Introduction
Systemic lupus erythematosus (SLE) is an autoimmune disease in which circulating auto-antibodies target a variety of body tissues including the brain. Neuropsychiatric (NP) manifestations in SLE are diverse and may include major syndromes, such as stroke and seizures, or other problems including headaches, mood disorders, and cognitive changes. Estimations of NP manifestations in SLE range from 37 to 95%, a variability that largely reflects differences in classification, subject selection, and attribution of cause (Hanly, Kuznetsova, & Fisk, Reference Hanly, Kuznetsova, Fisk, Wallace and HaHn2007). In many studies, SLE patients have been classified as having overt neurological or psychiatric disorders (NPSLE). NPSLE has been further classified as active or inactive (Denburg, Carbotte, & Denburg, Reference Denburg, Carbotte and Denburg1987) and from severe to mild. Those without NP syndromes or histories of central nervous system involvement have typically been classified as nonNPSLE or SLE-only. In the most recent nomenclature for NP disorders in SLE (“The American College of Rheumatology Nomenclature and Case Definitions for Neuropsychiatric Lupus Syndromes,” 1999), 19 syndromes were identified, including cognitive dysfunction. However, the distinction between nonNPSLE and NPSLE can be confusing because patients with nonNPSLE can still manifest cognitive dysfunction. While the ACR categories will be used in this review, the concept of mild cognitive dysfunction in SLE (MCD-SLE) will also be introduced to describe SLE patients who have cognitive impairment less severe than those with the cognitive dysfunction of NPSLE. The presence of MCD has clearly been established in nonNPSLE (Kozora, Reference Kozora, Morgan and Ricker2008), indicating that subtle brain pathology likely affects many SLE patients early in the disease. The course and prognosis of cognitive decline in SLE are largely unknown, and mechanisms underlying this problem have remained elusive, but the hypothesis that MCD-SLE is a precursor to the moderate to severe cognitive dysfunction suggested by NPSLE may help clarify these issues.
WM is essential for information transfer, enabling rapid axonal conduction and the synchronous, highly integrated activity of distributed neural networks. The cognitive dysfunction of SLE suggests that WM injury may be centrally involved in pathogenesis, as impaired attention, executive function, and information processing speed are commonly observed. These deficits, combined with the relative preservation of language, are more reminiscent of the frontal—subcortical dementias and closely resemble the cognitive profile of patients with multiple sclerosis (MS). Over 100 WM disorders are recognized, and all have been associated with some form of neurobehavioral disturbance (Filley, Reference Filley2001). Damage to WM tracts may thus be relevant to cognitive dysfunction in SLE. This review will offer a summary of neuropsychological and neuroimaging studies in SLE that support the association between neurobehavioral changes and cerebral WM damage. The hypothesis will also be presented that early myelinopathy may underlie MCD-SLE, in contrast to the more extensive white and gray matter pathology associated with NPSLE.
Neuropsychological and Neuroimaging Findings in SLE
Neuropsychological Studies
Studies using standard neuropsychological tests have reported cognitive impairment in 14–79% of SLE patients, and a comprehensive review of this literature has been published (Kozora, Reference Kozora, Morgan and Ricker2008). Inconsistencies across studies are largely methodological, including differences in the characteristics of the sample and the selection of tests used to classify impairment. Deficits in attention, verbal and nonverbal learning and recall, verbal and nonverbal fluency, complex psychomotor functions, visuospatial skills, and motor dexterity are identified when SLE patients are compared to healthy control subjects. The prevalence of cognitive dysfunction in SLE patients is higher in NPSLE (∼50%) compared to nonNPSLE (∼25%). Attention and processing speed are the most commonly impaired areas in these patients, with NPSLE patients demonstrating greater severity. The duration of disease, disease activity, use of prednisone, and depression have not been consistent mediators of cognitive dysfunction in SLE. In contrast, antiphospholipid antibodies (aPL) have been associated with cognitive dysfunction, and although rarely studied, measures of inflammation and pro-inflammatory cytokine activation have been similarly implicated (Kozora, Reference Kozora, Morgan and Ricker2008; Shucard, Gaines, Ambrus, & Shucard, Reference Shucard, Gaines, Ambrus and Shucard2007). Investigations of immune activity have primarily used peripheral measures in the serum, and continued analyses with cerebrospinal fluid (CSF) measures are likely to yield stronger correlations.
Neuroimaging Studies
Several neuroimaging techniques are available to investigate brain abnormalities in patients with SLE. The data generally support a substantial burden of both macrostrucutral and microstuctural WM disease in these patients.
Magnetic resonance imaging (MRI) analysis using conventional instruments reveals diffuse WM hyperintensities (WMHI), infarcts, hemorrhage, and cerebral atrophy in SLE. WMHI occur throughout the cerebral hemispheres, most often in the periventricular regions, in up to 70% of SLE patients (Appenzeller, Pike, & Clarke, Reference Appenzeller, Pike and Clarke2008). In several well designed studies with clearly defined characterizations of NP activity in adult SLE patients, the number of WMHI has been significantly higher in NPSLE patients compared to those with nonNPSLE (Ainiala et al., Reference Ainiala, Dastidar, Loukkola, Lehtimaki, Korpela, Peltola and Hietaharju2005; Appenzeller, Vasconcelos Faria, Li, Costallat, & Cendes, Reference Appenzeller, Vasconcelos Faria, Li, Costallat and Cendes2008; Handa et al., Reference Handa, Sahota, Kumar, Jagannathan, Bal, Gulati and Wali2003). With the use of both 1.5 and 2.0 T scanners, WMHI were documented in up to 35–50% of well-defined adult nonNPSLE patients, but data on control subjects in these studes were not presented (Abreu et al., Reference Abreu, Jakosky, Folgerini, Brenol, Xavier and Kapczinsky2005; Appenzeller, Vasconcelos Faria, et al., Reference Appenzeller, Vasconcelos Faria, Li, Costallat and Cendes2008). In a recent study using a 3.0 T scanner, no differences were noted in WM volume or number and volume of WM lesions in well-defined adult nonNPSLE patients compared to controls (Filley et al., Reference Filley, Kozora, Brown, Miller, West, Arciniegas and Zhang2009).
Magnetic resonance spectroscopy (MRS) is the only neuroimaging technique that identifies and quantifies chemicals in living tissue. MRS data are typically displayed as spectra of nuclear resonances from active compounds in a selected area, and are thought to demonstrate abnormalities in tissue metabolism. The peaks of MRS spectra reflect the chemical structure and concentration of individual metabolites. Creatine (Cr) is a storage form of high-energy phosphates and is used as a reference marker. N-acetylaspartic acid (NAA) is produced by neurons and is a marker of neuronal health; levels decrease markedly with brain insult and nonspecific neuronal loss. Choline (Ch) is a precursor to acetylcholine, an essential chemical for neuronal membrane integrity and synaptic transmission. Elevated Ch/Cr is associated with increased cell production (as occurs with brain tumors), and with increased membrane turnover related to inflammation, demyelination, ischemia, and gliosis. Most SLE studies have focused on NAA/Cr and Ch/Cr in an effort to assess neuronal and myelin integrity.
For this review, MRS studies were selected if they included only SLE patients who met ACR criteria, measurements of NAA/Cr and Ch/Cr in cerebral WM, and similar age controls; 16 such studies were found (Table 1, A–P). These studies generally reported decreased NAA and increased Ch/Cr in both the normal-appearing (NAWM) and abnormal-appearing WM (AAWM). The studies differed in their methods, however, including variation in subject selection, sample size, and neuroimaging techniques. Specific details regarding SLE disease activity, such as duration of disease, level of disease, and prednisone use were missing in many early studies (Table 1, A–D,G–I,N), and, when available, included confounders such as SLE samples with high disease activity (Table 1, B,D,J,M,O) or mean daily doses of prednisone over 15 mg (Table 1, B,C,F,M,O). Duration of disease varied significantly from 1 month to over 30 years, and was rarely studied in relation to MRS findings in studies with small sample sizes. Study groups were primarily women, and only two studies reported ethnicity (Table 1, K and P). Although mean age was between 30 and 40 years in all studies, several included subjects less than 18 (ages, 6–15; Table 1, D–G, H, J, L), and some included subjects over 50 (Table 1, A, B, N) and over 60 (Table 1, C, E, H). Classification of nonNPSLE and NPSLE varied as studies used different criteria to characterize NP activity. Some early studies (B, D, G) used previous NP classification (Denburg et al., Reference Denburg, Carbotte and Denburg1987), and most included active psychiatric and neurological conditions while not screening for or categorizing others such as head injury and substance abuse. Five later studies used the ACR NP classifications, four of which carefully screened out other problems such as head injury and substance abuse (Table 1, K, M, O, and P). A majority of the MRS studies were performed on 1.5 T scanners, with only K, M, O, and P using a more sensitive 2.0 T or 3.0 T magnet, potentially influencing and limiting MRS findings in earlier studies. As shown in Table 1, the areas studied also varied considerably, another factor that could influence results. In summary, WM abnormalities were often found in SLE, suggesting both microstructural and macrostructural WM damage, but methodological differences should be considered in data interpretation.
Table 1 Review of 16 MRS studies investigating (Ch/Cr) and NAA/Cr in SLE compared to controls
Note. NAA = n-acetylaspartate; Ch/Cr = choline/creatine; NAWM = normal-appearing white matter; AAWM = abnormal-appearing white matter; GM = gray matter; CC = corpus callosum; R = right; L = left; aPLs = antiphospholipid antibodies present.
Whereas decreased NAA and elevated Ch/Cr are both found in SLE, the earliest change may be higher Ch/Cr, which is more likely due to inflammatory myelinopathy than ischemia or gliosis at this disease stage (Filley et al., Reference Filley, Kozora, Brown, Miller, West, Arciniegas and Zhang2009). Studies with carefully characterized nonNPSLE patients suggest that increased Ch/Cr can occur without, and perhaps before lowered NAA/Cr (Appenzeller, Li, Costallat, & Cendes, Reference Appenzeller, Li, Costallat and Cendes2007; Filley et al., Reference Filley, Kozora, Brown, Miller, West, Arciniegas and Zhang2009; Sundgren et al., Reference Sundgren, Jennings, Attwood, Nan, Gebarski, McCune and Maly2005). MRS studies have also documented NAA/Cr and Ch/Cr abnormalities before MRI shows WMHI (Appenzeller et al., Reference Appenzeller, Li, Costallat and Cendes2007; Castellino et al., Reference Castellino, Govoni, Padovan, Colamussi, Borrelli and Trotta2005). Castellino and colleagues (Reference Castellino, Govoni, Padovan, Colamussi, Borrelli and Trotta2005) found that nonNPSLE patients with high Ch/Cr at baseline had more MRI WM abnormalities at follow-up. Appenzeller and colleagues (Reference Appenzeller, Li, Costallat and Cendes2007) reported that SLE patients (both NPSLE and nonNPSLE) whose disease activity increased over the course of the study had a concomitant decline in NAA/Cr, that Ch/Cr increased in SLE patients compared to controls, and that patients with elevated Ch/Cr and normal MRI at baseline developed WMHI approximately 1 year later.
Few studies have evaluated autoantibody activity in relation to MRS findings in SLE. SLE patients with antibodies against phospholipids (aPL) had higher levels of Ch/Cr than those without aPL (Sabet, Sibbitt, Stidley, Danska, & Brooks, Reference Sabet, Sibbitt, Stidley, Danska and Brooks1998), and a significant association between increased Ch/Cr and aPL was reported by Appenzeller and colleagues (Reference Appenzeller, Li, Costallat and Cendes2007). APL were also associated with increased number and volume of WMHI in SLE (Appenzeller, Vasconcelos Faria, et al., Reference Appenzeller, Vasconcelos Faria, Li, Costallat and Cendes2008).
Diffusion tensor imaging (DTI) provides an index of the structural integrity of WM by using quantitative directional diffusion properties of water molecules for each voxel. The technique is based on the principle of anisotropy, a term referring to the propensity for water in the normal state to diffuse along the direction of WM tracts. Damaged WM, in contrast, is characterized by isotropic diffusion, which is correspondingly less directional and more random. Key DTI measures include the apparent diffusion coefficient (ADC), mean diffusivity (MD), and fractional anisotropy (FA). DTI is particularly sensitive to microstructural changes in WM, including both myelin and axonal damage, and is an ideal tool for the study of WM because it can assess the structural integrity of specific tracts in relation to cognitive functions.
DTI studies suggest WM damage in well-defined NPSLE patients (Bosma et al., Reference Bosma, Steens, Petropoulos, Admiraal-Behloul, van den Haak, Doornbos and van Buchem2004; Hughes et al., Reference Hughes, Sundgren, Fan, Foerster, Nan, Welsh and Gebarski2007; Jung et al., Reference Jung, Caprihan, Chavez, Flores, Sharrar, Qualls and Roldan2010) and in relation to controls (Hughes et al., Reference Hughes, Sundgren, Fan, Foerster, Nan, Welsh and Gebarski2007; Jung et al., Reference Jung, Caprihan, Chavez, Flores, Sharrar, Qualls and Roldan2010; Zhang et al., Reference Zhang, Harrison, Heier, Zimmerman, Ravdin, Lockshin and Ulug2007) using 1.5 T scanners. A study of 34 NPSLE patients compared to controls reported increased MD in the frontal lobe and the internal capsule, and decreased FA in the corpus callosum (CC) (Zhang et al., Reference Zhang, Harrison, Heier, Zimmerman, Ravdin, Lockshin and Ulug2007). Abnormal findings were seen in patients with normal MRI scans, supporting the sensitivity of DTI to early WM changes. FA values were also decreased in frontobasal and temporal WM tracts of 12 SLE patients (NPSLE and nonNPSLE) (Emmer et al., Reference Emmer, Veer, Steup-Beekman, Huizinga, van der Grond and van Buchem2010) using a 3.0 T scanner. In a recent study, increased ADC was reported in the frontal lobe as well as the CC splenium and genu in 15 well-characterized nonNPSLE patients compared to controls (Ulug et al., Reference Ulug, Vo, Kozora, Ramos, Vega, Zimmerman and Lockshin2010) using a 3.0 T scanner. DTI thus shows promise as another method by which cerebral WM changes may be correlated with cognition in SLE (Filley, Reference Filley2009).
Cognition and White Matter Associations in SLE
Studies supporting an association between abnormal WM neuroimaging and cognitive deficits in SLE are beginning to accumulate. Early MRI investigations demonstrated that the number of WMHI was associated with cognition, particularly attention, in 20 nonNPSLE patients, while no relationship was found between cognitive change and cerebral atrophy (Kozora et al., Reference Kozora, West, Kotzin, Julian, Porter and Bigler1998). Volumetric WM analyses were then conducted, and the CC was studied as a readily identifiable WM tract. In 20 nonNPSLE patients compared to controls, decreased CC volume was noted and associated with decline in measures of visuomotor reasoning and speed (Johnson, Pinkston, Bigler, & Blatter, Reference Johnson, Pinkston, Bigler and Blatter1996). Appenzeller, Rondina, Li, Costallat, and Cendes (2005) reported that CC volumes were significantly smaller in 115 SLE patients compared to controls, with approximately 25% also demonstrating cerebral atrophy. The NPSLE patients in this study had greater CC volume loss than those with nonNPSLE, but did not show significantly more cerebral atrophy. SLE patients with cognitive impairment had reduced CC sizes; however, only a third of the patients were cognitively tested and no data were available for review.
MRS was then used as a method to study WM that does not show abnormality on conventional MRI (NAWM), and correlations with cognition were made. Elevated Ch/Cr was reported in frontal WM of eight nonNPSLE patients compared to eight demographically similar controls, and patients with greater cognitive impairment had significantly increased Ch/Cr (Kozora et al., Reference Kozora, Arciniegas, Filley, Ellison, West, Brown and Simon2005). In a later study from our group, there was no evidence of cerebral atrophy (based on volumetric analysis) or neuronal damage (based on NAA/Cr measurements) in 60 nonNPSLE patients compared to 24 demographically similar controls (Filley et al., Reference Filley, Kozora, Brown, Miller, West, Arciniegas and Zhang2009). However, higher Ch/Cr was found in the right frontal WM (p = .03) and left frontal WM (p = .05) in the nonNPSLE patients compared to controls. Figure 1 demonstrates higher Ch/Cr in an SLE patient compared to a control subject. A WM Cognitive Score (WMCS) was developed using a stepwise backward regression model to predict Ch/Cr from nine measures of attention and executive function derived from the nonNPSLE subjects. The total WMCS was found to correlate positively with total WM volume and negatively with left frontal WM Ch/Cr (Filley et al., Reference Filley, Kozora, Brown, Miller, West, Arciniegas and Zhang2009). In another disease affecting WM, human immunodeficiency virus (HIV) infection, elevated Ch/Cr in frontal WM was related to executive dysfunction (Chang et al., Reference Chang, Ernst, Witt, Ames, Gaiefsky and Miller2002), providing supportive evidence that elevated Ch/Cr may have a similar implication in this disease.
Fig. 1 Magnetic resonance spectroscopy spectra from right frontal lobe white matter showing normal choline (Ch) in a control subject (left) and elevated choline (Ch) in an SLE subject (right)
The available MRS data in SLE indicate that early changes in WM, particularly in myelin, may play an important role in cognitive impairment. Strong correlations between elevated frontal WM Ch/Cr and cognitive dysfunction in nonNPSLE patients have been evident in our studies thus far. Elevated frontal WM Ch/Cr may underlie MCD-SLE, which appears to be a precursor of more obvious brain pathology, advancing cognitive dysfunction, and clinical deterioration. The development of a WMCS that correlates with Ch/Cr may allow testing for the strength of this hypothesized brain-behavior relationship. One potential confound in this effort may be the stability of Ch/Cr in nonNPSLE patients, as some SLE studies suggest variability in neurometabolite levels over time (Sibbitt, Haseler, Griffey, Friedman, & Brooks, Reference Sibbitt, Haseler, Griffey, Friedman and Brooks1997; Sundgren et al., Reference Sundgren, Jennings, Attwood, Nan, Gebarski, McCune and Maly2005).
No studies have evaluated DTI parameters in relation to cognition in SLE patients. However, studies of MS, a disease sharing much in common with SLE, reported correlations between various DTI measures and indices of attention and executive function (Rovaris et al., Reference Rovaris, Riccitelli, Judica, Possa, Caputo, Ghezzi and Filippi2008). These findings, particularly in the NAWM, highlight the potential for DTI to shed light on the pathogenesis of cognitive dysfunction in SLE.
Although this review is not intended to discuss details of pathogenesis, it is possible that MCD-SLE results from immune-mediated myelinopathy whereas cognitive dysfunction in NPSLE tends to involve a combination of immune-mediated and vascular injury affecting both WM and gray matter. An autopsy and MRS study of patients who died with NPSLE found widespread vascular injury in the brain, and lowered NAA/Cr was associated with reduced neuronal density while higher Ch/Cr was related to gliosis (Brooks et al., Reference Brooks, Sibbitt, Kornfeld, Jung, Bankhurst and Roldan2010). Whereas these MRS changes are consistent with those observed in less affected SLE patients, the neuropathology of those dying with NPSLE is of course more extensive and severe than would be expected in nonNPSLE, and evidence exists to support early myelinopathy accounting for MCD-SLE before any other brain pathology develops (Filley et al., Reference Filley, Kozora, Brown, Miller, West, Arciniegas and Zhang2009).
Conclusions
Recent observations may compel a major shift in the conceptualization of cognitive dysfunction in autoimmune diseases such as SLE, bringing into focus pathology in the myelin and axons of cerebral WM. As one conclusion among many, damage to the myelin within frontal WM may lead to impaired attention, executive dysfunction, and slowed processing speed before more pervasive cognitive dysfunction develops.
A key implication of our review is the hypothesis that microstructural WM abnormalities produce cognitive dysfunction by affecting the operations of frontal systems in SLE. Correspondingly, these effects may account for deficits in attention, executive function, and speed of processing. Another less studied but intriguing possibility is that memory retrieval may be specifically affected, as has been observed in a wide range of WM disorders (Filley, Reference Filley2001). Advanced neuroimaging can help address all these issues. MRS has proven highly productive thus far in the study of WM and cognition in SLE (Filley et al., Reference Filley, Kozora, Brown, Miller, West, Arciniegas and Zhang2009; Kozora et al., Reference Kozora, Arciniegas, Filley, Ellison, West, Brown and Simon2005). DTI is also sensitive to microstructural WM changes (Zhang et al., Reference Zhang, Harrison, Heier, Zimmerman, Ravdin, Lockshin and Ulug2007), and recent evidence of frontal WM damage in patients with nonNPSLE (Ulug et al., Reference Ulug, Vo, Kozora, Ramos, Vega, Zimmerman and Lockshin2010) suggests that impaired cognition may be related to WM abnormalities as detected by this technique. The pathogenesis of WM injury putatively underlying MCD-SLE is uncertain, but an autoimmune process is suspected, and CSF studies may permit identification of inflammatory mediators that may contribute to microstructural WM damage. Ultimately, a combination of neuropsychological, neuroimaging, and immunological techniques appears likely to advance our understanding of the role of WM in the cognitive dysfunction of SLE at all stages of the disease.
Many other implications arise from this work. Evidence of WM damage (Emmer et al., Reference Emmer, Veer, Steup-Beekman, Huizinga, van der Grond and van Buchem2010) and deficits in attention and executive function (Kozora, Reference Kozora, Morgan and Ricker2008) are emerging in patients with rheumatoid arthritis and Sjogren's syndrome, suggesting that our studies of WM damage in SLE may also have implications for the mechanisms and treatment of cognitive dysfunction in other autoimmune diseases. In a broader sense, the study of WM in relation to cognitive dysfunction and dementia holds great promise for improving our knowledge of a wide variety of poorly understood neurologic and psychiatric disorders that cause significant human suffering (Filley, Reference Filley2001). In SLE, the concept of MCD-SLE offers a useful clinical model that may facilitate a deeper understanding of cognition in this disease, distinguishing these patients both from cognitively unaffected SLE patients, and those with the more severe cognitive dysfunction of NPSLE. Future studies in this area would benefit from selection of demographically diverse and well characterized NP activity in patient samples, use of reliable and valid cognitive tests, sophisticated neuroimaging (volumetric MRI, MRS, and DTI) across a variety of white and gray matter areas with scanners at field strength of 3.0 T or above, and investigations of CSF inflammatory activity. Finally, longitudinal studies are necessary, particularly with early, nonNPSLE patients, to understand the complex relationships between biological and neurobehavioral processes in SLE and to confirm that MCD-SLE is a precursor to more significant cognitive problems associated with NPSLE.
Acknowledgments
The authors report no conflicts of interest. This manuscript has not been published previously either electronically or in print. Supported in part by the National Institute of Musculoskeletal and Skin Diseases, (Grant RO1 AR049152).
