SEARCH

SEARCH BY CITATION

Abstract

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

Objective

To determine the histopathologic basis of altered brain neurometabolites in neuropsychiatric systemic lupus erythematosus (NPSLE).

Methods

Brain neurometabolite concentrations in a 20-voxel area of the brain were determined premortem by magnetic resonance spectroscopy (MRS) in 7 individuals with NPSLE. Absolute concentrations of neurometabolite for N-acetylaspartate (NAA), choline, creatine, and lactate were measured. After the death of the patients, histopathologic changes were determined at autopsy of the brain and were matched voxel-by-voxel with the neurometabolites.

Results

The mean ± SD absolute concentrations of NAA (9.15 ± 1.78 mM in patients versus 12.2 ± 0.8 mM in controls; P < 0.01) and creatine (6.43 ± 0.16 mM in patients versus 6.90 ± 0.60 mM in controls; P < 0.003) were significantly reduced and the concentration of choline (2.51 ± 0.42 mM in patients versus 1.92 ± 0.32 mM in controls; P < 0.04) was significantly elevated in NPSLE patients as compared with controls. Widespread heterogeneous changes in the histologic features of the brain were present, including microinfarcts, microhemorrhages, bland angiopathy, thrombotic angiopathy with platelet and fibrin thrombi, neuronal necrosis in various states of resolution, reduced numbers of axons and neurons, vacuole and space formation among the fibers, reduced numbers of oligodendrocytes, reactive microglia and astrocytes, lipid-laden macrophages, and cyst formation. Neurometabolite abnormalities were closely associated with underlying histopathologic changes in the brain: 1) elevated choline levels were independently associated with gliosis, vasculopathy, and edema (r = 0.75, P < 0.004 in the multivariate model); 2) reduced creatine levels with reduced neuronal–axonal density and gliosis (r = 0.72, P < 0.002 in the multivariate model); 3) reduced NAA levels with reduced neuronal–axonal density (r = 0.66, P < 0.001 in the multivariate model); and 4) the presence of lactate with necrosis, microhemorrhages, and edema (r = 0.996, P < 0.0001 in the multivariate model).

Conclusion

Altered neurometabolites in NPSLE patients, as determined by MRS, are a grave prognostic sign, indicating serious underlying histologic brain injury.

Neuropsychiatric systemic lupus erythematosus (NPSLE) remains a major cause of morbidity and mortality, but the diagnosis of this clinical entity remains challenging (1). Magnetic resonance (MR) techniques have provided insights into both the underlying pathologic processes in NPSLE and new approaches for diagnosis (2–4). Proton magnetic resonance spectroscopy (MRS) is an MR technique that permits study of the neurometabolites in NPSLE (5–11). MRS has demonstrated neurometabolite disturbances in the brain tissues of patients with NPSLE, including reduced levels of N-acetylaspartate (NAA) and creatine, and increased levels of choline, lactate, lipid, and macromolecules (5–11). Reduced NAA levels may indicate neuronal injury or death, increased choline has been associated with gliosis and membrane breakdown, and lactate has been associated with ischemia, anaerobic metabolism, and inflammation (12–14).

A major problem with neuroimaging studies in general and MRS studies in NPSLE in particular has been the absence of brain histopathologic correlates to assist in the interpretation of the data derived from noninvasive MR techniques. In the present study, we correlated brain concentrations of neurometabolites measured premortem by MRS with brain histopathologic findings determined postmortem in the same NPSLE patients.

PATIENTS AND METHODS

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

Study patients.

This study was approved by the Institutional Review Board of the University of New Mexico and was conducted in compliance with the Declaration of Helsinki. Each patient provided written informed consent for the clinical, imaging, and postmortem autopsy studies. The purpose of this study was to determine the specific effect of different histopathologic patterns in the brain on the profiles of individual brain neurometabolites.

The overall study design was to obtain baseline MRS and clinical evaluations of the NPSLE patients at study entry, to repeat the MRS and clinical evaluations in patients who experienced active episodes of NPSLE, and to obtain further MRS and clinical evaluations at resolution of the NPSLE episodes (defined as 3 months after the NPSLE episode). Finally, at 3 years after enrollment, all surviving patients were to undergo a final MRS. At least 2 MRS measures were obtained in each patient; however, in certain individuals with widely varying histopathologic features, extra MRS measures were obtained, which permitted the recording of additional data points with widely varying histopathology/neurometabolite values.

A total of 168 patients with NPSLE were studied prospectively using MRS over a 5-year period. During this time, 12 patients died of complications of NPSLE. An autopsy of the brain was obtained in 7 of them, permitting direct comparison of premortem MRS results with postmortem histopathologic findings. These MRS results were compared with those in a normative database of 75 normal controls. The demographic characteristics of the controls and the NPSLE patients were similar in terms of age (mean ± SD age 38.5 ± 12.4 years in controls and 39.6 ± 13.9 years in NPSLE patients; P = 0.1) and sex (91% of controls and 93% of NPSLE patients were female; P = 0.1).

SLE was defined according to the American College of Rheumatology (ACR; formerly, the American Rheumatism Association) revised diagnostic criteria (15, 16). The diagnosis of SLE was confirmed by a rheumatologist (WLS) after a face-to-face interview with the patient, medical history, physical examination, medical chart review, and appropriate laboratory testing. SLE disease activity was determined with the SLE Disease Activity Index (SLEDAI), and SLE disease severity was determined with the Systemic Lupus International Collaborating Clinics/ACR Damage Index (SDI). Subscores for the neurologic components of each of these instruments were also determined, with the Neuro-SLEDAI consisting of the sum of the scores for seizure, psychosis, organic brain syndrome, visual abnormality, headache, and cerebral infarct, and the Neuro-SDI consisting of the sum of the scores for retinal pathology, optic atrophy, cognitive disorder, psychosis, seizures, stroke, neuropathy, and transverse myelitis (4, 17, 18). NPSLE was characterized according to the ACR nomenclature and case definitions for the disease, as previously described (19, 20). The clinical characteristics of the NPSLE cohort are summarized in Table 1.

Table 1. Characteristics of the study patients with fatal NPSLE*
Patient/age/sexNPSLE manifestationsMajor non-neurologic manifestationsDisease activityDisease severityNPSLE event causing deathMRI findingsInterval from MRI to autopsy
SLEDAINeuro-SLEDAISDINeuro-SDI
  • *

    The mean ± SD age of the 7 study patients was 39.6 ± 13.9 years. Their mean ± SD scores on disease activity and severity measures were 35.6 ± 21.0 on the Systemic Lupus Erythematosus Disease Activity Index (SLEDAI), 17.1 ± 8.6 on the Neuro-SLEDAI, 11.3 ± 5.2 on the Systemic Lupus International Collaborating Clinics/American College of Rheumatology Damage Index (SDI), and 3.00 ± 1.15 on the Neuro-SDI. Their mean ± SD interval from magnetic resonance imaging (MRI) to autopsy was 8.6 ± 16.4 weeks. NPSLE = neuropsychiatric SLE; GN = glomerulonephritis; MI = myocardial infarction; DVT = deep venous thrombosis; WMAs = white matter abnormalities.

1/44/FCerebrovascular disease, epilepsy, cognitive disorder, depressionGN, arthritis, valvular heart disease, pericarditis, MI, DVT5724194Cerebrovascular disease (acute ischemic stroke associated with active SLE)Moderate cortical atrophy, severe ventricular dilation, severe multifocal WMAs, chronic diffuse WMAs, multiple old and new infarcts2 weeks
2/56/MCerebrovascular disease, seizure disorder, cognitive disorder, depressionPleural effusion, arthritis, pericarditis, valvular heart disease, MI, DVT6524183Acute confusional state, seizure disorder (associated with active SLE)Moderate cortical atrophy, moderate ventricular dilation, extensive focal WMAs, multiple calcifications, increased T2 signal in basal cisterns, periventricular WM, and centrum semiovale1 month
3/22/FAcute confusional state, seizure disorder, cognitive disorder, depressionGN, arthritis, pericarditis, valvular heart disease, MI, vasculitis, limb necrosis248113Acute confusional state, seizure disorder (associated with active SLE)Moderate cortical atrophy and ventricular dilation, minimal focal WMAs, few punctate hypointense lesions, late brain edema, tonsillar herniation4 days
4/43/FCerebrovascular disease, cognitive disorder, depressionArthritis, pericarditis, valvular heart disease, DVT, osteonecrosis12884Cerebrovascular disease (acute ischemic stroke)Severe cortical atrophy, severe ventricular dilation, multiple cortical infarcts, multiple small focal lesions, chronic frontoparietal diffuse WMAs1 month
5/19/FAcute confusional state, seizure disorder, cognitive disorder, depressionGN, arthritis, pericarditis, myocarditis, vasculitis, valvular heart disease442461Acute confusional state, seizure disorder (associated with active SLE)Moderate cortical atrophy and ventricular dilation, small focal lesions, few deep WM lesions, late cerebral edema2 weeks
6/42/FCerebrovascular disease, seizure disorder, cognitive disorder, depressionArthritis, valvular heart disease, finger necrosis12874Seizure disorder (epilepsy)Moderate cortical atrophy and ventricular dilation, focal WMAs, multiple resolved, cortical infarcts, cortical cysts1 year
7/45/FSeizure disorder, cognitive disorder, depressionGN, arthritis, pericarditis, chronic hepatitis, osteonecrosis3524102Seizure disorder (associated with active SLE)Minimal cortical atrophy, minimal ventricular dilation, moderate focal WMAs2 months

MRS data acquisition.

Premortem MRS data were acquired at 1.5T with a Signa scanner (GE Medical Systems), using a head coil for transmission of radiofrequency pulses and detection of signals. Sagittal T1-weighted images (echo time [TE] 16 msec; repetition time [TR] 600 msec) were used to select the location of the spectroscopic voxels. A stimulated echo acquisition mode pulse sequence, including water suppression, was used to sample 1 voxel (volume 12.6 cm3, TE 30 msec, and TR 2,000 msec, with 128 averages) within the left occipitoparietal white matter and another in primarily the occipital gray matter, as well as in other selected areas of interest. A spectrum from the same location using the same acquisition parameters, but without water suppression, was also acquired and used as a concentration reference.

Spectroscopic data were transferred to a Sun UltraSparcStation (Sun Microsystems) for analysis, using Magnetic Resonance User Interface (MRUI) software. Residual water resonances were removed using time-domain Hankel-Lanczos singular value decomposition filtering implemented in MRUI. Following water filtering, time-domain fitting of Gaussian line shapes to NAA, creatine, choline, and lactate was performed by variable projection. The areas corresponding to NAA, creatine, choline, and lactate were recorded. The presence of lactate was confirmed with a doublet peak at 1.3 parts per million using a 30-msec TE that inverted at 135 msec. The unsuppressed water spectrum was quantified using singular value decomposition and was used as an internal concentration reference after correcting for metabolite and water T1 and T2 effects and the contribution from the cerebrospinal fluid (21). Absolute concentrations of neurometabolites in lesional and nonlesional tissues were thus determined for NAA, choline, creatine, and lactate in all MRS and histopathology pairs. The values for neurometabolites in a standard left occipitoparietal white matter voxel in each NPSLE patient was compared with those in 75 normal control subjects.

Pathologic examination.

Brain autopsy was performed in each case, and gross pathologic changes were described and recorded. The reproducibility of the neuropathologist's interpretation has been shown to be consistently reliable, with 85–97% interobserver reproducibility (22). The neuropathologists were blinded to the clinical and neuroimaging data.

After fixation in 10% buffered formalin for 2 weeks, the brain was examined for gross pathologic changes, and then sectioned into standard coronal planes. Each coronal section was then examined for gross changes, including obvious hemorrhage, focal atrophy, cyst formation, calcifications, and meningeal abnormalities. After inspection of the gross pathology, regions of the brain corresponding to the spectroscopic voxels, as well as standard samples from the cerebral lobes and lesions, were sampled. Surface anatomy was used to determine the location of individual voxels. These blocks were embedded in paraffin, and the sections were stained with hematoxylin and eosin, Luxol fast blue/periodic acid–Schiff. The neuropathologist then detailed in a formal report the histopathologic changes in each sampled area. From these reports, the histopathologic changes in each voxel location were ranked on a categorical scale (0 = none, 1 = minimal, 2 = moderate, and 3 = severe) for each of the following 6 histopathologic features commonly seen postmortem in NPSLE patients: 1) reduced neuronal–axonal density (reduced cellularity, neuronal and/or axonal dropout, necrosis, rarification, or cellular absence [cyst formation]), 2) necrosis (infarction, cellular debris, neuronal/axonal death, cell swelling, or disintegration, and inflammatory infiltration), 3) microhemorrhages (small punctate hemorrhages), 4) gliosis (glial hyperplasia), 5) vasculopathy (microthrombi, vascular remodeling, perivascular inflammation, vasculitis), and 6) edema (obvious cellular swelling and/or edema by MR neuroimaging).

Statistical analysis.

Data were analyzed using StatView SE+Graphics, version 1.04 (Abacus Concepts). Measurement data were compared by Student's t-test. Univariate relationships between neurometabolites and histopathologic features were determined with Kendall correlations. Multivariate associations were determined with logistic regression modeling, with nonsignificant association variables deleted until the final model predicted significant relationships between the histopathologic predictors and the individual neurometabolites. In order to compare categorical data with categorical data, individual neurometabolite concentrations were converted to a categorical variable. The quartiles ranking of each MRS spectrum were determined from the database of spectra obtained under identical conditions from 168 SLE patients, where each metabolite was ranked into quartiles based on the absolute concentration of the metabolite within the ordered rankings. Each neurometabolite concentration from the neurometabolite/histopathology pairs fell into a ranking quartile in this database, thus providing a categorical neurometabolite score of 1, 2, 3, or 4, which corresponded to the first through fourth quartiles, respectively. These paired categorical neurometabolite/histopathology scores were then used in the correlation analyses.

RESULTS

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

The characteristics of the 7 patients with fatal NPSLE are shown in Tables 1 and 2. The majority of the NPSLE patients (6 of 7) were female. Their mean ± SD age was 39.6 ± 13.9 years. Their mean ± SD scores on the disease activity and severity assessments were 35.6 ± 21.0 for the SLEDAI, 17.1 ± 8.6 for the Neuro-SLEDAI, 11.3 ± 5.2 for the SDI, and 3.00 ± 1.15 for the Neuro-SDI. The mean ± SD MRS voxels per patient was 2.86 ± 1.86. The high SLEDAI and SDI scores indicated that these patients had very active and aggressive SLE directly prior to death, and all of them had some involvement of component of NPSLE prior to and/or during the episode associated with their death (Table 1). Table 2 summarizes the serologic profiles of the study patients.

Table 2. Serologic profiles of the study patients with fatal neuropsychiatric systemic lupus erythematosus*
 ANA titerdsDNA titerSmRNPSSASSBRibosomal PaCLLAC
IgG, GPL unitsIgM, MPL unitsIgA, APL units
  • *

    ANA = antinuclear antibody; dsDNA = double-stranded DNA; aCL = anticardiolipin antibodies; GPL = IgG phospholipid; MPL = IgM phospholipid; APL = IgA phospholipid; LAC = lupus anticoagulant (lupus-like inhibitor).

Patient
 11:3201:160PositiveNegativePositiveNegativeNegative456518Positive
 21:6401:640NegativeNegativeNegativeNegativeNegative65127Positive
 31:6401:640PositiveNegativePositiveNegativePositive80345Positive
 41:80NegativeNegativeNegativeNegativeNegativeNegative12459Positive
 51:6401:640PositiveNegativePositiveNegativeNegative322812Negative
 61:320NegativePositivePositivePositiveNegativeNegative145418Positive
 71:3201:40PositiveNegativeNegativeNegativePositive829Negative
Summary, %10071.471.414.357.10.0 28.657.171.728.371.4

There were significant differences in the levels of brain neurometabolites in the nonlesional occipitoparietal white matter of the NPSLE patients as compared with the controls. In the NPSLE patients, the absolute concentrations of NAA were 25% lower than those in the controls (mean ± SD 9.15 ± 1.78 mM versus 12.2 ± 0.8 mM [95% confidence interval (95% CI) –30.9, –19.1]; P < 0.01). The creatine level was 6.9% lower (mean ± SD 6.43 ± 0.16 in patients versus 6.90 ± 0.60 mM in controls [95% CI –13.5, –0.1]; P < 0.003), and the choline level was 30.7% higher (mean ± SD 2.51 ± 0.42 in patients versus 1.92 ± 0.32 mM in controls [95% CI 17.2, 44.3]; P < 0.04) in the patients than in the controls. Within the NPSLE group, voxels from lesional tissue (n = 13) as compared with nonlesional tissue (n = 7) assessed by MRS demonstrated 29.8% less NAA (mean ± SD 6.42 ± 2.3 mM in lesional versus 9.15 ± 1.78 mM in nonlesional tissue [95% CI –52.9, –6.8]; P < 0.009), 44.0% less creatine (mean ± SD 3.60 ± 2.40 mM in lesional versus 6.43 ± 0.16 mM in nonlesional tissue [95% CI –74.0, –14.0]; P < 0.001), and 60.2% more choline (mean ± SD 4.02 ± 0.52 mM in lesional versus 2.51 ± 0.42 mM in nonlesional tissue [95% CI 41.0, 79.3]; P < 0.001).

Examination of the MRS spectra revealed several different patterns of neurometabolite disturbance, including: 1) increased levels of lactate associated with acute ischemic injury, and with necrosis related to vascular thrombosis (patients 1 and 3); 2) reduced levels of NAA and creatine associated with old ischemic injury, with reduced cellularity (reduced neuronal–axonal density), cellular debris, and glial hyperplasia (patients 1, 2, 4, 6, and 7); 3) reduced levels of NAA associated with acute diffuse ischemic or cytotoxic cellular (neuronal–axonal) injury, without vascular thrombosis (patients 3 and 5); and 4) reduced levels of NAA and elevated levels of choline associated with chronic cellular rarification (neuronal–axonal dropout) related to chronic inflammation and accumulation of calcium concretions (patient 2). There were also widespread heterogeneous histologic changes, including microinfarcts, microhemorrhages, bland angiopathy, thrombotic angiopathy with platelet and fibrin thrombi, neuronal necrosis in various states of resolution, reduced numbers of axons and neurons, vacuole and space formation among the fibers, reduced numbers of oligodendrocytes, reactive microglia and astrocytes, lipid-laden macrophages, and cyst formation, and in 1 patient, calcium-containing concretions consistent with SLE-associated Fahr's disease (23).

Figure 1 shows paired MRS and histologic data from a patient with NPSLE (patient 4) who had antiphospholipid antibodies, increased SLE disease activity, and dementia. This patient experienced an acute cerebral infarction and died. The MRS spectrum in the region away from the acute infarct was located in the occipitoparietal white matter on neuroimaging, which demonstrated chronic diffuse increased intensity on T2-weighted and fluid-attenuated inversion recovery (FLAIR) images (results not shown). MRS (Figure 1, left) demonstrated reduced NAA levels, reduced creatine levels, and elevated choline levels, which on histologic examination (Figure 1, right), corresponded to linear areas of chronic neuronal and parenchymal hypocellularity associated with chronic vasculopathy of nearby blood vessels.

thumbnail image

Figure 1. Magnetic resonance spectroscopy (MRS) and histopathologic assessment of brain tissues in a patient with neuropsychiatric systemic lupus erythematosus (NPSLE) and dementia. Shown are paired data from an MRS performed premortem and histopathologic assessment performed postmortem in NPSLE patient 4, who had antiphospholipid antibodies, increased SLE disease activity, and dementia. This patient died of an acute cerebral infarction. The MRS spectrum (left), which was obtained in minimally abnormal occipitoparietal white matter, demonstrates reduced levels of N-acetylaspartate (NAA), reduced levels of creatine (Cre), and elevated levels of choline (Cho). A myoinositol (mI) peak is also shown. The histopathologic appearance of the same area of the brain (right) shows linear areas of chronic neuronal and parenchymal hypocellularity (short arrows) associated with chronic vasculopathy of nearby blood vessels (arrow) (Luxol fast blue/periodic acid–Schiff stained, original magnification × 100).

Download figure to PowerPoint

Figure 2 shows paired MRS and histologic data from a patient with NPSLE (patient 1) who had antiphospholipid antibodies, increased SLE disease activity, and multiple cerebral infarctions. This patient experienced an acute cerebral infarction and died. The neuroimaging results were characterized by hyperintensity on the FLAIR and T2-weighted images and by restricted diffusion on the diffusion-weighted image (results not shown). In the first few days after cerebral infarction, the MRS (Figure 2, left) showed the characteristic doublet of increased lactate levels (1.32 ppm) and low NAA levels during the early phase of infarction. During the resolution phase (2 weeks postinfarction), this same region was characterized by further reductions in NAA levels, increased choline and lipid levels, and persistent lactate (results not shown). After the patient died of further cerebrovascular events, the underlying histopathology in the brain (Figure 2, right) was assessed. Brain tissues demonstrated a typical resolving cerebral infarction, with a marked reduction in cellularity (neuronal–axonal numbers), necrotic amorphous material, infiltration with phagocytic cells, and glial hyperplasia. Surrounding blood vessels showed persistent thrombosis.

thumbnail image

Figure 2. Magnetic resonance spectroscopy (MRS) and histopathologic assessment of brain tissues in a patient with neuropsychiatric systemic lupus erythematosus (NPSLE) and acute cerebral infarction. Shown are paired data from an MRS performed premortem and histopathologic assessment performed postmortem in NPSLE patient 1, who had antiphospholipid antibodies, increased SLE disease activity, and multiple cerebral infarctions. This patient died of an acute cerebral infarction. Findings of MR imaging (results not shown) were characterized by hyperintensity on the fluid-attenuated inversion recovery and T2-weighted images and restricted diffusion on the diffusion-weighted image. During the first few days after a cerebral infarction, the MRS spectrum (left) showed the characteristic doublet of increased lactate at 1.32 ppm and low levels of N-acetylaspartate (NAA) during the early phase of infarction. Creatine (Cre), choline (Cho), and myoinositol (mI) peaks are also shown. During the resolving phase (2 weeks postinfarction), this same region was characterized by further reductions in NAA levels, increased levels of choline and lipid, and persistent lactate (results not shown). After the patient died of further cerebrovascular events, the histopathologic appearance of the area of the acute infarct of the brain was assessed (right) and demonstrated a typical resolving cerebral infarction, with marked reduction in cellularity (neuronal–axonal numbers) (long arrow), necrotic amorphous material (arrowheads and short arrow), infiltration with phagocytic cells, and glial hyperplasia. Surrounding blood vessels showed persistent thrombosis (thick arrows) (Luxol fast blue/periodic acid–Schiff stained, original magnification × 50).

Download figure to PowerPoint

Correlation and regression analyses across all samples were then used to explain the MRS versus histopathology relationships. Univariate correlations between the histopathologic findings and the brain neurometabolite levels demonstrated multiple associations. However, multivariate logistic analysis demonstrated that an elevated choline level was independently associated with gliosis (partial correlation coefficient β = 1.005, P < 0.0004), vasculopathy (β = 0.46, P < 0.03), and edema (β = 0.78, P < 0.006) (r = 0.75, P < 0.004 in the total model). Similarly, a reduced creatine level was independently associated with reduced neuronal–axonal density (β = –0.55, P < 0.03) and gliosis (β < –0.64, P < 0.03) (r = 0.72, P < 0.002 in the total model). A reduced level of NAA was independently associated with reduced neuronal–axonal density (r = 0.66, P < 0.001). The presence of lactate was independently associated with the presence of necrosis (β = 0.65, P < 0.0001), microhemorrhages (β = 0.31, P < 0.0001), and edema (β = 0.34, P < 0.0001) (r = 0.996, P < 0.002 in the total model).

DISCUSSION

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

In the present study, we compared the levels of brain neurometabolites, as determined by premortem MRS, with the findings of histopathologic examination of the brain, as determined postmortem, in patients with NPSLE. This experimental design, which paired MRS with specific histopathologic assessment, provides a direct, rather than inferred, pathologic basis for the profound changes in neurometabolite levels in patients with NPSLE (5–11, 24–27). The principal changes include reduced NAA and creatine levels and elevated choline (choline-containing compounds) and lactate levels associated with reduced neuronal density, increased glial elements, acute and resolving injury, and in the majority of patients, acute and chronic ischemic or nonspecific vasculopathy (Figures 1 and 2). These histopathologic changes are broadly consistent with the pathologic changes identified by microscopy as reported in classic autopsy series of NPSLE patients (28–35).

The neurometabolite NAA is located almost entirely in neurons and is the strongest peak in the MRS brain spectrum in adults (36–38). The precise function of NAA in neuronal metabolism as an osmolyte and metabolic reserve continues to be explored, but NAA appears to be essential for normal neurocognitive function (21). Reduced levels of NAA have been noted in many diseases, and this is generally interpreted as representing neuronal injury or death (5–14). NAA levels have previously been reported to be reduced in normal-appearing white matter and gray matter, as well as in focal lesions, of patients with NPSLE (5–14, 36–38). Reduced NAA levels in radiologically normal–appearing tissues in NPSLE patients have been correlated with small focal lesions seen on MRI elsewhere in the brain, suggesting that the decline in NAA might also be due to extensive microlesions, most likely microinfarcts, that are too small to be seen on MRI (9, 24). This is further supported by the close association between reduced NAA levels and the presence of IgG antiphospholipid antibodies in NPSLE patients, suggesting thrombotic microinfarcts as a common cause of this decline (26). However, decreased NAA levels have also been noted in patients with generalized seizures, psychosis, or confusional state, which are not always associated with thrombosis, indicating that other causes of injury, including global ischemia and nonischemic cytotoxic effects, might be involved (27).

The present study confirms the idea that the reduced NAA level in patients with NPSLE is a significant finding and is most closely associated with reduced cellularity (reduced neuronal–axonal density). A reduced creatine level was also associated with reduced neuronal–axonal density and the presence of gliosis and inflammation, which are markers of acute and chronic brain injury. These changes were present in both normal-appearing tissues and lesional tissues, but were much worse in the lesions.

Other neuronal metabolites seen on MRS might also play important roles in NPSLE (14, 38). The choline peak represents choline moieties (–N+[CH3]3) that are visible on MRS, primarily phosphocholine, glycerophosphocholine, and choline. Choline levels are often elevated in NPSLE patients and have been postulated to be related to disease activity, stroke, inflammation, or chronic white matter disease (5–14, 36–38). Increased choline is also observed in NPSLE patients without obvious stroke or in the normal-appearing white matter, indicating the presence of preclinical disease or considerable microscopic brain disease not easily identified using conventional imaging modalities (5, 26, 27). There is evidence that increased choline levels might be a measure of disease activity or evidence of reactive brain inflammation, and thus, elevated choline levels in NPSLE patients should be interpreted as an ominous sign (9, 27). This is further supported by the finding that choline levels are elevated in NPSLE patients with cognitive impairment, indicating that the functional impact of this disease may be reflected by both elevated choline and reduced NAA (20, 26). Increased choline was associated histologically with vascular and reactive changes, including gliosis, vasculopathy, inflammatory cell infiltration, and edema, similar to the findings in other diseases in which there are active or reactive brain processes (13, 14, 38).

The presence of elevated upfield peaks at 1.32 ppm arising from lipids, macromolecules, and lactate has previously been associated with elevated disease activity in patients with NPSLE (5–14, 36–38). This is similar to the neurometabolic changes found in diseases associated with persistent or slowly resolving membrane degradation, activation, and demyelination (14, 38). However, although ischemia is often important in NPSLE, definite lactate in nonlesional tissue has not previously been observed on MRS, suggesting that in anything but overt stroke or global ischemia, extensive fixed anaerobic metabolism is not a fundamental characteristic of NPSLE (5–15, 30–33, 36–39). It should also be noted that lactate and other upfield peaks can be observed in patients with nonvascular diseases, including brain abscess, carbon monoxide poisoning, and status epilepticus (36–38). In the present study, the presence of lactate was associated with severe histopathologic changes in the brain, including tissue necrosis, microhemorrhages, and cerebral edema (Figure 2).

Previous autopsy studies in NPSLE have demonstrated highly variable changes, from minimally to grossly abnormal features, with cortical atrophy, gross infarcts, gross hemorrhage, microhemorrhages, ischemic demyelination, multiple sclerosis–like demyelination, and rarely, subdural hematoma, vasculitis, or aneurysms (28–35, 39–46). Fatal cases of NPSLE have also been associated with stroke or complement consumption, leukostasis, leukoaggregation, ischemia, and extensive cortical microinfarcts, indicating acute immune-mediated vascular occlusion (32–40). Leukostasis was not observed in the present study, but infarcts, microinfarcts, neurocytotoxic changes, vasculopathy, and many nonspecific or minimal changes were present. The concept of certain forms of NPSLE manifesting as cerebral edema with only subtle correlates at autopsy is emerging, particularly for the diffuse manifestations of NPSLE, with edema having an ominous significance (3, 47).

Bland vasculopathy, characterized by vascular hyalinization, vessel tortuosity, endothelial proliferation, and perivascular gliosis, is a common, but nonspecific, finding in NPSLE that was also observed in patients in the present study. This is indistinguishable from a distinctive type of cerebral vasculopathy characterized by fibrin thrombi, widespread obstruction by a proliferation of intimal fibrous tissue and myointimal cells, varying stages of recanalization, and in the late stages, fibrous webs across arterial lumens, which may be identified in the small leptomeningeal arteries of patients with high levels of antiphospholipid antibodies or with Libman-Sacks endocarditis (41, 48, 49). Cortical microinfarcts have also been associated with thrombi or emboli due to antiphospholipid antibodies, cardiac lesions, dissection, fibromuscular dysplasia, vasculitis, or atherosclerosis (28–35). Perivascular cuffing of arterioles or venules with inflammatory cells may occur, but there is a general recognition now that true central nervous system vasculitis is rare in SLE (28–35, 40–46). In the present study, despite widespread vasculopathy, thrombosis, and microinfarcts, true vasculitis was observed in only 1 small focus in patient 2.

There are limitations to this study that are related both to the design and to the inherent difficulties in any autopsy study. SLE has highly variable mortality, thus, the predictability of death and the ability to obtain an autopsy within a fixed time interval from the previous imaging study was variable (Table 1). Therefore, in some patients, the histologic changes may have been more evolved and the pathologic changes more severe at autopsy than the MRS findings would suggest. Autopsy studies of this type are biased, since they rely on the death of patients who typically have more severe disease. Accordingly, our results cannot necessarily be translated to less severe or nonfatal cases of NPSLE. Such a study would require studying patients with less active NPSLE who died of other causes.

Furthermore, NPSLE as defined by the ACR case definitions does not include subclinical injury; therefore, altered neurometabolite levels and/or histopathologic changes are not pathognomonic of clinical NPSLE (19). The study design also did not specifically address the role of antineuronal, excitotoxic, or anti–N-methyl-D-aspartic acid receptor antibodies; however, if present, it is likely that these antibodies would amplify the neuronal injury initiated by what appears to be the primary vascular insult evident in these study patients (50). Another limitation of this study is the relatively small number of patients. Clearly, a much larger multisite collaborative study is required to define the individual neurometabolite and histopathologic patterns associated with each NPSLE subtype (19). Finally, this study did not use specific stains and quantitative stereology to precisely quantify neuronal–axonal density, glial numbers, numbers of obstructed blood vessels, and the type, quantity, and immunotypes of inflammatory infiltrates; however, this study provides the scientific justification for including quantitative brain stereology in future autopsy studies in NPSLE (13).

Despite the inherent limitations and technical difficulties of autopsy studies, the findings of the present study confirm that abnormal brain neurometabolites measured premortem by MRS are indicative of significant underlying histopathologic changes consistent with acute and chronic brain injury. The findings of this study further support the concept of NPSLE as an aggressive and progressive brain disease and suggest that abnormal brain neurometabolites measured by MRS may indicate underlying histopathologic evidence of brain injury.

AUTHOR CONTRIBUTIONS

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

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Sibbitt 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 conception and design. Brooks, Sibbitt, Kornfeld, Jung, Bankhurst, Roldan.

Acquisition of data. Brooks, Sibbitt, Kornfeld, Jung, Bankhurst, Roldan.

Analysis and interpretation of data. Brooks, Sibbitt, Roldan.

Acknowledgements

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

The authors thank Drs. David Weers, Paul Mullins, and Charles Gasparovic for related developmental work, Ms Helen Petropoulos for assistance with data analysis, Dr. Blaine Hart for screening the localizing images, and Janeen Sharrar, RN, for her work as study coordinator.

REFERENCES

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