosus (SLE) experience neuropsychiatric symp-
toms indicative of central nervous system (CNS) involve-
ment. In primary neuropsychiatric SLE (NPSLE), these
NP manifestations are attributed to the SLE disease
process itself, rather than to secondary factors such as
infections or metabolic disorders (1). Focal neurologic
NP syndromes are often associated with antiphospho-
lipid antibodies; however, the cause of diffuse neuro-
logic and psychiatric symptoms is largely unknown (2).
Findings of conventional magnetic resonance imaging
(MRI) of the brain are frequently unremarkable, and
abnormalities are nonspecific (3,4). However, quantita-
tive MRI techniques, such as magnetization transfer
imaging (MTI), diffusion-weighted imaging, and MR
spectroscopy (MRS), are now being assessed for their
diagnostic value (4).
Using MTI, a quantitative MRI technique that is
sensitive to both macroscopic and microscopic CNS
damage (4), abnormalities that had been invisible by
conventional MRI were demonstrated in the brain pa-
renchyma of NPSLE patients (5). Significant associa-
tions with parameters of neurologic, psychiatric, and
neuropsychological function demonstrated the clinical
relevance of these findings (6). Although the underlying
pathogenesis of diffuse brain damage in NPSLE is
largely unknown, recent findings of neuronal and astro-
cytic degradation products in the cerebrospinal fluid
(CSF) of SLE patients have focused attention on the
role of neuronal damage (7). Apart from small-vessel
2878 STEENS ET AL

disease as a suggested pathogenetic factor in diffuse

brain damage (2), an anti-DNA antibody that cross-
reacts with a neuronal receptor, and potentially induces
neuronal damage, has recently been identified in serum
and CSF of patients with SLE (8).
Neuronal damage may be the key to diffuse brain
damage in NPSLE. According to this hypothesis, the
gray matter (GM) would be preferentially affected in
NPSLE because of its high concentration of neurons. To
date, MTI analyses in NPSLE patients have focused on
the brain as a whole and have not been applied to
selectively study the GM compartment. To test the
hypothesis of preferential GM involvement in diffuse
NPSLE, we assessed the GM and white matter (WM)
compartments separately by MTI, in SLE patients with a
history of NP symptoms without MRI evidence of
explanatory focal disease.

PATIENTS AND METHODS Figure 1. Segmented magnetization transfer ratio (MTR) maps of
Twenty-four female patients diagnosed as having SLE gray matter (GM) and white matter (WM) at 4 different levels.
according to the 1982 American College of Rheumatology Locations of the axial MTR maps are displayed on sagittal T1-
revised criteria (9) and 24 healthy female controls were weighted magnetic resonance images shown to the left.
included in the study (age range 19–65 years in both groups;
mean age 35 years in the patient group and 39 years in the
control group). All subjects provided written informed con- voxel was calculated by the equation (M0 ⫺ Ms)/M0, with M0
sent. Patients were recruited from the Leiden University representing the intensity of voxels in a nonsaturated state and
Medical Center Department of Rheumatology. NPSLE was Ms the intensity in a saturated state (6) (in-house software;
diagnosed based on clinical findings, after exclusion of other Division of Image Processing, Department of Radiology, Lei-
possible causes of NP symptoms (10). The mean duration of den University Medical Center). Using Statistical Parametric
SLE was 8 years (range 1–29); time since the onset of NPSLE Mapping ’99 (Wellcome Department of Cognitive Neurology,
varied from 1 month to 18 years (mean 5 years). At the time of Institute of Neurology, London, UK [11]), Ms images were
scanning, no active NP symptoms or concurrent neurologic or segmented automatically based on a cluster analysis using prior
psychiatric diseases were present, and there was no indication spatial probabilities, signal intensity information, and maxi-
of previous neurologic or psychiatric disease in any of the mum image inhomogeneity correction. Probability maps for
subjects. Patients were included only if MRI revealed no GM, WM, and CSF were inspected visually to confirm ade-
abnormalities that could possibly explain a history of NP
symptoms. Patient characteristics, specific NP manifestations,
and abnormalities on conventional MRI are described in a
previous report on the same patient group (6).
MRI was carried out with a 1.5T MR scanner (Philips,
Best, The Netherlands). Scans were aligned parallel to the
axial plane through the anterior and posterior commissure and
covered the whole brain in all sequences. Conventional T1-
weighted spin-echo, fluid-attenuated inversion recovery and
dual (proton density and T2-weighted) images (6) were ac-
quired in all patients and interpreted by an experienced
neuroradiologist (MAvB). For MTI, a 3-dimensional gradient-
echo pulse sequence with an echo time/repetition time of 6/106
msec and a flip angle of 12° was chosen, resulting in proton
density contrast in the absence of MT saturation pulses (6). A
256 ⫻ 256 matrix with 50% acquisition and 220-mm field of
view was used for 28 contiguous 5-mm slices. Two consecutive Figure 2. Histograms of the magnetization transfer ratio (MTR) (in
sets of axial images were acquired, with and without a sinc- percent units [pu], after correction for volumes) for gray matter (solid
shaped radiofrequency saturation pulse 1,100 Hz upfield of lines) and white matter (broken lines) in healthy controls (gray lines)
H2O resonance (6). Scanning time for MTI was ⫾11 minutes. and patients with neuropsychiatric systemic lupus erythematosus
The magnetization transfer ratio (MTR) for every (black lines).
MTI STUDY REVEALS GM INVOLVEMENT IN NPSLE 2879

Table 1. MTR histogram parameters and results of univariate analysis and multivariate analysis with age as an additional covariate*
Univariate analysis Multivariate analysis
Healthy NPSLE
controls, patients, Mean ⫾ SEM Mean ⫾ SEM
Parameter mean ⫾ SD mean ⫾ SD difference† P difference† P
GM PL, pu 34.4 ⫾ 0.5 34.3 ⫾ 0.9 0.13 ⫾ 0.21 0.553 0.25 ⫾ 0.19 0.210
WM PL, pu 37.5 ⫾ 0.6 37.9 ⫾ 1.1 0.46 ⫾ 0.26 0.078 ⫺0.35 ⫾ 0.25 0.165
GM PH, AU 151 ⫾ 13 136 ⫾ 22 14.6 ⫾ 5.2 0.007 16.8 ⫾ 5.1 0.002
WM PH, AU 189 ⫾ 16 181 ⫾ 23 8.3 ⫾ 5.6 0.146 9.0 ⫾ 5.7 0.122
GM, mean MTR, pu 33.6 ⫾ 0.5 33.3 ⫾ 1.0 0.31 ⫾ 0.23 0.195 0.46 ⫾ 0.21 0.033
WM, mean MTR, pu 37.4 ⫾ 0.5 37.8 ⫾ 1.0 ⫺0.36 ⫾ 0.23 0.126 ⫺0.25 ⫾ 0.22 0.258

* MTR ⫽ magnetization transfer ratio; NPSLE ⫽ neuropsychiatric systemic lupus erythematosus; GM ⫽ gray matter; PL ⫽ peak location; pu ⫽
percent units; WM ⫽ white matter; PH ⫽ peak height; AU ⫽ arbitrary units (corrected for volumes).
† Estimated by the least squares method.

quate extraction of intracranial contents. To avoid partial
volume effect on the segmentation of GM, WM, and CSF,
voxels with a ⬎10% probability of belonging to the CSF were
excluded. Of the remaining voxels, only those for which the
probability of belonging to GM or WM reached ⬎80% were
considered for the histogram analysis. Based on these conser-
vative thresholds, binary masks were produced for the GM and
WM and were applied to the original MTR maps, producing
pure GM and WM MTR maps. From these MTR maps,
histograms were generated and the corresponding peak loca-
tion (PL), peak height (PH), and mean MTR were derived for
GM and WM separately. Since the PH is dependent on both
lesion load and the physiologic variation in volumes of the
different compartments, a volume correction was performed
by dividing the number of voxels for each MTR value by the
total number of voxels for that tissue type (6).
Univariate analyses and multivariate linear regression
analyses with age as an additional covariate were performed to
test for significant differences in GM and WM MTR para-
meters between healthy controls and NPSLE patients. Vari-
ances in MTR parameters between NPSLE patients and
controls were assessed using Levene’s test. P values less than
0.05 were considered significant.
RESULTS
All images showed an accurate segmentation of
GM and WM, as seen in the example presented in
Figure 1. The stringent probability thresholds resulted in
an exclusion of voxels with partial volume effects at the
interfaces of GM, WM, and CSF (Figure 1), providing
pure GM and WM MTR maps. WM mean MTR values
were higher than those in the GM, although there was
considerable overlap of MTR values between GM and
WM in both the patient group and the control group
(Figure 2).
Univariate analysis revealed a significantly lower
GM PH in NPSLE patients as compared with healthy
controls (P ⫽ 0.007) (Table 1). A subsequent multivar-
iate analysis with age as an additional covariate con-
firmed the significantly reduced GM PH in NPSLE
patients (P ⫽ 0.002), and also revealed a significantly
reduced GM mean MTR as compared with healthy
controls (P ⫽ 0.033). The degree of variation among the
NPSLE patients was significantly greater than that
among controls for GM PH (P ⫽ 0.012) and GM mean
MTR (P ⫽ 0.005) only. For GM PL and for all WM
parameters, no significant differences were observed
(P ⬎ 0.05).
DISCUSSION
Recently, MTI has been applied to detect and
quantify global, functionally relevant brain damage in
patients with NPSLE (5,6). However, no information on
the anatomic distribution of MTI abnormalities has
previously been obtained. This is the first study to
demonstrate MTI abnormalities indicative of parenchy-
mal brain damage specifically in the GM in SLE patients
with a history of NP symptoms but without explanatory
focal abnormalities on MRI. When NPSLE patients
were compared with healthy controls, differences in
volumetric MTR measures indicating loss of structural
integrity of tissue were observed in the GM, but not the
WM. Moreover, the increased variance of PH and mean
MTR values in the NPSLE patients, reflecting variation
of cerebral lesion load in this group of patients who were
heterogeneous in terms of symptoms and disease dura-
tion, was found only in the GM. These findings of GM
involvement correspond with results of histologic, immu-
nologic, and other quantitative neuroimaging studies.
The exact histologic changes in NPSLE remain
unclear. However, several histopathologic studies have
demonstrated vasculopathy and microinfarcts in patients
with NPSLE (2). Strikingly, the classic histopathologic
study by Johnson and Richardson showed vascular
changes predominantly in the cerebral cortex and brain
stem of SLE patients (12). Moreover, a recent immuno-
2880
chemical study demonstrated increased levels of the
light subunit of the neurofilament triplet protein and the
glial fibrillary acidic protein in the CSF of NPSLE
patients, indicating neuronal and astrocytic degenera-
tion (7). Due to the high concentration of neurons in
GM, neuronal and astrocytic damage will probably lead
to a decrease of mainly GM structural integrity, which is
in accordance with the findings of the present study.
Pathogenetically, thrombotic vasculopathy and
cerebral infarcts may account for some focal NP syn-
dromes; however, it is unlikely that all diffuse neurologic
and psychiatric manifestations in NPSLE are attribut-
able solely to these factors. In fact, there has been a
gradual realization that SLE and NPSLE may represent
a more diffuse disease process, including nonthrombotic
vascular injury at the endothelial level and involvement
of complement cascades, cytokines, autoantibodies, and
hormones (2). Since GM is more susceptible than WM
to many disease processes, such as hypoperfusion (13), it
will be particularly affected by any kind of CNS injury.
Small-vessel disease itself may also increase blood–brain
barrier permeability, which facilitates the entrance of
antineuronal antibodies such as the recently identified
anti–N-methyl-D-aspartate antibody into the brain pa-
renchyma (2,8). In either case, because of the higher
concentration of neurons in GM, the GM will be partic-
ularly affected.
Abnormal GM has also been observed using
spin-spin (T2) relaxation time measurements and MRS
(3,4). An increase in T2 relaxation time indicated cere-
bral edema and significant metabolic disturbance (14),
while MRS revealed a decreased N-acetylaspartate:crea-
tine (NAA:Cre) ratio, indicative of neuronal injury (15).
With both techniques, however, WM abnormalities in
NPSLE patients were also observed. Two explanations
with respect to the lack of abnormal MTR findings in the
WM in the present study can be suggested. First, our
patients may comprise a subgroup of NPSLE patients
without WM involvement, since only patients who did
not have focal lesions that could explain the history of
NPSLE were included. Second, MTI and MRS reflect
different histologic conditions. MTI mainly reflects my-
elin and axonal concentrations, whereas the NAA:Cre
ratio is a neuronal marker. If neuronal damage is the
dominant underlying histologic change, it comes as no
surprise that MTR values in the WM are normal, since
in the WM these values are determined mainly by the
integrity of myelin and axonal concentrations. In the
GM, however, myelin is present in a much lower con-
centration and therefore contributes less to the observed
STEENS ET AL
MTR values. The effect of neuronal loss might therefore
be easier to detect with MTI in that compartment.
It could be argued that decreases in GM MTR
could be due to retrograde neuronal degeneration re-
sulting from transection of axons traversing WM lesions.
However, no significant differences in WM MTR histo-
gram parameters were observed between NPSLE pa-
tients and healthy controls, and therefore it is unlikely
that our GM MTR findings are secondary to a WM
abnormality. Another potential confounder is the pres-
ence of cerebral atrophy, which is known to occur in
NPSLE (3,4). Cerebral atrophy increases the number of
voxels at the parenchymal cortex containing a mixture of
GM and CSF, and classification of those voxels as GM
could also have resulted in decreased GM MTR values.
However, with the stringent classification thresholds
used, partial voluming effects were reduced as much as
possible.
Patients and healthy controls were matched for
age (P ⫽ 0.193 by independent-samples t-test). How-
ever, since MTR parameters may be differently associ-
ated with age in healthy subjects and SLE patients, a
multivariate analysis was performed in which age was
included as an additional covariate, thereby reducing the
residual variance. The significantly reduced GM PH in
NPSLE patients was confirmed in this analysis, and
additionally, a significantly reduced GM mean MTR was
noted. Both the higher least square mean, estimating the
mean difference between the groups, and a lower stan-
dard error contributed to this significant difference. A
potential difference in association between age and
MTR parameters in controls and NPSLE patients will be
the subject of further study.
In conclusion, this is the first study to demon-
strate that in SLE patients with a history of NP symp-
toms and without explanatory focal abnormalities seen
on MRI, damage occurs in the GM. This observation
supports the model of neuronal damage in diffuse
NPSLE, and can be explained by the greater suscepti-
bility of GM to the sequelae of small-vessel disease or to
the presence of antineuronal antibodies, for which new
evidence (8) has recently been found.
REFERENCES
1. Hermosillo-Romo D, Brey RL. Diagnosis and management of
patients with neuropsychiatric systemic lupus erythematosus
(NPSLE). Best Pract Res Clin Rheumatol 2002;16:229–44.
2. Scolding NJ, Joseph FG. The neuropathology and pathogenesis of
systemic lupus erythematosus. Neuropathol Appl Neurobiol 2002;
28:173–89.
3. Sibbitt WL, Sibbitt RR, Brooks WM. Neuroimaging in neuropsy-
MTI STUDY REVEALS GM INVOLVEMENT IN NPSLE
chiatric systemic lupus erythematosus. Arthritis Rheum 1999;42:
2026–38.
4. Huizinga TW, Steens SC, van Buchem MA. Imaging modalities in
central nervous system systemic lupus erythematosus. Curr Opin
Rheumatol 2001;13:383–8.
5. Bosma GP, Rood MJ, Huizinga TW, de Jong BA, Bollen EL, van
Buchem MA. Detection of cerebral involvement in patients with
active neuropsychiatric systemic lupus erythematosus by the use of
volumetric magnetization transfer imaging. Arthritis Rheum 2000;
43:2428–36.
6. Bosma GP, Middelkoop HA, Rood MJ, Bollen EL, Huizinga TW,
van Buchem MA. Association of global brain damage and clinical
functioning in neuropsychiatric systemic lupus erythematosus.
Arthritis Rheum 2002;46:2665–72.
7. Trysberg E, Nylen K, Rosengren LE, Tarkowski A. Neuronal and
astrocytic damage in systemic lupus erythematosus patients with
central nervous system involvement. Arthritis Rheum 2003;48:
2881–7.
8. DeGiorgio LA, Konstantinov KN, Lee SC, Hardin JA, Volpe BT,
Diamond B. A subset of lupus anti-DNA antibodies cross-reacts
with the NR2 glutamate receptor in systemic lupus erythematosus.
Nat Med 2001;7:1189–93.
9. Tan EM, Cohen AS, Fries JF, Masi AT, McShane DJ, Rothfield
2881
NF, et al. The 1982 revised criteria for the classification of systemic
lupus erythematosus. Arthritis Rheum 1982;25:1271–7.
10. ACR Ad Hoc Committee on Neuropsychiatric Lupus Syndromes.
The American College of Rheumatology nomenclature and case
definitions for neuropsychiatric lupus syndromes. Arthritis Rheum
1999;42:599–608.
11. Ashburner J, Friston KJ. Voxel-based morphometry: the methods.
Neuroimage 2000;11:805–21.
12. Johnson RT, Richardson EP. The neurological manifestations of
systemic lupus erythematosus. Medicine (Baltimore) 1968;47:
337–69.
13. Takahashi S, Higano S, Ishii K, Matsumoto K, Sakamoto K,
Iwasaki Y, et al. Hypoxic brain damage: cortical laminar necrosis
and delayed changes in white matter at sequential MR imaging.
Radiology 1993;189:449–56.
14. Sibbitt WL, Brooks WM, Haseler LJ, Griffey RH, Frank LM, Hart
BL, et al. Spin-spin relaxation of brain tissues in systemic lupus
erythematosus: a method for increasing the sensitivity of magnetic
resonance imaging for neuropsychiatric lupus. Arthritis Rheum
1995;38:810–8.
15. Brooks WM, Sabet A, Sibbitt WL, Barker PB, van Zijl PC, Duyn
JH, et al. Neurochemistry of brain lesions determined by spectro-
scopic imaging in systemic lupus erythematosus. J Rheumatol
1997;24:2323–9.