Abstract
Childhood-onset systemic lupus erythematosus (cSLE) is a multisystem autoimmune disease characterized by immune dysregulation affecting patients less than 18 years old. One-fifth of SLE cases are diagnosed during childhood. cSLE presents differently from adults and has a more severe and aggressive course. We describe the clinical and antibody profiles in our cSLE Singapore cohort. All cSLE patients who satisfied the 1997 American College of Rheumatology diagnostic criteria were captured in our lupus registry from January 2009 to January 2014. Data including demographic, cumulative clinical, serologic data, and damage indices were collected. Adjusted mean SLEDAI-2K (AMS) was used to summarize disease activity over multiple visits. Cluster analysis using non-hierarchical K-means procedure was performed on eight selected antibodies. The 64 patients (female:male ratio 5:1; Chinese 45.3%, Malay 28.1%, Indian 9.4%, and other races 17.2%) had a mean onset age of 11.5 years (range 2.1–16.7) and mean age at diagnosis was 11.9 years (range 2.6–18.0). Our study demonstrated differences in clinical manifestations for which hematologic involvement was the most common manifestation with less renal disease and uncommon neurologic manifestation as compared to other cSLE cohorts reported in our region. Antibody clusters were identified in our cohort but their clinical association/discrimination and outcome prediction required further validation study. Outcomes of our cohort in regard to disease activity after therapy and organ damages were comparable if not better to other cSLE cohorts elsewhere. Steroid-related damage, including symptomatic multifocal avascular necrosis and cataract, were not uncommon locally. Infection remains the major cause of death for the continent. Nevertheless, the five year survival rate of our cohort (98.4%) was high.
Introduction
Childhood-onset systemic lupus erythematosus (cSLE) is a multisystem autoimmune disease characterized by immune dysregulation affecting patients less than 18 years old. Approximately one-fifth of patients with SLE develop the disease in the first two decades of life.1,2 Clinical presentation and organ involvement in cSLE are largely heterogeneous. Differences in immunologic, serologic, and clinical phenotypes between cSLE and adult-onset systemic lupus erythematosus (aSLE) were previously described.1,2 At presentation, cSLE patients often have more acute and severe diseases with more renal, neurologic, and hematologic disorders.3–7 These organ involvements, when present, are more severe with higher SLE disease activity than aSLE.8,9 When compared to aSLE, patients with cSLE have more permanent damage.7,10–12
SLE is known to be more prevalent among Asians.13–16 Differences in clinical characteristics, management, and overall outcomes in cSLE have been reported from three developing countries in Southeast Asia (SEA), i.e. Thailand, Philippines, and Vietnam.17–19 Singapore, a member of 11 SEA countries, is one of the two developed countries (Singapore and Brunei) with a multi-ethnic population of 5.47 million situated on an island to the south of Malaysia. 20 Three quarter of its population is Chinese (74.3%), followed by Malays (13.3%), Indians (9.1%), and other races (3.3%). 20 Singapore healthcare cost is heavily subsidized by the government, thus healthcare accessibility is not an issue. There is only one free-standing children’s hospital, KK Women’s and Children’s Hospital – an over 400-bed tertiary children’s hospital, in Singapore where the majority of sick children are referred. Due to unique differences in geography, socioeconomic, and healthcare system, we aim to describe our cSLE cohort in regard to clinical characteristics and outcomes and compare those to what were reported in the region and the West. Adult studies suggest autoantibodies tend to cluster and each cluster predicts different clinical phenotype.21,22 To date, there has been no report in SEA that such antibody clustering exists in cSLE for which we aim to identify the autoantibody clusters in our cSLE cohort and to evaluate the associations between the clusters and clinical phenotypes.
Methods
Patients
Patients 18 years old and younger, who fulfilled the revised 1997 American College of Rheumatology (ACR) SLE diagnostic criteria, were defined as having cSLE. 23 They were identified from our web-based lupus registry which has been maintained since the Pediatric Rheumatology Clinical Program inception in 2009. All cSLE patients evaluated and managed by pediatric rheumatologists at KK Women’s and Children’s Hospital between January 1, 2009 and January 31, 2014 were recruited. They were followed up at 1–3 month intervals. The SingHealth Cluster Institutional Review Board approved the study.
Clinical data
Patients’ demographic, clinical and laboratory, management, and outcome data were captured in the registry. The cSLE clinical manifestations were defined based on the revised 1997 ACR SLE criteria and American Rheumatism Association glossary committee consensus.23,24 For those with renal disorder, renal biopsy was performed in patients with significant proteinuria and histopathology was classified according to World Health Organization (WHO) classification. 25 Cumulative clinical manifestations was defined as accumulation or addition of successive clinical manifestations that patient developed during the study period.
Autoantibody detections
Autoantibodies, which include antinuclear antibodies (ANA), anti-double stranded DNA (anti-dsDNA), anti-Smith (anti-Sm), anti-U1RNP, anti-Ro/SSA and anti-La/SSB, anticardiolipin (aCL), anti-beta 2 glycoprotein I (anti-β2GPI) antibodies, and lupus anticoagulant (LAC) were included in the analysis. ANA was detected by indirect immunofluorescence using Hep-2 cells. Anti-dsDNA and other autoantibodies were analyzed by ELISA. Positive aCL IgM and IgG were defined as titer >40 MPL and >40 GPL, respectively. Eight autoantibodies (anti-dsDNA, anti-Sm, anti-U1RNP, anti-Ro/SSA, anti-La/SSB, aCL IgM, aCL IgG, LAC) were selected for cluster analysis.
Disease outcomes
Disease activity was assessed by the SLE Disease Activity Index-2K (SLEDAI-2K). 24 Adjusted mean SLEDAI-2K (AMS) was used to measure the burden of ongoing disease activity over time.27,28 The AMS score was calculated by the average of area under the curve of the SLEDAI scores over time.27,28 The area under the curve was obtained by adding the area of each of the blocks of visit interval and then dividing them by the length of time for the whole period.27,28 Both SLEDAI and AMS have the same unit and use the same interpretation. The levels are categorized as mild if score <6, moderate if score between 6–10, and severe if score >10. Damage was measured by Systemic Lupus International Collaborating Clinics (SLICC)/American College of Rheumatology Damage Index for SLE (SDI). 29
Statistical analysis
Descriptive statistics were used to describe the data. Chi-square or Fisher exact test were used to detect association between autoantibodies and clinical manifestations where appropriate. Odd ratios and 95% confidence intervals (CI) were provided. Kaplan–Meier analysis was used to analyze survival rate over time. Cluster analysis using K-means (non-hierarchical) procedure was performed on the eight selected antibodies. All statistical analyses were performed by SPSS software version 17 (SPSS Inc., Chicago, IL) and p values < 0.05 were considered significant.
Results
Patient characteristics
Sixty-four consecutive cSLE patients, which consisted of 52 (81.3%) females giving a female:male ratio of 5:1, were recruited. The majority of patients were Chinese (45.3%), followed by Malays (28.1%), Indians (9.4%), and other races (17.2%), which were Indonesians (7.9%), Burmese (3.3%), Pakistani (1.5%), Sikh (1.5%), Middle-Eastern (1.5%), and Vietnamese (1.5%). None of patients was of mixed ethnicity. The mean age of onset was 11.5 years (range 2.1–16.7). The mean age at diagnosis was 11.9 years (range 2.6–18.0). It took an average of 6 months before the diagnosis was made. The mean length of follow-up was 48.5 months (range 0.8–181.6).
Clinical manifestations
Clinical manifestations of cSLE in Singapore
Autoantibody profile and the clusters
Almost all patients had positive ANA (98.4%) and anti-dsDNA antibody (90.6%). About half of the cohort had positive anti-U1RNP (45.3%) and anti-Ro/SSA antibodies (53.1%), while one third had positive anti-Sm antibody (37.5%). Only 17.2% had a positive anti-La/SSB antibody. Antiphospholipid antibodies were found in 25 of 64 patients (39.1%), which included aCL IgM in 10.9%, aCL Ig G in 20.3%, and LAC in 23.4% of the patients. At diagnosis, the mean and range of the erythrocyte sedimentation rate (ESR) was 74.4 mm/hr (5.0–165.0) and that of C-reactive protein (CRP) was 9.1 mg/L (0.2–61.4). The mean and range of C3 level was 0.56 G/L (0.09–1.73) and C4 level was 0.08 G/L (0.02–0.35).
Cluster analysis: ethnicity and autoantibody frequencies and clinical associations
Values are number (%) of patients.
Others include Filipino, Indonesian, Burmese, Caucasian.
The prevalence of Raynaud’s phenomenon, cutaneous vasculitis, hemolytic anemia and myositis was significantly different among the clusters. Cluster 1 was associated with high frequencies of cutaneous vasculitis and hemolytic anemia while cluster 2 was characterized by the presence of Raynaud’s phenomenon. Patients in cluster 3 had low frequency of all manifestations except leukopenia and lymphopenia which presented in high frequencies across all the clusters.
Medications
Almost all patients were taking prednisolone (93.8%), hydroxychloroquine (96.9%), and calcium/vitamin D supplement (92.2%). For immunosuppressive agents, azathioprine (56.3%) was the most commonly used drug, followed by mycophenolate mofetil (45.3%), cyclophosphamide (31.3%), and methotrexate (6.3%). Four patients (6.3%) received rituximab, each for refractory nephritis, resistant thrombocytopenia, refractory autoimmune hemolytic anemia, and recurrent cutaneous vasculitis.
Out of 21 patients who had biopsy proven LN, 1 patient was on azathioprine, 17 patients were on cyclophosphamide, 18 patients were on mycophenolate mofetil, and 1 patient was on rituximab. Seven patients (10.9%) were taking alendronate for treatment of osteoporosis/osteonecrosis. Other medications used were aspirin (14.1%), angiotensin converting enzyme (ACE) inhibitor (29.7%), and thalidomide (3.1%).
Disease outcomes
The mean (range) of SLEDAI values at diagnosis, 6 months, and 12 months were 16.7 (2–33), 5.4 (0–33), and 1.9 (0–11), respectively. The AMS values at 6 months and 12 months were 1.9 (0.2–5.2) and 1.3 (0.2–3.8). Four patients (6.3%) had avascular necrosis, two (3.1%) with cataract, one (1.6%) with digital necrosis, and one (1.6%) with bowel infarction requiring resection. One patient (1.6%) died from overwhelming sepsis, secondary to multi-drug resistant Acinetobacter baumannii, resulting in a 5-year survival rate of 98.4% for our cohort.
Discussion
We describe the spectrum of cSLE in Singapore over the past five years, demonstrating the differences in clinical manifestations and outcome when compared to others in the region and the West which will be discussed further. Singapore is the only developed country with pediatric rheumatologists and has only one tertiary children’s hospital on the island, thus referral bias was expected to be minimal, in comparison to a few cSLE cohorts in the region which were based on single center of large countries whereby referral bias was inevitable.17–19 Learning the variant in cSLE manifestations and outcomes in each cohort will alert local physicians of the disease leading to sooner diagnosis and management.
Cumulative clinical and laboratory findings based on ACR criteria
θMedian (range). ΦMean (IQR), in years. †At disease onset. ††At presentation. ‡Centers from Belgium, United Kingdom, Netherlands, Slovakia. *Seizure and behavioral change. #Seizure only. ##psychosis only. δOnly anti-dsDNA. §only anti-Sm.
Hematologic manifestation was the most common manifestation for cSLE in Singapore, followed by arthritis and malar rash (Table 1 and 3). Lymphopenia was found in over three quarters of our patients while one-third of the cohort developed hemolytic anemia and thrombocytopenia. The presence of lymphopenia should alert local pediatricians to include cSLE as one of the differential diagnosis options. Hematologic involvement remains the top three most common manifestations of cSLE reported in the region (Table 3). The frequency of arthritis seen in Singapore cSLE cohort was similar to that reported from Vietnam, which was more prevalent than that reported from Thailand and Philippines.
Cutaneous manifestations in cSLE differ according to ethnicity.32,35,36 Malar rash, the most common cutaneous manifestations in our cohort, was surprisingly less common than those reported from our neighbor countries and the West.8,17,18,32,37 Photosensitivity and discoid rash was rather uncommon, which was also observed by Thai and Taiwanese investigators.19,33 Discoid rash was reported more frequently in African-Americans. 35 Nevertheless, it was unexpectedly found to be more common among Chinese than Malays or Indians, despite being statistically insignificant (p = 0.284) locally. In Singapore, oral ulcer was common and was comparable to other studies.
LN is a more frequent presenting feature in cSLE as compared to aSLE. 1 It not only determines the long-term outcome, but also is a major cause of death, in patients suffering from SLE worldwide. However, renal involvement was seen in only 40% of cSLE patients in Singapore as compared to that reported from Philippines, Vietnam, Thailand, and China, 72–86%.17–19,38 We believe this reflected our true prevalence of LN, since all cSLE patients were evaluated and managed primarily by pediatric rheumatologists in our hospital. Referral bias could not be disregarded in other reports, as cSLE especially with renal involvement were managed primarily and/or reported by pediatric nephrologists due to the lack of a pediatric rheumatology clinical program in the region.24,28 Unsurprisingly, WHO class IV LN was the most common histopathology identified, in accordance with other cSLE cohorts elsewhere. 1 However, none of our patients developed end-stage renal diseases. Early diagnosis and an effective treatment regimen remain crucial in cSLE management in order to improve long-term outcome.
Serositis and neurologic involvement were less prevalent locally.18,19,32,37 The former could be under reported since we do not routinely investigate asymptomatic patients. Neurologic disorder was relatively uncommon as compared to other studies (20.8–34.6%).8,18,19,33 The higher frequency mainly resulted from different definitions used for which not only psychosis and seizure, but also headache, and other peripheral and central nervous system involvements were included.8,17,33 Nonetheless, the frequency of our neurologic manifestation was accurate as other defined neurologic manifestations were not present although routine neurocognitive assessment was not done locally. The prevalence of hemolytic anemia, thrombocytopenia, lymphadenopathy and pleuritis were not increased during follow-up which was also observed by Rood et al. and Hiraki et al.5,8
Nearly all (98%) of our cSLE patients had ANA in accordance with other cSLE reports elsewhere.8,18,19,33 Anti-dsDNA antibody was detected in a majority of patients but it was not associated with higher frequency of renal disease locally (p = 0.702). The differences in anti-dsDNA antibody prevalence were explained by different techniques of detection.8,17–19,33 The distribution of other SLE-related autoantibodies were similar to those reported from Taiwan for which anti-U1RNP and anti-Ro/SSA antibodies were found in one-half of the patients as compared to one-third or less prevalence reported by others.8,32,33,37 These two autoantibodies were found more frequently in Asian and African Americans than Caucasians in a Canadian cSLE cohort. 8 However, there was no difference in autoantibody distribution among ethnic backgrounds.
As suggested by aSLE studies and one cSLE report, autoantibodies, not only are the hallmark of the disease feature, but also tend to appear in clusters.21,22,39 The autoantibody clusters were shown to associate with certain clinical features and outcomes.21,22,39 We then explored such clusters in our cohort. Three distinct clusters using eight autoantibodies were identified by K-mean cluster analysis. Since anti-dsDNA antibody was positive in nearly all our cSLE patients (90.6%) and it distributed in similar frequency across all clusters, thus it had no role in differentiating the clusters. This phenomenon was also observed with LAC, despite its low frequency. Patients in cluster 1, having all the remaining six autoantibodies – “full-house” cluster, were characterized by the high frequencies of cutaneous vasculitis, hemolytic anemia, and myositis. Cluster 2 was characterized by prominent anti-Sm/anti-U1RNP antibodies – “mixed connective tissue disease (MCTD)-like cluster – and associated with Raynaud’s phenomenon but low frequency for hemolytic anemia. Cluster 3 was characterized by low prevalence of all other antibodies except anti-dsDNA and anti-Ro antibodies. No patients in this cluster developed myositis and psychosis. This may represent a milder cSLE subgroup in our cohort. Interestingly, autoantibody clusters reported by Jurencak et al. in their cSLE cohort had similar autoantibody distribution when compared to our clusters. 39 Although the clinical association of each cluster was not identical between the two cohorts, certain similarities existed. The “full-house” cluster which is similar to cluster 2 of their report tended to have highest frequency of hemolytic anemia, serositis and nephritis (Table 2) as compared to other clusters. Difference in ethnic backgrounds of the two cohorts may in part be responsible for the clinical association discrepancy observed.
Disease activity assessed by SLEDAI is well accepted and is a crucial tool to guide therapy and predicts damage in children. 40 The only cSLE disease activity reported in SEA was from Vietnam for whose SLEDAI at diagnosis was significantly higher than that of our cohort, 23.8 ± 11.6 vs. 16.7 ± 7.6. 17 This large discrepancy was likely resulted from high prevalence of renal involvement in the Vietnamese cohort (82%) as compared to the Singapore cohort (40.6%). Our SLEDAI was comparable to that reported from Canada (16.8 ± 10.1). 9 After treatment, our SLEDAI dropped significantly at 6 months (5.4 ± 7.1) and 12 month (2.0 ± 2.7). Due to different time-interval during follow-up, summarizing patients’ experience over time is better described by using AMS. 27 However, the AMS and mean SLEDAI are identical once follow-up intervals are the same. We calculated AMS at 6 months and 12 months and demonstrated low AMS at both time-intervals, 1.9 ± 1.2 and 1.3 ± 0.8, respectively. This persistent SLEDAI was secondary to persistent positive anti-dsDNA in majority of our patients after therapy.
Long term outcomes are summarized by either assessing organ damage (SLICC/SDI) or mortality rate. More damage was seen in cSLE as compared to that of aSLE. 9 Organ damage was more common in patients with neuropsychiatric manifestations at diagnosis and had longer disease activity. 41 Disease damage was rarely reported in SEA. Brunner et al reported 56.1% of their Canadian cSLE patients had SDI score > 0. 1 During the five year study period, we observed 12.6% of our patients with SDI > 0. Symptomatic multifocal avascular necrosis (AVN) was the most common organ damage identified (6.3%), followed by posterior subcapsular cataract (3.1%), digital necrosis (1.6%), and bowel infarction requiring resection (1.6%). Hiraki et al. reported incidence of AVN and cataract was as high as 10% and 19% respectively. 8 The common use of corticosteroids in our cohort (93.8%) as the cause of these complications is yet to be determined. Our Filipino colleagues reported higher damages affecting neurologic (19.0%) and musculoskeletal (8.7%) systems. 18 Early treatment, fewer patients with renal and neurologic disorders and judicious use of immunosuppressive agents may explain our less non-steroid-related SDI.
Mortality rate in cSLE varies according to countries and health institutions. However, the five year survival rate in cSLE has improved significantly in the past decades (>90%) reported from developed countries in the West. 1 Major causes of death in adult and children with SLE are renal disease, severe disease flares and infections. 1 In SEA, Gulay et al. reported 9/78 of their patients died, giving an estimated five year survival rate of 88.5% and infections were causes of death in 77%. 18 In a Vietnamese’s cohort, 4/45 patients died at 6 months follow-up from circulatory collapse and infection. 17 In our study, 1/64 patients died from overwhelming multi-drug resistant bacterial sepsis in the adult hospital two months after transfer, giving the overall five year survival rate of 98.4%. Infection remains the major cause of death in Asia, not limited to SEA, as reported by Wang et al. in their retrospective analysis of mortality in cSLE from Taiwan for which infection was the leading cause of death in both of their 1980–1990 (78.1%) and 1991–2001 (72.2%) cohorts. 33 On the contrary, SLE and related complications seemed to be the major causes of death reported by Hiraki et al. and Ramirez Gomez et al. in their Canadian and Latin American cSLE cohorts, respectively.8,32 Lower socioeconomic status, poor patient education/awareness and healthcare accessibility were all partly contributed to variation in morbidity rate in our region.
Our current study was limited by the small sample size. The autoantibody and clinical features association could not be demonstrated. Moreover, the validity of the autoantibody clusters may not be robust. However, autoantibody clusters were identified but their clinical association/discrimination and outcome prediction required further study.
Our study demonstrated differences in clinical manifestations for which hematologic involvement was the most common manifestation with less renal disease and uncommon neurologic manifestation as compared to other cSLE cohorts reported in our region. We confirm that autoantibodies tend to appear in clusters in cSLE and their clinical associations were noted. Outcomes of the current cohort in regard to disease activity after therapy and organ damages were comparable if not better than other cSLE cohorts elsewhere. Steroid related damage including AVN and cataract were not uncommon locally. Infection remains the major cause of death for the continent. Nevertheless, the five year survival rate of our cohort is high.
Footnotes
Funding
The study and the registry were supported by grants from the National Arthritis Foundation of Singapore 2009 and the ILAR grant for developing countries 2011.
Conflict of interest statement
The authors have no conflicts of interest to declare.