Neuropsychiatric manifestations are commonly. Neuropsychiatric Syndromes in Systemic Lupus Erythematosus: A Meta-Analysis SLE

Transcription

1 SLE Neuropsychiatric Syndromes in Systemic Lupus Erythematosus: A Meta-Analysis Avraham Unterman, MD,*, Johannes E.S. Nolte, MD,*, Mona Boaz, PhD,, Maya Abady, MPT, Yehuda Shoenfeld, MD,*, and Gisele Zandman-Goddard, MD, Objectives: To assess the prevalence of the 19 neuropsychiatric (NP) syndromes in systemic lupus erythematosus (SLE) patients, as defined by the American College of Rheumatology (ACR) in 1999, and better understand the reasons for interstudy variability of prevalence estimates, by performing a meta-analysis of relevant publications. Methods: A literature search from April 1999 to May 2008 was performed to identify studies investigating NP syndromes in patients with definite SLE, applying the 1999 ACR case definitions and having a sample size of at least 30 patients. Excluded were studies that did not relate to all 19 NPSLE syndromes, presented duplicate data, or were irrelevant. Results: Seventeen of 112 identified studies matched the inclusion criteria, reporting on a total of 5057 SLE patients, including 1439 NPSLE patients, with 2709 NPSLE syndromes. In a subanalysis of the 10 higher quality prospective and elicited studies (2049 patients) using the randomeffects model, the prevalence of NP syndromes in SLE patients was estimated to be 56.3% (95% CI 42.5%-74.7%), and the most frequent NP syndromes were headache 28.3% (18.2%-44.1%), mood disorders 20.7% (11.5%-37.4%), cognitive dysfunction 19.7% (10.7%-36%), seizures 9.9% (4.8%-20.5%), and cerebrovascular disease 8.0% (4.5%-14.3%), although significant between-study heterogeneity was present (P 0.05). Autonomic disorder and Guillain-Barré syndrome carried a prevalence of less than 0.1%. No case of plexopathy was reported. Conclusions: NP syndromes were estimated to exist in more than half of SLE patients. The most prevalent manifestations were headache, mood disorders, and cognitive dysfunction. A major limitation of the study was the significant heterogeneity of prevalence estimates between studies Elsevier Inc. All rights reserved. Semin Arthritis Rheum 41:1-11 Keywords: meta-analysis, prevalence, neuropsychiatric, systemic lupus erythematosus, American College of Rheumatology Neuropsychiatric manifestations are commonly found in patients with SLE and are an important cause of morbidity and mortality in these patients (1). However, for many years, lack of consensus in defining neuropsychiatric syndromes in systemic lupus *Department of Medicine B and Center of Autoimmune Diseases, Sheba Medical Center, Tel-Hashomer, Israel. Sackler Faculty of Medicine, Tel-Aviv University, Tel-Aviv, Israel. Epidemiology Unit, Wolfson Medical Center, Holon, Israel. Edgecliff Physiotherapy Sports & Spinal Centre, Sydney, Australia. Department of Medicine C, Wolfson Medical Center, Holon, Israel. The authors have no conflicts of interest to disclose. Address reprint requests to Yehuda Shoenfeld, MD, Department of Medicine B and Center of Autoimmune Diseases, Sheba Medical Center, Tel-Hashomer 52621, Israel. shoenfel@post.tau.ac.il /11/$-see front matter 2011 Elsevier Inc. All rights reserved. doi: /j.semarthrit erythematosus (NPSLE) presented a major obstacle in their research. In 1999, the American College of Rheumatology (ACR) Ad Hoc Committee on neuropsychiatric lupus nomenclature developed case definitions for 19 NPSLE syndromes to standardize definitions, diagnostic criteria, exclusions, associations, and diagnostic testing (2). Since its publication in 1999, this nomenclature has been widely accepted and many studies have investigated the prevalence of the 19 neuropsychiatric (NP) syndromes in both adult and pediatric SLE populations. Despite using these standardized definitions, the prevalence of NP manifestations in SLE patients has ranged widely from 12 to 95% in different series (3-19). This vast interstudy difference was previously attributed to multiple factors, including study design (prospective vs retrospective), 1

2 2 Neuropsychiatric syndromes in SLE diverse screening strategies used across studies, different baseline characteristics of the patients (adult vs pediatric patients, disease duration, disease severity, ethnicity etc), and duration of follow-up (1). Hence, the overall prevalence of NPSLE is currently unclear. A decade after the first publication of the 1999 ACR case definitions, we conducted a meta-analysis of relevant studies that assessed the prevalence of NPSLE by employing these case definitions. Our goals were to estimate the overall prevalence of NPSLE; evaluate the relative prevalence of each of the 19 NP syndromes, with special attention to the rare syndromes that are not represented in most series; and better understand the reasons for the marked variance of prevalence estimates between studies. METHODS This study was conducted according to the guidelines outlined by the Meta-analysis of Observational Studies in Epidemiology (MOOSE) group (20). Data Sources and Searches We conducted a PubMed/Medline and EMBASE search from April 1999 to May 2008 utilizing the following commonly used keywords: neuropsychiatric and lupus or CNS and lupus, combined with American College of Rheumatology or ACR. A follow-up of the relevant bibliography in articles was also undertaken to identify additional relevant articles. In certain cases, we corresponded with the authors of the articles for further information or clarification. Study Selection Abstracts of all identified reports published in all languages were reviewed independently by 2 investigators (A.U. and J.N.). Case reports, review articles, and papers not dealing with SLE patients were excluded and were not further reviewed. In case an abstract was considered potentially relevant by either 1 of the 2 investigators, the article was reviewed in full-text and assessed for relevance by applying a checklist of inclusion and exclusion criteria. Included were studies investigating NP syndromes in patients with definite systemic lupus erythematosus (SLE) (ie, each patient fulfilling at least 4 ACR criteria for SLE), applying the 1999 ACR case definitions, and having a sample size of at least 30 patients. Excluded were studies that did not relate to all 19 NPSLE syndromes, studies that presented duplicate data, or irrelevant studies that were not identified in the initial review of abstracts (eg, case reports or review articles). Data Extraction and Quality Assessment For each relevant study, we identified and recorded the total number of SLE patients, the number of NP patients, the cumulative number of NP syndromes (each patient could have more than 1 NP syndrome during the course of the disease), the study design and characteristics of patients, including age group (adult or pediatric), mean age, disease duration, age at SLE diagnosis, gender distribution, ethnicity, and nationality. We also recorded the cumulative prevalence of each of the 19 NPSLE syndromes in each cohort, as well as the SLE Disease Activity Index and the Systemic Lupus International Collaborative Clinics/American College of Rheumatology Damage Index when available. When possible, we extracted the exact numbers of syndromes as reported in the articles, and when not available we calculated them from proportions provided. The process of data extraction was performed by a single investigator (A.U.) to assure uniformity of the data and then re-evaluated by a second investigator (J.N.). Data Synthesis and Analysis Prevalence estimates for each of the 19 NP syndromes and the total NPSLE prevalence for each study were calculated together with their 95% confidence intervals (CI). Additionally, prevalence rates for prospective or elicited, and separately, retrospective studies were estimated with 95% CI and an overall prevalence rate with 95% CI was also calculated. Except where indicated otherwise, prevalence rates are shown as percentage of the entire SLE patient population. Comparison of the prevalence estimates of each of the 19 NP syndromes between prospective/elicited and retrospective studies was conducted using the 2 test. Tests were 2-sided and considered significant at P to account for multiple comparisons and maintain the overall alpha level at approximately 5%. Because of the significant difference in reported prevalence between these 2 types of studies, between-study heterogeneity of each NP syndrome was assessed only for the prospective/ elicited studies, using the 2 test (P 0.05 indicating significant heterogeneity) and the Higgins and Thompson s heterogeneity index (21). In addition to arithmetic means, prevalence estimates were combined across prospective/elicited studies, using both fixed-effect (Mantel-Haenszel) and random-effects (DerSimonian-Laird) models (22-24). Random-effects models tend to provide wider confidence intervals (24) and are preferable in the presence of between-study heterogeneity. Except where otherwise indicated, randomeffects estimates are displayed. Statistical calculations were done with WinPepi software version 7.4 (25) and with Microsoft Excel RESULTS Our search identified 112 articles (Fig. 1), of which 17 met the inclusion criteria and were incorporated in the final analysis (Table 1). The final cohort comprised of 5057 SLE patients. Of these, 1439 patients (28.5%, range: 12.2%-94.7%) had 2709 NP syndromes (mean of 1.88 syndromes per NPSLE patient, range: ).

3 A. Unterman et al. 3 Figure 1 Flowchart showing inclusion process. Most studies were single-center (13/17 studies) and reported on an adult population (8/17, N 3782 patients), although some studies reported on either pediatric (5/17, N 550) or mixed adult and pediatric populations (4/17, N 725). The majority of patients were women (76.4%- 95.4% of patients in the various studies). Data on ethnicity were available for 85.6% of the patients (4329 of 5057) in 11 of 17 studies (Table 1), with 72.4% Asians, 16.3% Caucasians, 5.8% Hispanic, 3.5% African American, and 2% of other ethnicities. Ethnicity was reported for 93.8% (2821/3008) of the patients in the retrospective studies compared with 73.6% (1508/2049) in the prospective or elicited ones. Of note is a 96% predominance of Asian patients in retrospective studies, compared with only 28.3% in prospective or elicited studies. Data regarding SLE disease duration (Table 1) were available for 95.8% (1962/2049) of the patients in 9 of 10 prospective and elicited studies (mean duration 4.8 years, range years), but was only available for 33% (993/ 3008) of the patients in 5 of 7 retrospective studies (mean duration 6.0 years, range years). The correlation coefficient of disease duration and NPSLE prevalence for the prospective and elicited studies was The cumulative prevalence of each of the 19 NPSLE syndromes in each of the 17 studies, as well as the combined prevalence, is shown in Table 2. When all 17 studies were pooled together, the most frequent NP syndromes among the 5057 patients were headache (12.2%), mood disorders (7.4%), seizures (7.0%), cognitive dysfunction (6.6%), and cerebrovascular disease (5.0%). Autonomic disorder and Guillain-Barré syndrome (GBS) each carried a prevalence rate of less than 0.1%. Plexopathy (plexus injury) was not reported in any of the 5057 patients. Central nervous system manifestations were more frequently observed than peripheral nervous system ones (93.1% vs 6.9% of all manifestations, respectively). The differences in NPSLE prevalence between the 10 prospective or elicited studies (encompassing 2049 patients) and the 7 retrospective, chart-driven ones (3008 patients) are depicted in Table 3. Of note is the significantly higher prevalence of most syndromes in the prospective or elicited studies. In fact, aseptic meningitis was the only syndrome with a significantly higher prevalence in retrospective chart-driven studies. The most frequent NP syndromes among the 10 prospective and elicited studies were headache (23.3%), mood disorder (14.9%), cognitive dysfunction (13.9%), seizure disorder (8.0%), and cerebrovascular disease (7.2%). The total NPSLE prevalence was 44.5% (or 56.3% using random-effects model) in the prospective or elicited studies, compared with only 17.6% in the retrospective studies (P 0.001). Table 4 depicts both arithmetic means and random-effects estimates (when appropriate) of the prevalence of each NP syndrome in the 10 prospective or elicited studies, along with the 95% CI. The random-effects estimate of the total NPSLE prevalence in the prospective and elicited studies (56.3%, 95% CI: 42.5%-74.7%) was higher than the arithmetic one (44.5%, 95% CI: 42.3%-46.6%). Between-study heterogeneity was assessed for the 10 prospective and elicited studies (Fig. 2), yielding significant heterogeneity in the total NPSLE prevalence (P 0.001), as well as in the prevalence of each individual NP syndrome, except for plexopathy (with a uniform zero prevalence in all studies). DISCUSSION Methodologic Differences Between Studies We identified 2 major methodologic approaches used in the 17 studies. The first (applied in 7/17 studies) was purely a retrospective, chart-driven one and therefore prone to biases such as low-quality patient records and underestimation of the subjective manifestations (eg, headache). The second approach (used in 10/17 studies) combined a retrospective chart review with either a prospective follow-up or additional elicited information (eg, questionnaires about past medical history, additional physical or neuropsychologic evaluation, etc). The latter approach is generally more accurate but still has its limitations, eg, recall biases. Many syndromes were more prevalent in prospective or elicited

4 4 Neuropsychiatric syndromes in SLE Table 1 Properties of the 17 Analyzed Studies No. Source Name of Cohort Study Design Age Group Prospective or elicited 1 Hanly et al 2008 (19) Systemic Lupus International Collaborating Clinics (SLICC) P/E, MC (24 centers), only NP syndromes around SLE diagnosis were included P/E, SC, IP, and OP 2 Abdel-Nasser et al 2008 (3) El-Minia University Hospital, Egypt Adult and pediatric 3 Mok et al 2006 (9) Tuen Mun Hospital, Hong-Kong, P/E, SC, data on NP syndromes that Adult occurred before 2000 is R 4 Schenatto et al 2006 Hospital de Clınicas de Porto P/E, SC, only 1 main active NP Adult (18) Alegre (HCPA), Brazil syndrome was reported per patient 5 Appenzeller et al University of Campinas, Sao P/E, SC, patients unable to undergo Adult and 2005 (4) Paulo, Brazil MRI were excluded pediatric 6 Afeltra et al 2003 Rome, Italy P/E, NS Adult (10) 7 Sanna et al 2003 (13) St. Thomas Hospital Lupus Clinic, P/E, SC Adult UK 8 Sibbitt et al 2002 (8) University of New Mexico Lupus P/E, SC, IP, and OP, used an Pediatric Cohort adaptation of the ACR 1999 case 9 Brey et al 2002 (15) San Antonio Lupus Study of NP Disease (SALUD) 10 Ainiala et al 2001 Finnish Population-Based Cohort (16) Retrospective, chart-driven 11 Zhou et al 2008 (5) Peking Union Medical College (PUMC) Hospital 12 Avcin et al 2008 (7) Hospital for Sick Children, Toronto 13 Spinosa et al 2007 (6) 14 Yu et al 2006 (12) Rheumatology Clinic of National definitions P/E, MC, data from study entry P/E, MC, population-based, IP, and OP Sum (prospective/elicited) Adult Adult Adult R, SC, IP Adult R, SC, only patients with apl levels Pediatric available were included Pequeno Principe Hospital, Brazil R, SC Pediatric Taiwan University Hospital 15 Harel et al 2006 (11) Stanford University M.C. & UCSF M.C. R, SC Pediatric R, MC (2 centers), headache underestimated: only severe, unresponsive to narcotics was considered Pediatric 16 Robert et al 2006 (14) Medical College Hospital, India R, SC, IP Adult and pediatric 17 Mok et al 2001 (17) Queen Mary Hospital, Hong Kong R, SC Adult and pediatric Sum (retrospective) Total (prospective and retrospective) NP, neuropsychiatric; SLE, systemic lupus erythematosus; P/E, prospective or elicited; R, retrospective chart-driven; SC, single-center; MC, multicenter; SPP, syndromes per patient; n/a, not available; C, Caucasian; H, Hispanic; AA, African American; As, Asian; O, other; MRI, magnetic resonance imaging; IP, inpatients; OP, outpatients; ACR, American College of Rheumatology; apl, antiphospholipid antibodies. a Ethnicity of patients is available for 11 of 17 studies. When not available, nationality is shown in parentheses. studies than in retrospective chart-driven ones (Table 3), especially those that have a somewhat subjective nature or the ones that require active testing to diagnose. Examples include headache (23.3% in prospective or elicited studies vs 4.7% in retrospective chartdriven ones), mood disorders (14.9% vs 2.3%), cognitive dysfunction (13.9% vs 1.6%), and anxiety disorder (5.3% vs 0.4%). It is conceivable that even prospective and elicited studies may have underestimated the prevalence of some syndromes due to incomplete evaluation methods used in some of the studies. Conversely, objective syndromes such as seizures and movement disorders, which are relatively easy to diagnose and are generally well recorded in patient charts,

5 A. Unterman et al. 5 Table 1 Continued No. of SLE Patients Prevalence of NPSLE SPP Ratio Female Age at SLE Gender (%) Ethnicity a Onset (yr) Disease Duration (yr) % % C 53.6%; H 12%; As 16%; AA 14.4%; O 3.9% % % n/a (Egypt) % % As 100% % % C 76%; O 24% n/a n/a % % n/a (Brazil) % % n/a (Italy) % % n/a (UK) % % H 61%; C 27%; As 4%; AA 4%; O 4% % % H 56%; C 30%; AA 8%; O 3% % % C 100% % % 1.63 n/a As 100% n/a n/a % % n/a (Canada) % % C 74%; O 26% (mulattos) % % As 100% % % As 37%; H 26%; C 20%; AA 8.5%; O 8.5% % % n/a (India) n/a n/a % % As 100% % % 1.88 did not show significant difference between the 2 study types and are less likely to be underestimated. NPSLE Prevalence and Reasons for Heterogeneity Between Studies In this study, the cumulative prevalence of NPSLE was 17.6% in retrospective chart-driven studies and 44.5% (or 56.3% using random-effects model) in prospective or elicited ones. However, even in the methodologically superior prospective and elicited studies, a wide heterogeneity of prevalence exists between studies (range: 23.0%- 94.7%, P for heterogeneity between studies). A thorough evaluation of the articles reporting the highest (8,16) and the lowest prevalence (9,18,19) revealed some

6 6 Neuropsychiatric syndromes in SLE Table 2 Prevalence of NP Syndromes in Each Study Source SLE pts. NPSLE prev. Prevalence of CNS Syndromes (%) Asept men. CVD Demyl. Headache Mov dis. Myelopathy Prospective or elicited Hanly 2008 (19) % Abdel-Nasser 2008 (3) % Mok 2006 (9) % Schenatto 2006 (18) % Appenzeller 2005 (4) % Alfeltra 2003 (10) % Sanna 2003 (13) % Sibbitt 2002 (8) % Brey 2002 (15) % Ainiala 2001 (16) % Retrospective, chart-driven Zhou 2008 (5) % Avcin 2008 (7) % Spinosa 2007 (6) % Yu 2006 (12) % Harel 2006 (11) % Robert 2006 (14) % Mok 2001 (17) % Total % pts., patients; prev., prevalence; asept men., aseptic meningitis; CVD, cerebrovascular disease; demyl., demyelinative syndrome; mov, movement; dis., disorder; seiz, seizure; ACS, acute confusional state; anx, anxiety; cogn dysf., cognitive dysfunction; AIDP, acute inflammatory demyelinating polyneuropathy; GBS, Guillain-Barre syndrome; auton, aoutonomic; MN, mononeuropathy; myas grav., myasthenia gravis; CN, cranial neuropathy; PN, polyneuropathy. Seiz dis. ACS major differences, which may account for this wide difference. The lowest reported prevalence among the prospective and elicited studies was 23% (9); this article examined Chinese patients newly diagnosed with SLE between 1990 and 2004 and was prospective only in part, since data regarding NP manifestations that occurred before the year 2000 was reviewed retrospectively from medical records. Additionally, headache was infrequently reported in this Chinese cohort, a population known to report a lower rate of headache (9). Another study (18) reported a prevalence rate of 26.4%, but included only the major NP syndromes; headache, for example, was not reported at all. A third study (19) researched only NP syndromes around the time of SLE diagnosis, without long-term follow-up of patients, resulting in a relatively low prevalence (30.4%). On the other end of the spectrum, a study of pediatric patients (8) used an adaptation of the ACR 1999 case definitions, with broader diagnostic criteria, resulting in higher prevalence (94.7%). The young age of the patients may also have influenced the prevalence, as pediatric SLE has been associated with greater severity and poorer outcome than in adults (8,26). A prevalence of 91.3% was reported in a study of relatively older SLE patients with longer disease duration (16). These patients were tested with a comprehensive (3-4 hours long) neurocognitive battery, leading to a relatively high cognitive dysfunction prevalence estimate of 80.4%, which constituted the main component of NPSLE in that study. With some of the reports underestimating and some possibly overestimating the prevalence of NPSLE, the randomeffects prevalence estimate of the prospective and elicited studies (ie, 56.3%) is the best estimate so far of the true prevalence of NPSLE, as defined by the 1999 ACR criteria, although even this figure may represent an underestimation. Frequent Syndromes Headache, mood disorders, seizures, cognitive dysfunction, and cerebrovascular disease were the 5 most frequently reported syndromes in our meta-analysis, each with a prevalence rate greater than 5%. Headache in SLE patients was systematically reviewed (27) and the researchers concluded that the prevalence of all headache types, particularly that of tension-type headache and migraine, does not differ between SLE patients and the general population. Of interest is the large difference between the pooled prevalence of 50.2% in the aforementioned article, as opposed to 12.2% in our pooled cohort, and 23.3% (28.3% in random-effects model) if only prospective and elicited studies are considered. A major part of this difference can be explained by the relatively large Asian population in our cohort, known to report a lower prevalence of headache (9). This interethnic difference is very prominent in our cohort (only 1-3.8% headache in 4 studies of ethnic Chinese) (5,9,12,17). Another factor is the lower prevalence of

7 A. Unterman et al. 7 Table 2 Continued Prevalence of CNS Syndromes (%) Prevalence of PNS Syndromes (%) Anx dis. cogn dysf. mood dis. psychosis AIDP (GBS) auton dis. MN Myas Grav. CN plexopathy PN headache in retrospective chart-driven studies (representing 3008/5057 patients in our cohort). When analyzing the prevalence of headache in prospective studies excluding one of Chinese subjects (9) and one where headache was not reported (18), the prevalence using random-effects model is 37.0% (range: 19.2%-72%; 95% CI: 24.3%-56.3%), a figure much closer to the aforementioned 50.2%. Cognitive dysfunction was defined by the ACR as significant deficits in any or all of the following cognitive functions: simple or complex attention, reasoning, executive skills, memory, visual-spatial processing, language, and psychomotor speed (2). The ACR committee also proposed a standard 1-hour battery of neuropsychologic tests. The reliability and validity of this battery were later tested and established (28). The wide range of cognitive dysfunction prevalence in the various studies (0-80%) is mainly the result of different rates of neuropsychologic testing. All retrospective chart-driven studies and some of the elicited or prospective studies did not perform neuropsychologic testing (or tested only some of the patients), resulting in lower reported prevalence of cognitive dysfunction. Therefore, the pooled prevalence of 6.6% in our metaanalysis (and even the random-effects model estimate of 19.7% in prospective and elicited studies) is most probably an underestimate of the true prevalence of cognitive dysfunction in SLE. When proper neuropsychologic testing is conducted in unselected SLE patients, the prevalence of cognitive dysfunction is much higher (23-60% in most series) (3,10,28-31). Infrequent Syndromes We performed a review of the literature to assess the relationship between the least frequent syndromes (plexopathy, GBS, myasthenia gravis, and autonomic dysfunction) and SLE. As for the plexopathy, a literature search revealed only 3 case reports (32-34) of SLE patients with brachial plexus injury. In 1 case (33) bilateral brachial plexus neuropathy was discovered 2 weeks later to be the first sign of polyneuropathy. Therefore, the association between neural plexus injury and SLE requires further consideration. Approximately 40 cases of GBS were described in SLE patients, according to a 1994 case report and review of literature (35). A retrospective series of 1100 patients with GBS showed 7 patients to have SLE (36). The relative rarity of GBS reported in our meta-analysis ( 0.1%) is notable since GBS is a fairly common syndrome, with a median annual incidence of 1.3 cases/100,000 population (37). Autonomic disorders were quite rare in our meta-analysis ( 0.1%). However, a growing body of literature in recent years suggests a much higher prevalence in SLE patients, ranging from 6 to 93% in different studies (38). This wide variability may reflect different methodology, with some authors even reporting the absence of any significant autonomic dysfunction (39). A recent study (38),

8 8 Neuropsychiatric syndromes in SLE Table 3 Prevalence of NP Manifestations According to Study Design All Patients (N 5057) P/E (N 2049) RC (N 3008) NPSLE Syndrome Number of Synd. Prevalence Prevalence Prevalence P Value Headache % 23.3% 4.7% Mood disorder % 14.9% 2.3% Seizure disorder % 8.0% 6.4% 0.03 Cognitive dysfunction % 13.9% 1.6% CVD % 7.2% 3.6% Psychosis % 3.9% 2.8% 0.03 Acute confusional state % 3.9% 2.5% Anxiety disorder % 5.3% 0.4% Polyneuropathy % 3.0% 0.5% Cranial neuropathy % 1.7% 0.5% Aseptic meningitis % 0.3% 1.3% Myelopathy % 1.0% 0.8% 0.40 Mononeuropathy (single, multiplex) % 1.5% 0.5% Movement disorder % 0.9% 0.6% 0.25 Demyelinating syndrome % 0.3% 0.3% 0.96 Myasthenia gravis 8 0.2% 0.4% 0.0% AIDP (GBS) % 0.1% 0.1% 0.70 Autonomic disorder % 0.1% 0.0% 0.16 Plexopathy 0 0.0% 0.0% 0.0% Total % 44.5% 17.6% CVD, cerebrovascular disease; AIDP, acute inflammatory demyelinating polyneuropathy; GBS, Guillain-Barre syndrome; synd., syndrome; P/E, prospective or elicited studies; RC, retrospective chart-driven studies. P values are for the comparison between P/E and RC. comparing 51 SLE patients to 30 matched controls, revealed abnormal noninvasive autonomic tests in 18% of SLE patients compared with 3% of controls. In the same study, 37% of SLE patients exhibited autonomic symptoms when specifically asked, although none of these symptoms appeared in the patients charts as recorded by their attending physicians. Therefore, autonomic disorders are probably underestimated in our cohort, due to inappropriate history taking and autonomic testing in most studies. The possible association between myasthenia gravis and SLE has been raised in various case reports and reviews (40-42). In a recent study evaluating 78 unselected patients with myasthenia gravis, 6 patients (7.7%) met the criteria for SLE (41). Methodologic Problems and Biases in the Present Study An important limitation of our study is the significant between-study heterogeneity, as discussed above. We addressed heterogeneity by using a random-effects model that incorporated the uncertainty arising from betweenstudy differences to provide better estimates of the true prevalence than the arithmetic mean. Nevertheless, it cannot be overlooked that the prevalence estimates reported herein are derived from heterogenic studies, necessitating caution in interpreting or generalizing them. The potential biases inherent in the 17 surveyed studies have been addressed above. These biases most likely lead to underestimation of some syndromes, especially in the retrospective chart-driven studies. This in turn affects the total prevalence of NPSLE described in our cohort. The prospective and elicited studies, despite significant between-study heterogeneity, generally provide better estimates to the relative frequency of NPSLE syndromes, although some may still be underappreciated. Another limitation is the high percentage of Asian patients in our cohort. Asian predominance was more prominent in the retrospective chart-driven studies, with 96% Asian patients, compared with only 28.3% in prospective or elicited studies. Variations in expression and clinical course in patients from different ethnic groups are recognized in many acute and chronic diseases (43) and may also exist in SLE patients. For example, in the LUMINA (LUpus in MInorities: NAture vs nurture) cohort, African Americans had a more serious disease and worse outcomes as compared with Caucasians (43). Another example is the relatively low prevalence of headache in 4 studies of ethnic Chinese included in our cohort (5,9,12,17). The prevalence of some NP syndromes, such as headache, as well as the total prevalence of NPSLE, is likely to be affected by this Asian predominance, especially in the subanalysis of the retrospective chart-driven studies and to a lesser extent of the prospective and elicited ones. Since pediatric SLE has been associated with greater severity and poorer outcome than in adults (8,26), it would have been preferable to analyze these 2 populations separately.

9 A. Unterman et al. 9 Table 4 Estimates of NPSLE Prevalence in the Prospective and Elicited Studies NPSLE Syndrome Prevalence Range in P/E Studies (%) Arithmetic Mean Prevalence (%) 95% CI Random-Effects Model Prevalence (%) 95%CI Headache Mood disorder Cognitive dysfunction Seizure disorder CVD Anxiety disorder Acute confusional state Psychosis Polyneuropathy Cranial neuropathy Mononeuropathy (single, multiplex) Myelopathy Movement disorder Myasthenia gravis NA NA Demyelinating syndrome NA NA Aseptic meningitis NA NA Autonomic disorder NA NA AIDP (GBS) NA NA Plexopathy NA NA Total P/E, prospective or elicited; CVD, cerebrovascular disease; AIDP, acute inflammatory demyelinating polyneuropathy; GBS, Guillain-Barre syndrome; CI, confidence interval; NA, nonapplicable. Unfortunately, separating adult from pediatric patients in our cohort was problematic since some of the articles had mixed adult and pediatric populations, and since only 1 of the prospective studies (8) described a purely pediatric population. Although publication bias cannot be ruled out, it is unlikely, due to the descriptive nature of prevalence studies, with no right or wrong results. Figure 2 NPSLE prevalence in the 10 prospective and elicited studies. Black squares denote the prevalence of neuropsychiatric lupus (NPSLE) in 10 prospective and elicited studies, while their size is proportional to the sample size in each study. Horizontal bars represent 95% confidence intervals (CI). The diamond and dashed line mark the total prevalence and 95% CI, while the dotted line represents the total prevalence according to random-effects model. Despite obvious differences in quality across studies, we chose not to use quality scores to weigh the contribution of each study to the overall estimate. Recent evidence indicates that contemporary quality scores have little or no value in improving the utility of a meta-analysis and may introduce bias (44,45). Therefore, as recommended by the MOOSE guidelines (20), only subgrouping according to study design was applied. Neuropsychiatric lupus is probably not a single disease entity (46), but rather a mosaic of different syndromes with possible different pathophysiologic mechanisms, including autoantibody development (47). None of the 19 NP syndromes, as defined by the ACR, is specific to SLE; they are reported in association with many other autoimmune disorders, including rheumatoid arthritis, antiphospholipid syndrome (APS), Behçet s disease, Sjögren s syndrome, systemic vasculitides, and many more, as well as in individuals without autoimmune disease (16,48-50). Since most articles did not report prevalence rates of the 19 syndromes in comparison to matched healthy controls, we were unable to determine the fraction of neuropsychiatric morbidity specifically attributable to SLE. Moreover, since most articles did not report prevalence rates separately in SLE patients with APS, the contribution of APS to the overall neuropsychiatric morbidity could not be assessed. CONCLUSION Ours is the first study to compile the body of data on NPSLE prevalence using the 1999 ACR case definitions,

10 10 Neuropsychiatric syndromes in SLE a decade after their publication. NP syndromes were estimated to exist in more than half of SLE patients, making it one of the most common and significant manifestations of SLE. Many of the NP syndromes, such as cognitive dysfunction and autonomic disorders, may be difficult to detect and may require careful testing of all patients to avoid underdiagnosis. Determining the prevalence of each NP syndrome is only the first step; further research is needed to determine the relative prevalence of each syndrome in comparison to matched controls, including patients with other autoimmune diseases or apparently healthy subjects. Detecting significant differences in neuropsychiatric morbidity in comparison to other rheumatic diseases and between different populations of SLE patients may be a step toward understanding the pathophysiology of NPSLE. Moreover, some of these syndromes may not be more common in SLE patients than in the general population, allowing clinicians and researchers to concentrate efforts on fewer high-yield syndromes. ACKNOWLEDGMENT We would like to thank Dr. Yedidya Dotan for his important remarks and assistance with the statistical analysis. REFERENCES 1. Borchers AT, Aoki CA, Naguwa SM, Keen CL, Shoenfeld Y, Gershwin ME. Neuropsychiatric features of systemic lupus erythematosus. Autoimmun Rev 2005;4(6): The American College of Rheumatology nomenclature and case definitions for neuropsychiatric lupus syndromes. Arthritis Rheum 1999;42(4): Abdel-Nasser AM, Ghaleb RM, Mahmoud JA, Khairy W, Mahmoud RM. Association of anti-ribosomal P protein antibodies with neuropsychiatric and other manifestations of systemic lupus erythematosus. Clin Rheumatol 2008;27: Appenzeller S, Rondina JM, Li LM, Costallat LT, Cendes F. Cerebral and corpus callosum atrophy in systemic lupus erythematosus. Arthritis Rheum 2005;52(9): Zhou HQ, Zhang FC, Tian XP, Leng XM, Lu JJ, Zhao Y, et al. Clinical features and outcome of neuropsychiatric lupus in Chinese: analysis of 240 hospitalized patients. Lupus 2008;17(2): Spinosa MJ, Bandeira M, Liberalesso PB, Vieira SC, Janz LL Jr, Sa EG, et al. Clinical, laboratory and neuroimage findings in juvenile systemic lupus erythematosus presenting involvement of the nervous system. Arq Neuropsiquiatr 2007;65(2B): Avcin T, Benseler SM, Tyrrell PN, Cucnik S, Silverman ED. A follow up study of antiphospholipid antibodies and associated neuropsychiatric manifestations in 137 children with systemic lupus erythematosus. Arthritis Rheum 2008;59(2): Sibbitt WL Jr, Brandt JR, Johnson CR, Maldonado ME, Patel SR, Ford CC, et al. The incidence and prevalence of neuropsychiatric syndromes in pediatric onset systemic lupus erythematosus. J Rheumatol 2002;29(7): Mok CC, To CH, Mak A. Neuropsychiatric damage in Southern Chinese patients with systemic lupus erythematosus. Medicine (Baltimore) 2006;85(4): Afeltra A, Garzia P, Mitterhofer AP, Vadacca M, Galluzzo S, Del Porto F, et al. Neuropsychiatric lupus syndromes: relationship with antiphospholipid antibodies. Neurology 2003;61(1): Harel L, Sandborg C, Lee T, von Scheven E. Neuropsychiatric manifestations in pediatric systemic lupus erythematosus and association with antiphospholipid antibodies. J Rheumatol 2006; 33(9): Yu HH, Lee JH, Wang LC, Yang YH, Chiang BL. Neuropsychiatric manifestations in pediatric systemic lupus erythematosus: a 20-year study. Lupus 2006;15(10): Sanna G, Bertolaccini ML, Cuadrado MJ, Laing H, Khamashta MA, Mathieu A, et al. Neuropsychiatric manifestations in systemic lupus erythematosus: prevalence and association with antiphospholipid antibodies. J Rheumatol 2003;30(5): Robert M, Sunitha R, Thulaseedharan NK. Neuropsychiatric manifestations systemic lupus erythematosus: a study from South India. Neurol India 2006;54(1): Brey RL, Holliday SL, Saklad AR, Navarrete MG, Hermosillo- Romo D, Stallworth CL, et al. Neuropsychiatric syndromes in lupus: prevalence using standardized definitions. Neurology 2002;58(8): Ainiala H, Loukkola J, Peltola J, Korpela M, Hietaharju A. The prevalence of neuropsychiatric syndromes in systemic lupus erythematosus. Neurology 2001;57(3): Mok CC, Lau CS, Wong RW. Neuropsychiatric manifestations and their clinical associations in southern Chinese patients with systemic lupus erythematosus. J Rheumatol 2001;28(4): Schenatto CB, Xavier RM, Bredemeier M, Portela LV, Tort AB, Dedavid e Silva TL, et al. Raised serum S100B protein levels in neuropsychiatric lupus. Ann Rheum Dis 2006;65(6): Hanly JG, Urowitz MB, Su L, Sanchez-Guerrero J, Bae SC, Gordon C, et al. Short-term outcome of neuropsychiatric events in systemic lupus erythematosus upon enrollment into an international inception cohort study. Arthritis Rheum 2008;59(5): Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD, Rennie D, et al. Meta-analysis of observational studies in epidemiology: a proposal for reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA 2000; 283(15): Higgins JP, Thompson SG. Quantifying heterogeneity in a metaanalysis. Stat Med 2002;21(11): DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986;7(3): Mantel N, Haenszel W. Statistical aspects of the analysis of data from retrospective studies of disease. J Natl Cancer Inst 1959; 22(4): Egger M, Smith GD, Phillips AN. Meta-analysis: principles and procedures. BMJ 1997;315(7121): Abramson JH. WINPEPI (PEPI-for-Windows): computer programs for epidemiologists. Epidemiol Perspect Innov 2004;1(1): Muscal E, Myones BL. The role of autoantibodies in pediatric neuropsychiatric systemic lupus erythematosus. Autoimmun Rev 2007;6(4): Mitsikostas DD, Sfikakis PP, Goadsby PJ. A meta-analysis for headache in systemic lupus erythematosus: the evidence and the myth. Brain 2004;127(Pt5): Kozora E, Ellison MC, West S. Reliability and validity of the proposed American College of Rheumatology neuropsychological battery for systemic lupus erythematosus. Arthritis Rheum 2004; 51(5): Hanly JG, Robichaud J, Fisk JD. Anti-NR2 glutamate receptor antibodies and cognitive function in systemic lupus erythematosus. J Rheumatol 2006;33(8): Sabbadini MG, Manfredi AA, Bozzolo E, Ferrario L, Rugarli C, Scorza R, et al. Central nervous system involvement in systemic lupus erythematosus patients without overt neuropsychiatric manifestations. Lupus 1999;8(1): Tomietto P, Annese V, D Agostini S, Venturini P, La Torre G, De Vita S, et al. General and specific factors associated with severity of

11 A. Unterman et al. 11 cognitive impairment in systemic lupus erythematosus. Arthritis Rheum 2007;57(8): Assal F, Blanche P, Lamy C, Sicard D, De Recondo J. [Brachial plexus involvement disclosing systemic lupus erythematosus]. Presse Med 1993;22(12): Bloch SL, Jarrett MP, Swerdlow M, Grayzel AI. Brachial plexus neuropathy as the initial presentation of systemic lupus erythematosus. Neurology 1979;29: Siekert RG, Clark EC. Neurologic signs and symptoms as early manifestations of systemic lupus erythematosus. Neurology 1955; 5(2): Robson MG, Walport MJ, Davies KA. Systemic lupus erythematosus and acute demyelinating polyneuropathy. Br J Rheumatol 1994;33(11): Leneman F. The Guillain-Barre syndrome. Definition, etiology, and review of 1,100 cases. Arch Intern Med 1966;118(2): Hughes RA, Rees JH. Clinical and epidemiologic features of Guillain-Barre syndrome. J Infect Dis 1997;176(Suppl 2):S Shalimar, Handa R, Deepak KK, Bhatia M, Aggarwal P, Pandey RM. Autonomic dysfunction in systemic lupus erythematosus. Rheumatol Int 2006;26(9): Omdal R, Jorde R, Mellgren SI, Husby G. Autonomic function in systemic lupus erythematosus. Lupus 1994;3(5): Vaiopoulos G, Sfikakis PP, Kapsimali V, Boki K, Panayiotidis P, Aessopos A, et al. The association of systemic lupus erythematosus and myasthenia gravis. Postgrad Med J 1994;70(828): Sthoeger Z, Neiman A, Elbirt D, Zinger H, Magen E, Burstein R, et al. High prevalence of systemic lupus erythematosus in 78 myasthenia gravis patients: a clinical and serologic study. Am J Med Sci 2006;331(1): Bhinder S, Majithia V, Harisdangkul V. Myasthenia gravis and systemic lupus erythematosus: truly associated or coincidentaltwo case reports and review of the literature. Clin Rheumatol 2006; 25(4): Uribe AG, McGwin G Jr, Reveille JD, Alarcon GS. What have we learned from a 10-year experience with the LUMINA (Lupus in Minorities; Nature vs. nurture) cohort? Where are we heading? Autoimmun Rev 2004;3(4): Herbison P, Hay-Smith J, Gillespie WJ. Adjustment of metaanalyses on the basis of quality scores should be abandoned. J Clin Epidemiol 2006;59(12): Juni P, Witschi A, Bloch R, Egger M. The hazards of scoring the quality of clinical trials for meta-analysis. JAMA 1999;282(11): Sanchez-Guerrero J, Aranow C, Mackay M, Volpe B, Diamond B. Neuropsychiatric systemic lupus erythematosus reconsidered. Nat Clin Pract Rheumatol 2008;4(3): Zandman-Goddard G, Chapman J, Shoenfeld Y. Autoantibodies involved in neuropsychiatric SLE and antiphospholipid syndrome. Semin Arthritis Rheum 2007;36(5): Sofat N, Malik O, Higgens CS. Neurological involvement in patients with rheumatic disease. QJM 2006;99(2): Hanly JG, Fisk JD, McCurdy G, Fougere L, Douglas JA. Neuropsychiatric syndromes in patients with systemic lupus erythematosus and rheumatoid arthritis. J Rheumatol 2005;32(8): Ainiala H, Hietaharju A, Loukkola J, Peltola J, Korpela M, Metsanoja R, et al. Validity of the new American College of Rheumatology criteria for neuropsychiatric lupus syndromes: a population-based evaluation. Arthritis Rheum 2001;45(5):